diff --git "a/checkpoint-40000/trainer_state.json" "b/checkpoint-40000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-40000/trainer_state.json"
@@ -0,0 +1,24091 @@
+{
+  "best_metric": 0.882483720779419,
+  "best_model_checkpoint": "/content/drive/MyDrive/Data Mining and Analysis/coursework/bert_fine_tune/logs/report_1/checkpoint-35000",
+  "epoch": 3.4937549130928467,
+  "eval_steps": 5000,
+  "global_step": 40000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9995632806358636e-05,
+      "loss": 1.0403,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999126561271727e-05,
+      "loss": 0.9465,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998689841907591e-05,
+      "loss": 0.9756,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9982531225434535e-05,
+      "loss": 0.9459,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9978164031793175e-05,
+      "loss": 1.0372,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99737968381518e-05,
+      "loss": 0.9496,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996942964451044e-05,
+      "loss": 0.9111,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996506245086907e-05,
+      "loss": 0.9431,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996069525722771e-05,
+      "loss": 0.9688,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995632806358634e-05,
+      "loss": 0.9417,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9951960869944974e-05,
+      "loss": 0.9999,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9947593676303614e-05,
+      "loss": 1.0532,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994322648266224e-05,
+      "loss": 0.8558,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993885928902088e-05,
+      "loss": 0.9832,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993449209537951e-05,
+      "loss": 0.9702,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993012490173815e-05,
+      "loss": 0.9343,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992575770809678e-05,
+      "loss": 1.0262,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9921390514455414e-05,
+      "loss": 0.8871,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991702332081405e-05,
+      "loss": 1.0159,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991265612717268e-05,
+      "loss": 0.9799,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990828893353131e-05,
+      "loss": 1.0253,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9903921739889946e-05,
+      "loss": 0.9639,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9899554546248586e-05,
+      "loss": 0.9653,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989518735260721e-05,
+      "loss": 0.9899,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989082015896585e-05,
+      "loss": 1.0361,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9886452965324486e-05,
+      "loss": 0.9525,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988208577168312e-05,
+      "loss": 1.0019,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.987771857804175e-05,
+      "loss": 0.9388,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9873351384400386e-05,
+      "loss": 0.9571,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986898419075902e-05,
+      "loss": 0.9769,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986461699711765e-05,
+      "loss": 0.8649,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986024980347629e-05,
+      "loss": 0.8633,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9855882609834925e-05,
+      "loss": 1.0568,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985151541619356e-05,
+      "loss": 0.9646,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984714822255219e-05,
+      "loss": 0.899,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9842781028910825e-05,
+      "loss": 0.8334,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983841383526946e-05,
+      "loss": 0.9937,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983404664162809e-05,
+      "loss": 0.8855,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9829679447986724e-05,
+      "loss": 0.8992,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.982531225434536e-05,
+      "loss": 1.104,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982094506070399e-05,
+      "loss": 0.9004,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981657786706263e-05,
+      "loss": 0.9589,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9812210673421264e-05,
+      "loss": 0.9941,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98078434797799e-05,
+      "loss": 0.8478,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980347628613853e-05,
+      "loss": 0.9095,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9799109092497164e-05,
+      "loss": 0.9783,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97947418988558e-05,
+      "loss": 1.0161,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979037470521443e-05,
+      "loss": 0.912,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978600751157306e-05,
+      "loss": 0.9125,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9781640317931696e-05,
+      "loss": 1.0088,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9777273124290336e-05,
+      "loss": 0.9104,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977290593064897e-05,
+      "loss": 1.0435,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97685387370076e-05,
+      "loss": 0.8887,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9764171543366236e-05,
+      "loss": 0.8453,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975980434972487e-05,
+      "loss": 0.9301,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97554371560835e-05,
+      "loss": 0.8271,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9751069962442136e-05,
+      "loss": 1.0066,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9746702768800776e-05,
+      "loss": 1.0448,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97423355751594e-05,
+      "loss": 0.9184,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973796838151804e-05,
+      "loss": 0.9653,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973360118787667e-05,
+      "loss": 0.9661,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972923399423531e-05,
+      "loss": 1.0116,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972486680059394e-05,
+      "loss": 0.8806,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9720499606952575e-05,
+      "loss": 0.9603,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971613241331121e-05,
+      "loss": 0.9405,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971176521966984e-05,
+      "loss": 1.0418,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970739802602848e-05,
+      "loss": 0.8486,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970303083238711e-05,
+      "loss": 0.9366,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.969866363874575e-05,
+      "loss": 0.9031,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9694296445104374e-05,
+      "loss": 1.0206,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9689929251463014e-05,
+      "loss": 1.0355,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968556205782165e-05,
+      "loss": 0.9031,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968119486418028e-05,
+      "loss": 1.0127,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9676827670538914e-05,
+      "loss": 1.0261,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967246047689755e-05,
+      "loss": 0.8633,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966809328325619e-05,
+      "loss": 1.0313,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966372608961481e-05,
+      "loss": 0.9448,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965935889597345e-05,
+      "loss": 0.8489,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965499170233208e-05,
+      "loss": 0.9398,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965062450869072e-05,
+      "loss": 0.901,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9646257315049346e-05,
+      "loss": 0.938,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9641890121407986e-05,
+      "loss": 0.9577,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963752292776662e-05,
+      "loss": 0.8499,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963315573412525e-05,
+      "loss": 0.9224,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962878854048389e-05,
+      "loss": 0.8919,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962442134684252e-05,
+      "loss": 0.9534,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962005415320116e-05,
+      "loss": 0.9699,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9615686959559785e-05,
+      "loss": 1.0054,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9611319765918425e-05,
+      "loss": 0.9884,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.960695257227705e-05,
+      "loss": 1.0228,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.960258537863569e-05,
+      "loss": 1.0274,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9598218184994325e-05,
+      "loss": 0.9991,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.959385099135296e-05,
+      "loss": 0.9862,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.958948379771159e-05,
+      "loss": 0.8869,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9585116604070224e-05,
+      "loss": 0.9661,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9580749410428864e-05,
+      "loss": 0.9623,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.957638221678749e-05,
+      "loss": 0.9473,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957201502314613e-05,
+      "loss": 0.9598,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9567647829504764e-05,
+      "loss": 0.9164,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95632806358634e-05,
+      "loss": 0.9441,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955891344222203e-05,
+      "loss": 1.0582,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9554546248580664e-05,
+      "loss": 0.9841,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95501790549393e-05,
+      "loss": 0.8739,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954581186129793e-05,
+      "loss": 0.9216,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954144466765657e-05,
+      "loss": 0.9129,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9537077474015196e-05,
+      "loss": 1.0256,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9532710280373836e-05,
+      "loss": 0.9285,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952834308673247e-05,
+      "loss": 0.9412,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.95239758930911e-05,
+      "loss": 0.9598,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9519608699449736e-05,
+      "loss": 0.9926,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.951524150580837e-05,
+      "loss": 0.981,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9510874312167e-05,
+      "loss": 0.9916,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9506507118525636e-05,
+      "loss": 0.9936,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.950213992488427e-05,
+      "loss": 0.9909,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.94977727312429e-05,
+      "loss": 0.8959,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.949340553760154e-05,
+      "loss": 0.9105,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9489038343960175e-05,
+      "loss": 1.0268,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.948467115031881e-05,
+      "loss": 0.9223,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.948030395667744e-05,
+      "loss": 0.8919,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9475936763036075e-05,
+      "loss": 1.0177,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.947156956939471e-05,
+      "loss": 0.9374,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.946720237575334e-05,
+      "loss": 0.9679,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9462835182111974e-05,
+      "loss": 0.8975,
+      "step": 1230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9458467988470614e-05,
+      "loss": 0.9361,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.945410079482925e-05,
+      "loss": 1.0163,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.944973360118788e-05,
+      "loss": 1.0202,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9445366407546514e-05,
+      "loss": 1.0489,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.944099921390515e-05,
+      "loss": 0.918,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.943663202026378e-05,
+      "loss": 0.9559,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9432264826622414e-05,
+      "loss": 1.0352,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.942789763298105e-05,
+      "loss": 0.9161,
+      "step": 1310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.942353043933968e-05,
+      "loss": 0.9595,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.941916324569832e-05,
+      "loss": 0.9689,
+      "step": 1330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9414796052056946e-05,
+      "loss": 0.9298,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9410428858415586e-05,
+      "loss": 0.9834,
+      "step": 1350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.940606166477422e-05,
+      "loss": 1.0293,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.940169447113285e-05,
+      "loss": 0.9291,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9397327277491486e-05,
+      "loss": 0.9656,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.939296008385012e-05,
+      "loss": 0.9721,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.938859289020876e-05,
+      "loss": 0.8988,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9384225696567386e-05,
+      "loss": 0.8917,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9379858502926026e-05,
+      "loss": 0.9286,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.937549130928465e-05,
+      "loss": 0.9466,
+      "step": 1430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.937112411564329e-05,
+      "loss": 0.9644,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9366756922001925e-05,
+      "loss": 1.015,
+      "step": 1450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.936238972836056e-05,
+      "loss": 0.9779,
+      "step": 1460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.935802253471919e-05,
+      "loss": 0.855,
+      "step": 1470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9353655341077825e-05,
+      "loss": 0.9618,
+      "step": 1480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9349288147436465e-05,
+      "loss": 0.9882,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.934492095379509e-05,
+      "loss": 0.9304,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.934055376015373e-05,
+      "loss": 1.0002,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.933618656651236e-05,
+      "loss": 0.8864,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9331819372871e-05,
+      "loss": 0.9655,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9327452179229624e-05,
+      "loss": 0.9529,
+      "step": 1540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9323084985588264e-05,
+      "loss": 0.8572,
+      "step": 1550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.93187177919469e-05,
+      "loss": 0.918,
+      "step": 1560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.931435059830553e-05,
+      "loss": 1.0434,
+      "step": 1570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.930998340466417e-05,
+      "loss": 0.8926,
+      "step": 1580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.93056162110228e-05,
+      "loss": 1.0405,
+      "step": 1590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.930124901738144e-05,
+      "loss": 0.9279,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.929688182374006e-05,
+      "loss": 0.948,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.92925146300987e-05,
+      "loss": 0.8886,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.928814743645733e-05,
+      "loss": 1.016,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.928378024281597e-05,
+      "loss": 0.9776,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.92794130491746e-05,
+      "loss": 0.9765,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9275045855533236e-05,
+      "loss": 1.0266,
+      "step": 1660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.927067866189187e-05,
+      "loss": 0.8994,
+      "step": 1670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.92663114682505e-05,
+      "loss": 0.9955,
+      "step": 1680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.926194427460914e-05,
+      "loss": 0.8983,
+      "step": 1690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.925757708096777e-05,
+      "loss": 0.986,
+      "step": 1700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.925320988732641e-05,
+      "loss": 0.959,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9248842693685035e-05,
+      "loss": 1.0302,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9244475500043675e-05,
+      "loss": 0.9265,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.924010830640231e-05,
+      "loss": 0.9686,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.923574111276094e-05,
+      "loss": 0.9412,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9231373919119575e-05,
+      "loss": 0.9361,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.922700672547821e-05,
+      "loss": 1.0435,
+      "step": 1770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.922263953183685e-05,
+      "loss": 1.0142,
+      "step": 1780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9218272338195474e-05,
+      "loss": 0.9519,
+      "step": 1790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9213905144554114e-05,
+      "loss": 0.9459,
+      "step": 1800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.920953795091275e-05,
+      "loss": 0.9458,
+      "step": 1810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.920517075727138e-05,
+      "loss": 1.0128,
+      "step": 1820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9200803563630014e-05,
+      "loss": 0.9978,
+      "step": 1830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.919643636998865e-05,
+      "loss": 1.0136,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.919206917634728e-05,
+      "loss": 0.9648,
+      "step": 1850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9187701982705914e-05,
+      "loss": 0.9627,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.918333478906455e-05,
+      "loss": 0.9315,
+      "step": 1870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.917896759542318e-05,
+      "loss": 0.9317,
+      "step": 1880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.917460040178182e-05,
+      "loss": 0.9756,
+      "step": 1890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.917023320814045e-05,
+      "loss": 0.8983,
+      "step": 1900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9165866014499086e-05,
+      "loss": 0.9227,
+      "step": 1910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.916149882085772e-05,
+      "loss": 1.0251,
+      "step": 1920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.915713162721635e-05,
+      "loss": 0.9369,
+      "step": 1930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9152764433574986e-05,
+      "loss": 1.0062,
+      "step": 1940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.914839723993362e-05,
+      "loss": 1.0771,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.914403004629225e-05,
+      "loss": 0.9042,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9139662852650886e-05,
+      "loss": 0.9503,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9135295659009526e-05,
+      "loss": 0.914,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.913092846536816e-05,
+      "loss": 0.906,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.912656127172679e-05,
+      "loss": 0.9457,
+      "step": 2000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9122194078085425e-05,
+      "loss": 0.8958,
+      "step": 2010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.911782688444406e-05,
+      "loss": 0.952,
+      "step": 2020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.911345969080269e-05,
+      "loss": 0.9643,
+      "step": 2030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9109092497161325e-05,
+      "loss": 0.8752,
+      "step": 2040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.910472530351996e-05,
+      "loss": 0.9656,
+      "step": 2050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.91003581098786e-05,
+      "loss": 0.9661,
+      "step": 2060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9095990916237224e-05,
+      "loss": 1.0128,
+      "step": 2070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9091623722595864e-05,
+      "loss": 0.9888,
+      "step": 2080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.90872565289545e-05,
+      "loss": 0.96,
+      "step": 2090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.908288933531313e-05,
+      "loss": 0.9634,
+      "step": 2100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9078522141671764e-05,
+      "loss": 0.9381,
+      "step": 2110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.90741549480304e-05,
+      "loss": 0.8479,
+      "step": 2120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.906978775438903e-05,
+      "loss": 0.9178,
+      "step": 2130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9065420560747664e-05,
+      "loss": 0.8983,
+      "step": 2140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9061053367106304e-05,
+      "loss": 0.8988,
+      "step": 2150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.905668617346493e-05,
+      "loss": 0.812,
+      "step": 2160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.905231897982357e-05,
+      "loss": 0.949,
+      "step": 2170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.90479517861822e-05,
+      "loss": 0.9629,
+      "step": 2180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9043584592540836e-05,
+      "loss": 1.0334,
+      "step": 2190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.903921739889947e-05,
+      "loss": 1.0536,
+      "step": 2200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.90348502052581e-05,
+      "loss": 0.9478,
+      "step": 2210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.903048301161674e-05,
+      "loss": 0.9571,
+      "step": 2220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.902611581797537e-05,
+      "loss": 0.9597,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.902174862433401e-05,
+      "loss": 0.791,
+      "step": 2240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9017381430692636e-05,
+      "loss": 0.9065,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9013014237051276e-05,
+      "loss": 1.04,
+      "step": 2260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.90086470434099e-05,
+      "loss": 0.9144,
+      "step": 2270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.900427984976854e-05,
+      "loss": 0.8441,
+      "step": 2280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8999912656127175e-05,
+      "loss": 0.996,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.899554546248581e-05,
+      "loss": 0.8924,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.899117826884445e-05,
+      "loss": 0.9115,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8986811075203075e-05,
+      "loss": 0.9701,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8982443881561715e-05,
+      "loss": 0.9551,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.897807668792034e-05,
+      "loss": 0.9846,
+      "step": 2340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.897370949427898e-05,
+      "loss": 0.9265,
+      "step": 2350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.896934230063761e-05,
+      "loss": 0.9866,
+      "step": 2360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.896497510699625e-05,
+      "loss": 1.0033,
+      "step": 2370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.896060791335488e-05,
+      "loss": 0.8552,
+      "step": 2380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8956240719713514e-05,
+      "loss": 0.8984,
+      "step": 2390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.895187352607215e-05,
+      "loss": 1.0317,
+      "step": 2400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.894750633243078e-05,
+      "loss": 0.8503,
+      "step": 2410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.894313913878942e-05,
+      "loss": 1.0195,
+      "step": 2420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.893877194514805e-05,
+      "loss": 1.0256,
+      "step": 2430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.893440475150669e-05,
+      "loss": 1.0027,
+      "step": 2440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.893003755786531e-05,
+      "loss": 0.9532,
+      "step": 2450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.892567036422395e-05,
+      "loss": 1.0081,
+      "step": 2460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8921303170582586e-05,
+      "loss": 0.9046,
+      "step": 2470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.891693597694122e-05,
+      "loss": 0.9324,
+      "step": 2480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.891256878329985e-05,
+      "loss": 0.9082,
+      "step": 2490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8908201589658486e-05,
+      "loss": 0.9223,
+      "step": 2500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8903834396017126e-05,
+      "loss": 0.9729,
+      "step": 2510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.889946720237575e-05,
+      "loss": 0.9788,
+      "step": 2520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.889510000873439e-05,
+      "loss": 0.975,
+      "step": 2530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.889073281509302e-05,
+      "loss": 1.0072,
+      "step": 2540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.888636562145166e-05,
+      "loss": 1.0536,
+      "step": 2550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.888199842781029e-05,
+      "loss": 0.9831,
+      "step": 2560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8877631234168925e-05,
+      "loss": 0.9724,
+      "step": 2570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.887326404052756e-05,
+      "loss": 0.962,
+      "step": 2580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.886889684688619e-05,
+      "loss": 0.9508,
+      "step": 2590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8864529653244825e-05,
+      "loss": 0.9489,
+      "step": 2600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.886016245960346e-05,
+      "loss": 1.0164,
+      "step": 2610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.88557952659621e-05,
+      "loss": 0.9989,
+      "step": 2620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8851428072320725e-05,
+      "loss": 0.8959,
+      "step": 2630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8847060878679364e-05,
+      "loss": 0.9129,
+      "step": 2640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8842693685038e-05,
+      "loss": 0.9048,
+      "step": 2650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.883832649139663e-05,
+      "loss": 0.9269,
+      "step": 2660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8833959297755264e-05,
+      "loss": 0.9609,
+      "step": 2670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.88295921041139e-05,
+      "loss": 1.0289,
+      "step": 2680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.882522491047253e-05,
+      "loss": 0.9198,
+      "step": 2690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8820857716831164e-05,
+      "loss": 0.9428,
+      "step": 2700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8816490523189804e-05,
+      "loss": 0.8933,
+      "step": 2710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.881212332954844e-05,
+      "loss": 0.9559,
+      "step": 2720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.880775613590707e-05,
+      "loss": 0.9619,
+      "step": 2730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.88033889422657e-05,
+      "loss": 1.0151,
+      "step": 2740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8799021748624337e-05,
+      "loss": 0.9329,
+      "step": 2750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.879465455498297e-05,
+      "loss": 0.9951,
+      "step": 2760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.87902873613416e-05,
+      "loss": 0.9182,
+      "step": 2770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8785920167700236e-05,
+      "loss": 0.8708,
+      "step": 2780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.878155297405887e-05,
+      "loss": 0.9361,
+      "step": 2790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.87771857804175e-05,
+      "loss": 1.0204,
+      "step": 2800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.877281858677614e-05,
+      "loss": 0.9972,
+      "step": 2810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8768451393134776e-05,
+      "loss": 0.958,
+      "step": 2820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.876408419949341e-05,
+      "loss": 0.9688,
+      "step": 2830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.875971700585204e-05,
+      "loss": 1.0413,
+      "step": 2840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8755349812210675e-05,
+      "loss": 0.9984,
+      "step": 2850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.875098261856931e-05,
+      "loss": 1.0529,
+      "step": 2860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.874661542492794e-05,
+      "loss": 0.9634,
+      "step": 2870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.874224823128658e-05,
+      "loss": 0.9478,
+      "step": 2880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.873788103764521e-05,
+      "loss": 0.8547,
+      "step": 2890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.873351384400385e-05,
+      "loss": 0.9699,
+      "step": 2900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.872914665036248e-05,
+      "loss": 0.8683,
+      "step": 2910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8724779456721115e-05,
+      "loss": 0.911,
+      "step": 2920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.872041226307975e-05,
+      "loss": 0.8428,
+      "step": 2930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.871604506943838e-05,
+      "loss": 0.842,
+      "step": 2940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8711677875797014e-05,
+      "loss": 0.8891,
+      "step": 2950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.870731068215565e-05,
+      "loss": 0.9217,
+      "step": 2960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.870294348851429e-05,
+      "loss": 0.9264,
+      "step": 2970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8698576294872914e-05,
+      "loss": 0.9735,
+      "step": 2980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8694209101231554e-05,
+      "loss": 0.9314,
+      "step": 2990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.868984190759018e-05,
+      "loss": 0.9456,
+      "step": 3000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.868547471394882e-05,
+      "loss": 1.0547,
+      "step": 3010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.868110752030745e-05,
+      "loss": 0.934,
+      "step": 3020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8676740326666087e-05,
+      "loss": 0.8343,
+      "step": 3030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.867237313302472e-05,
+      "loss": 1.0218,
+      "step": 3040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.866800593938335e-05,
+      "loss": 1.0145,
+      "step": 3050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.866363874574199e-05,
+      "loss": 0.8718,
+      "step": 3060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.865927155210062e-05,
+      "loss": 0.901,
+      "step": 3070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.865490435845926e-05,
+      "loss": 0.8845,
+      "step": 3080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8650537164817886e-05,
+      "loss": 0.961,
+      "step": 3090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8646169971176526e-05,
+      "loss": 0.9009,
+      "step": 3100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.864180277753516e-05,
+      "loss": 0.9523,
+      "step": 3110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.863743558389379e-05,
+      "loss": 0.8963,
+      "step": 3120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8633068390252425e-05,
+      "loss": 0.8846,
+      "step": 3130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.862870119661106e-05,
+      "loss": 0.9082,
+      "step": 3140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.86243340029697e-05,
+      "loss": 0.9656,
+      "step": 3150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8619966809328325e-05,
+      "loss": 0.8836,
+      "step": 3160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8615599615686965e-05,
+      "loss": 1.0194,
+      "step": 3170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.861123242204559e-05,
+      "loss": 1.0275,
+      "step": 3180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.860686522840423e-05,
+      "loss": 0.8716,
+      "step": 3190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.860249803476286e-05,
+      "loss": 0.9184,
+      "step": 3200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.85981308411215e-05,
+      "loss": 1.0129,
+      "step": 3210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.859376364748013e-05,
+      "loss": 0.8573,
+      "step": 3220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8589396453838764e-05,
+      "loss": 0.9787,
+      "step": 3230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8585029260197404e-05,
+      "loss": 1.0416,
+      "step": 3240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.858066206655603e-05,
+      "loss": 0.9824,
+      "step": 3250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.857629487291467e-05,
+      "loss": 0.9861,
+      "step": 3260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.85719276792733e-05,
+      "loss": 0.9741,
+      "step": 3270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.856756048563194e-05,
+      "loss": 0.8959,
+      "step": 3280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.856319329199057e-05,
+      "loss": 1.037,
+      "step": 3290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.85588260983492e-05,
+      "loss": 0.9341,
+      "step": 3300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8554458904707837e-05,
+      "loss": 0.9072,
+      "step": 3310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.855009171106647e-05,
+      "loss": 0.9548,
+      "step": 3320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.85457245174251e-05,
+      "loss": 0.9305,
+      "step": 3330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8541357323783736e-05,
+      "loss": 1.0128,
+      "step": 3340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8536990130142376e-05,
+      "loss": 0.9327,
+      "step": 3350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8532622936501e-05,
+      "loss": 0.992,
+      "step": 3360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.852825574285964e-05,
+      "loss": 0.9628,
+      "step": 3370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8523888549218276e-05,
+      "loss": 0.9177,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.851952135557691e-05,
+      "loss": 0.9348,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.851515416193554e-05,
+      "loss": 0.9206,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8510786968294175e-05,
+      "loss": 0.8575,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.850641977465281e-05,
+      "loss": 0.9916,
+      "step": 3420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.850205258101144e-05,
+      "loss": 1.1207,
+      "step": 3430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.849768538737008e-05,
+      "loss": 0.9739,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.849331819372871e-05,
+      "loss": 0.8795,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.848895100008735e-05,
+      "loss": 0.863,
+      "step": 3460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.848458380644598e-05,
+      "loss": 0.9418,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8480216612804615e-05,
+      "loss": 0.9192,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.847584941916325e-05,
+      "loss": 1.0152,
+      "step": 3490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.847148222552188e-05,
+      "loss": 0.9408,
+      "step": 3500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8467115031880514e-05,
+      "loss": 0.8909,
+      "step": 3510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.846274783823915e-05,
+      "loss": 0.965,
+      "step": 3520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.845838064459778e-05,
+      "loss": 0.9394,
+      "step": 3530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.845401345095642e-05,
+      "loss": 1.0399,
+      "step": 3540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8449646257315054e-05,
+      "loss": 0.9975,
+      "step": 3550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.844527906367369e-05,
+      "loss": 1.0759,
+      "step": 3560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.844091187003232e-05,
+      "loss": 0.8918,
+      "step": 3570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.843654467639095e-05,
+      "loss": 1.0044,
+      "step": 3580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8432177482749587e-05,
+      "loss": 0.9593,
+      "step": 3590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.842781028910822e-05,
+      "loss": 0.8939,
+      "step": 3600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.842344309546685e-05,
+      "loss": 1.0201,
+      "step": 3610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8419075901825486e-05,
+      "loss": 0.9759,
+      "step": 3620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8414708708184126e-05,
+      "loss": 0.9575,
+      "step": 3630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.841034151454276e-05,
+      "loss": 0.9803,
+      "step": 3640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.840597432090139e-05,
+      "loss": 1.0374,
+      "step": 3650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8401607127260026e-05,
+      "loss": 1.0574,
+      "step": 3660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.839723993361866e-05,
+      "loss": 0.9993,
+      "step": 3670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.839287273997729e-05,
+      "loss": 0.9252,
+      "step": 3680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8388505546335925e-05,
+      "loss": 0.9333,
+      "step": 3690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8384138352694565e-05,
+      "loss": 0.9202,
+      "step": 3700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.837977115905319e-05,
+      "loss": 0.9498,
+      "step": 3710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.837540396541183e-05,
+      "loss": 0.9133,
+      "step": 3720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8371036771770465e-05,
+      "loss": 0.9092,
+      "step": 3730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.83666695781291e-05,
+      "loss": 0.9405,
+      "step": 3740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.836230238448773e-05,
+      "loss": 0.9134,
+      "step": 3750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8357935190846365e-05,
+      "loss": 0.9513,
+      "step": 3760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8353567997205e-05,
+      "loss": 1.0138,
+      "step": 3770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.834920080356363e-05,
+      "loss": 0.9744,
+      "step": 3780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.834483360992227e-05,
+      "loss": 0.8957,
+      "step": 3790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.83404664162809e-05,
+      "loss": 0.9095,
+      "step": 3800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.833609922263954e-05,
+      "loss": 0.9416,
+      "step": 3810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8331732028998164e-05,
+      "loss": 0.9695,
+      "step": 3820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8327364835356804e-05,
+      "loss": 0.9969,
+      "step": 3830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.832299764171544e-05,
+      "loss": 0.9227,
+      "step": 3840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.831863044807407e-05,
+      "loss": 0.9607,
+      "step": 3850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.83142632544327e-05,
+      "loss": 1.0871,
+      "step": 3860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8309896060791337e-05,
+      "loss": 0.9033,
+      "step": 3870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8305528867149977e-05,
+      "loss": 0.9323,
+      "step": 3880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.83011616735086e-05,
+      "loss": 0.9549,
+      "step": 3890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.829679447986724e-05,
+      "loss": 0.9288,
+      "step": 3900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.829242728622587e-05,
+      "loss": 0.9792,
+      "step": 3910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.828806009258451e-05,
+      "loss": 0.9596,
+      "step": 3920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.828369289894314e-05,
+      "loss": 0.9292,
+      "step": 3930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8279325705301776e-05,
+      "loss": 0.9185,
+      "step": 3940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.827495851166041e-05,
+      "loss": 0.9718,
+      "step": 3950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.827059131801904e-05,
+      "loss": 0.9121,
+      "step": 3960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.826622412437768e-05,
+      "loss": 1.0239,
+      "step": 3970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.826185693073631e-05,
+      "loss": 0.9005,
+      "step": 3980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.825748973709495e-05,
+      "loss": 0.8409,
+      "step": 3990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8253122543453575e-05,
+      "loss": 0.9646,
+      "step": 4000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8248755349812215e-05,
+      "loss": 0.8558,
+      "step": 4010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.824438815617084e-05,
+      "loss": 0.9921,
+      "step": 4020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.824002096252948e-05,
+      "loss": 0.9726,
+      "step": 4030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8235653768888115e-05,
+      "loss": 0.9622,
+      "step": 4040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.823128657524675e-05,
+      "loss": 0.8935,
+      "step": 4050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.822691938160539e-05,
+      "loss": 0.9742,
+      "step": 4060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8222552187964014e-05,
+      "loss": 0.937,
+      "step": 4070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8218184994322654e-05,
+      "loss": 0.938,
+      "step": 4080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.821381780068128e-05,
+      "loss": 1.0625,
+      "step": 4090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.820945060703992e-05,
+      "loss": 0.8882,
+      "step": 4100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8205083413398554e-05,
+      "loss": 0.9611,
+      "step": 4110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.820071621975719e-05,
+      "loss": 0.9687,
+      "step": 4120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.819634902611582e-05,
+      "loss": 1.0842,
+      "step": 4130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.819198183247445e-05,
+      "loss": 0.9597,
+      "step": 4140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8187614638833087e-05,
+      "loss": 0.9726,
+      "step": 4150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.818324744519172e-05,
+      "loss": 0.9093,
+      "step": 4160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.817888025155036e-05,
+      "loss": 0.9404,
+      "step": 4170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8174513057908986e-05,
+      "loss": 0.9821,
+      "step": 4180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8170145864267626e-05,
+      "loss": 0.9522,
+      "step": 4190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.816577867062626e-05,
+      "loss": 0.782,
+      "step": 4200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.816141147698489e-05,
+      "loss": 0.9857,
+      "step": 4210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8157044283343526e-05,
+      "loss": 1.0036,
+      "step": 4220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.815267708970216e-05,
+      "loss": 0.9922,
+      "step": 4230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.814830989606079e-05,
+      "loss": 1.0202,
+      "step": 4240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8143942702419425e-05,
+      "loss": 0.9557,
+      "step": 4250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8139575508778065e-05,
+      "loss": 1.0429,
+      "step": 4260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.813520831513669e-05,
+      "loss": 0.9351,
+      "step": 4270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.813084112149533e-05,
+      "loss": 1.027,
+      "step": 4280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8126473927853965e-05,
+      "loss": 0.964,
+      "step": 4290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.81221067342126e-05,
+      "loss": 1.0001,
+      "step": 4300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.811773954057123e-05,
+      "loss": 0.8441,
+      "step": 4310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8113372346929865e-05,
+      "loss": 0.9485,
+      "step": 4320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.81090051532885e-05,
+      "loss": 0.9037,
+      "step": 4330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.810463795964713e-05,
+      "loss": 0.9977,
+      "step": 4340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8100270766005764e-05,
+      "loss": 1.0251,
+      "step": 4350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8095903572364404e-05,
+      "loss": 0.8914,
+      "step": 4360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.809153637872304e-05,
+      "loss": 0.9359,
+      "step": 4370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.808716918508167e-05,
+      "loss": 0.9597,
+      "step": 4380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8082801991440304e-05,
+      "loss": 0.9289,
+      "step": 4390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.807843479779894e-05,
+      "loss": 0.9925,
+      "step": 4400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.807406760415757e-05,
+      "loss": 0.8852,
+      "step": 4410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.80697004105162e-05,
+      "loss": 1.0007,
+      "step": 4420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8065333216874837e-05,
+      "loss": 0.9665,
+      "step": 4430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.806096602323347e-05,
+      "loss": 0.8963,
+      "step": 4440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.805659882959211e-05,
+      "loss": 0.9186,
+      "step": 4450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.805223163595074e-05,
+      "loss": 0.9967,
+      "step": 4460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8047864442309376e-05,
+      "loss": 0.8759,
+      "step": 4470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.804349724866801e-05,
+      "loss": 0.9268,
+      "step": 4480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.803913005502664e-05,
+      "loss": 1.068,
+      "step": 4490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8034762861385276e-05,
+      "loss": 0.8693,
+      "step": 4500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.803039566774391e-05,
+      "loss": 0.9903,
+      "step": 4510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.802602847410254e-05,
+      "loss": 0.9137,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8021661280461175e-05,
+      "loss": 1.0447,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8017294086819815e-05,
+      "loss": 0.985,
+      "step": 4540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.801292689317844e-05,
+      "loss": 0.9566,
+      "step": 4550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.800855969953708e-05,
+      "loss": 0.8894,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8004192505895715e-05,
+      "loss": 0.9889,
+      "step": 4570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.799982531225435e-05,
+      "loss": 0.9232,
+      "step": 4580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.799545811861298e-05,
+      "loss": 1.0342,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7991090924971615e-05,
+      "loss": 0.8374,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7986723731330255e-05,
+      "loss": 0.8928,
+      "step": 4610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.798235653768888e-05,
+      "loss": 0.8429,
+      "step": 4620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.797798934404752e-05,
+      "loss": 1.0211,
+      "step": 4630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.797362215040615e-05,
+      "loss": 1.0058,
+      "step": 4640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.796925495676479e-05,
+      "loss": 0.9645,
+      "step": 4650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.796488776312342e-05,
+      "loss": 0.956,
+      "step": 4660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7960520569482054e-05,
+      "loss": 1.0115,
+      "step": 4670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.795615337584069e-05,
+      "loss": 1.0118,
+      "step": 4680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.795178618219932e-05,
+      "loss": 0.9815,
+      "step": 4690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.794741898855796e-05,
+      "loss": 0.8431,
+      "step": 4700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7943051794916587e-05,
+      "loss": 0.928,
+      "step": 4710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7938684601275227e-05,
+      "loss": 0.8808,
+      "step": 4720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.793431740763385e-05,
+      "loss": 0.9732,
+      "step": 4730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.792995021399249e-05,
+      "loss": 0.8384,
+      "step": 4740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.792558302035112e-05,
+      "loss": 1.0879,
+      "step": 4750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.792121582670976e-05,
+      "loss": 0.8439,
+      "step": 4760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.791684863306839e-05,
+      "loss": 0.9156,
+      "step": 4770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7912481439427026e-05,
+      "loss": 0.833,
+      "step": 4780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7908114245785666e-05,
+      "loss": 0.9546,
+      "step": 4790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.790374705214429e-05,
+      "loss": 0.8693,
+      "step": 4800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.789937985850293e-05,
+      "loss": 1.1034,
+      "step": 4810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.789501266486156e-05,
+      "loss": 1.0163,
+      "step": 4820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.78906454712202e-05,
+      "loss": 0.9428,
+      "step": 4830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7886278277578825e-05,
+      "loss": 0.9196,
+      "step": 4840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7881911083937465e-05,
+      "loss": 0.8901,
+      "step": 4850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.78775438902961e-05,
+      "loss": 0.9336,
+      "step": 4860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.787317669665473e-05,
+      "loss": 0.9136,
+      "step": 4870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7868809503013365e-05,
+      "loss": 1.0059,
+      "step": 4880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7864442309372e-05,
+      "loss": 0.8842,
+      "step": 4890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.786007511573064e-05,
+      "loss": 0.8995,
+      "step": 4900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7855707922089264e-05,
+      "loss": 0.888,
+      "step": 4910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7851340728447904e-05,
+      "loss": 0.9411,
+      "step": 4920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.784697353480653e-05,
+      "loss": 1.0011,
+      "step": 4930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.784260634116517e-05,
+      "loss": 0.9201,
+      "step": 4940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7838239147523804e-05,
+      "loss": 0.9689,
+      "step": 4950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.783387195388244e-05,
+      "loss": 0.9215,
+      "step": 4960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.782950476024107e-05,
+      "loss": 0.8547,
+      "step": 4970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7825137566599703e-05,
+      "loss": 0.9999,
+      "step": 4980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.782077037295834e-05,
+      "loss": 0.8561,
+      "step": 4990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.781640317931697e-05,
+      "loss": 0.915,
+      "step": 5000
+    },
+    {
+      "epoch": 0.44,
+      "eval_accuracy": 0.5413186597229046,
+      "eval_loss": 0.9457740783691406,
+      "eval_runtime": 84.182,
+      "eval_samples_per_second": 120.893,
+      "eval_steps_per_second": 15.122,
+      "step": 5000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.781203598567561e-05,
+      "loss": 1.0026,
+      "step": 5010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.780766879203424e-05,
+      "loss": 1.1102,
+      "step": 5020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7803301598392876e-05,
+      "loss": 0.9471,
+      "step": 5030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.779893440475151e-05,
+      "loss": 1.0117,
+      "step": 5040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.779456721111014e-05,
+      "loss": 0.9408,
+      "step": 5050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7790200017468776e-05,
+      "loss": 0.9945,
+      "step": 5060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.778583282382741e-05,
+      "loss": 0.9939,
+      "step": 5070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.778146563018604e-05,
+      "loss": 0.9611,
+      "step": 5080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7777098436544675e-05,
+      "loss": 0.9477,
+      "step": 5090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7772731242903315e-05,
+      "loss": 0.9704,
+      "step": 5100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.776836404926195e-05,
+      "loss": 1.046,
+      "step": 5110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.776399685562058e-05,
+      "loss": 0.9343,
+      "step": 5120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7759629661979215e-05,
+      "loss": 0.8697,
+      "step": 5130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.775526246833785e-05,
+      "loss": 0.9833,
+      "step": 5140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.775089527469648e-05,
+      "loss": 1.0198,
+      "step": 5150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7746528081055115e-05,
+      "loss": 0.9693,
+      "step": 5160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.774216088741375e-05,
+      "loss": 0.8275,
+      "step": 5170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.773779369377239e-05,
+      "loss": 0.9569,
+      "step": 5180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.773342650013102e-05,
+      "loss": 0.907,
+      "step": 5190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7729059306489654e-05,
+      "loss": 0.9482,
+      "step": 5200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772469211284829e-05,
+      "loss": 0.9195,
+      "step": 5210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772032491920692e-05,
+      "loss": 0.7862,
+      "step": 5220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7715957725565554e-05,
+      "loss": 0.9334,
+      "step": 5230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.771159053192419e-05,
+      "loss": 0.8626,
+      "step": 5240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.770722333828282e-05,
+      "loss": 0.9073,
+      "step": 5250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7702856144641453e-05,
+      "loss": 1.0073,
+      "step": 5260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.769848895100009e-05,
+      "loss": 1.1421,
+      "step": 5270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.769412175735872e-05,
+      "loss": 0.8536,
+      "step": 5280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.768975456371736e-05,
+      "loss": 1.0058,
+      "step": 5290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.768538737007599e-05,
+      "loss": 0.9426,
+      "step": 5300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7681020176434626e-05,
+      "loss": 0.9221,
+      "step": 5310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.767665298279326e-05,
+      "loss": 0.9836,
+      "step": 5320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.767228578915189e-05,
+      "loss": 0.9412,
+      "step": 5330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7667918595510526e-05,
+      "loss": 0.9653,
+      "step": 5340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.766355140186916e-05,
+      "loss": 0.9413,
+      "step": 5350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.76591842082278e-05,
+      "loss": 0.8595,
+      "step": 5360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7654817014586425e-05,
+      "loss": 0.9277,
+      "step": 5370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7650449820945065e-05,
+      "loss": 0.9825,
+      "step": 5380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.76460826273037e-05,
+      "loss": 0.9038,
+      "step": 5390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.764171543366233e-05,
+      "loss": 0.9425,
+      "step": 5400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7637348240020965e-05,
+      "loss": 1.0464,
+      "step": 5410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.76329810463796e-05,
+      "loss": 0.9711,
+      "step": 5420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.762861385273824e-05,
+      "loss": 0.9361,
+      "step": 5430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7624246659096865e-05,
+      "loss": 1.003,
+      "step": 5440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7619879465455505e-05,
+      "loss": 0.9661,
+      "step": 5450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.761551227181413e-05,
+      "loss": 0.9532,
+      "step": 5460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.761114507817277e-05,
+      "loss": 0.9735,
+      "step": 5470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.76067778845314e-05,
+      "loss": 0.9799,
+      "step": 5480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.760241069089004e-05,
+      "loss": 0.8876,
+      "step": 5490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.759804349724867e-05,
+      "loss": 0.9545,
+      "step": 5500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7593676303607304e-05,
+      "loss": 0.9032,
+      "step": 5510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7589309109965944e-05,
+      "loss": 0.9161,
+      "step": 5520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.758494191632457e-05,
+      "loss": 0.9136,
+      "step": 5530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.758057472268321e-05,
+      "loss": 0.935,
+      "step": 5540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.757620752904184e-05,
+      "loss": 0.8862,
+      "step": 5550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7571840335400477e-05,
+      "loss": 0.9883,
+      "step": 5560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.75674731417591e-05,
+      "loss": 0.9101,
+      "step": 5570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.756310594811774e-05,
+      "loss": 0.9613,
+      "step": 5580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7558738754476376e-05,
+      "loss": 0.9583,
+      "step": 5590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.755437156083501e-05,
+      "loss": 1.0256,
+      "step": 5600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.755000436719364e-05,
+      "loss": 0.8804,
+      "step": 5610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7545637173552276e-05,
+      "loss": 1.0011,
+      "step": 5620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7541269979910916e-05,
+      "loss": 0.9145,
+      "step": 5630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.753690278626954e-05,
+      "loss": 1.0676,
+      "step": 5640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.753253559262818e-05,
+      "loss": 1.015,
+      "step": 5650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.752816839898681e-05,
+      "loss": 1.0333,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.752380120534545e-05,
+      "loss": 0.9977,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.751943401170408e-05,
+      "loss": 0.9501,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7515066818062715e-05,
+      "loss": 0.8981,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.751069962442135e-05,
+      "loss": 0.8934,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.750633243077998e-05,
+      "loss": 0.9149,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.750196523713862e-05,
+      "loss": 0.9701,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.749759804349725e-05,
+      "loss": 0.7643,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.749323084985589e-05,
+      "loss": 0.9105,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7488863656214514e-05,
+      "loss": 1.0524,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7484496462573154e-05,
+      "loss": 1.0357,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.748012926893179e-05,
+      "loss": 0.8244,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.747576207529042e-05,
+      "loss": 0.9587,
+      "step": 5780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7471394881649054e-05,
+      "loss": 1.042,
+      "step": 5790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.746702768800769e-05,
+      "loss": 0.8206,
+      "step": 5800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.746266049436632e-05,
+      "loss": 0.9515,
+      "step": 5810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7458293300724953e-05,
+      "loss": 0.9164,
+      "step": 5820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7453926107083593e-05,
+      "loss": 0.831,
+      "step": 5830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.744955891344223e-05,
+      "loss": 0.9254,
+      "step": 5840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.744519171980086e-05,
+      "loss": 0.9678,
+      "step": 5850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.744082452615949e-05,
+      "loss": 0.9928,
+      "step": 5860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7436457332518126e-05,
+      "loss": 0.9701,
+      "step": 5870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.743209013887676e-05,
+      "loss": 1.0539,
+      "step": 5880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.742772294523539e-05,
+      "loss": 0.9236,
+      "step": 5890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7423355751594026e-05,
+      "loss": 0.958,
+      "step": 5900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.741898855795266e-05,
+      "loss": 1.0044,
+      "step": 5910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.74146213643113e-05,
+      "loss": 1.0415,
+      "step": 5920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.741025417066993e-05,
+      "loss": 0.914,
+      "step": 5930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7405886977028565e-05,
+      "loss": 0.935,
+      "step": 5940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.74015197833872e-05,
+      "loss": 0.9326,
+      "step": 5950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.739715258974583e-05,
+      "loss": 0.8993,
+      "step": 5960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7392785396104465e-05,
+      "loss": 0.88,
+      "step": 5970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.73884182024631e-05,
+      "loss": 0.9679,
+      "step": 5980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.738405100882173e-05,
+      "loss": 1.1006,
+      "step": 5990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.737968381518037e-05,
+      "loss": 0.9666,
+      "step": 6000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7375316621539e-05,
+      "loss": 1.0173,
+      "step": 6010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737094942789764e-05,
+      "loss": 0.9477,
+      "step": 6020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.736658223425627e-05,
+      "loss": 0.8937,
+      "step": 6030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7362215040614904e-05,
+      "loss": 0.9316,
+      "step": 6040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.735784784697354e-05,
+      "loss": 1.0215,
+      "step": 6050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.735348065333217e-05,
+      "loss": 0.9685,
+      "step": 6060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7349113459690804e-05,
+      "loss": 0.8727,
+      "step": 6070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.734474626604944e-05,
+      "loss": 0.9117,
+      "step": 6080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.734037907240808e-05,
+      "loss": 1.0533,
+      "step": 6090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7336011878766703e-05,
+      "loss": 1.0263,
+      "step": 6100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7331644685125343e-05,
+      "loss": 0.9017,
+      "step": 6110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.732727749148398e-05,
+      "loss": 0.9113,
+      "step": 6120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732291029784261e-05,
+      "loss": 0.8363,
+      "step": 6130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.731854310420124e-05,
+      "loss": 0.9279,
+      "step": 6140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7314175910559876e-05,
+      "loss": 0.9293,
+      "step": 6150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.730980871691851e-05,
+      "loss": 0.884,
+      "step": 6160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.730544152327714e-05,
+      "loss": 0.9532,
+      "step": 6170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.730107432963578e-05,
+      "loss": 0.8419,
+      "step": 6180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.729670713599441e-05,
+      "loss": 0.8287,
+      "step": 6190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.729233994235305e-05,
+      "loss": 0.826,
+      "step": 6200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7287972748711675e-05,
+      "loss": 0.9862,
+      "step": 6210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7283605555070315e-05,
+      "loss": 0.8946,
+      "step": 6220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.727923836142895e-05,
+      "loss": 1.0277,
+      "step": 6230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727487116778758e-05,
+      "loss": 0.885,
+      "step": 6240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727050397414622e-05,
+      "loss": 0.9274,
+      "step": 6250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.726613678050485e-05,
+      "loss": 0.9631,
+      "step": 6260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.726176958686349e-05,
+      "loss": 0.9603,
+      "step": 6270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7257402393222115e-05,
+      "loss": 0.8425,
+      "step": 6280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7253035199580755e-05,
+      "loss": 0.8951,
+      "step": 6290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.724866800593938e-05,
+      "loss": 0.8985,
+      "step": 6300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.724430081229802e-05,
+      "loss": 1.0314,
+      "step": 6310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7239933618656654e-05,
+      "loss": 0.968,
+      "step": 6320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.723556642501529e-05,
+      "loss": 0.9126,
+      "step": 6330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.723119923137392e-05,
+      "loss": 0.9209,
+      "step": 6340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7226832037732554e-05,
+      "loss": 0.9288,
+      "step": 6350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7222464844091194e-05,
+      "loss": 0.9363,
+      "step": 6360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.721809765044982e-05,
+      "loss": 0.9826,
+      "step": 6370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.721373045680846e-05,
+      "loss": 1.0211,
+      "step": 6380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.720936326316709e-05,
+      "loss": 1.0569,
+      "step": 6390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.720499606952573e-05,
+      "loss": 0.9363,
+      "step": 6400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.720062887588435e-05,
+      "loss": 0.8699,
+      "step": 6410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.719626168224299e-05,
+      "loss": 1.0077,
+      "step": 6420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7191894488601626e-05,
+      "loss": 1.0768,
+      "step": 6430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.718752729496026e-05,
+      "loss": 0.9443,
+      "step": 6440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.71831601013189e-05,
+      "loss": 0.8926,
+      "step": 6450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7178792907677526e-05,
+      "loss": 0.885,
+      "step": 6460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7174425714036166e-05,
+      "loss": 1.0487,
+      "step": 6470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.717005852039479e-05,
+      "loss": 1.0416,
+      "step": 6480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.716569132675343e-05,
+      "loss": 0.9057,
+      "step": 6490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7161324133112065e-05,
+      "loss": 0.9122,
+      "step": 6500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.71569569394707e-05,
+      "loss": 0.8959,
+      "step": 6510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.715258974582933e-05,
+      "loss": 0.9956,
+      "step": 6520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7148222552187965e-05,
+      "loss": 0.9792,
+      "step": 6530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.71438553585466e-05,
+      "loss": 0.8862,
+      "step": 6540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.713948816490523e-05,
+      "loss": 0.9126,
+      "step": 6550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.713512097126387e-05,
+      "loss": 0.9894,
+      "step": 6560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.71307537776225e-05,
+      "loss": 0.938,
+      "step": 6570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.712638658398114e-05,
+      "loss": 0.9393,
+      "step": 6580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712201939033977e-05,
+      "loss": 0.9629,
+      "step": 6590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7117652196698404e-05,
+      "loss": 0.873,
+      "step": 6600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.711328500305704e-05,
+      "loss": 0.8973,
+      "step": 6610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.710891780941567e-05,
+      "loss": 0.9372,
+      "step": 6620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7104550615774304e-05,
+      "loss": 0.9139,
+      "step": 6630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.710018342213294e-05,
+      "loss": 0.9415,
+      "step": 6640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.709581622849158e-05,
+      "loss": 0.9936,
+      "step": 6650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.709144903485021e-05,
+      "loss": 0.9515,
+      "step": 6660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7087081841208843e-05,
+      "loss": 0.9849,
+      "step": 6670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.708271464756748e-05,
+      "loss": 1.0061,
+      "step": 6680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.707834745392611e-05,
+      "loss": 0.9298,
+      "step": 6690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707398026028474e-05,
+      "loss": 0.9501,
+      "step": 6700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7069613066643376e-05,
+      "loss": 0.9054,
+      "step": 6710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.706524587300201e-05,
+      "loss": 0.9472,
+      "step": 6720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.706087867936064e-05,
+      "loss": 0.924,
+      "step": 6730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7056511485719276e-05,
+      "loss": 0.935,
+      "step": 6740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7052144292077916e-05,
+      "loss": 0.9046,
+      "step": 6750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.704777709843655e-05,
+      "loss": 0.9272,
+      "step": 6760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.704340990479518e-05,
+      "loss": 0.9817,
+      "step": 6770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7039042711153815e-05,
+      "loss": 0.919,
+      "step": 6780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.703467551751245e-05,
+      "loss": 0.8867,
+      "step": 6790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.703030832387108e-05,
+      "loss": 1.0659,
+      "step": 6800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7025941130229715e-05,
+      "loss": 1.0513,
+      "step": 6810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.702157393658835e-05,
+      "loss": 0.9983,
+      "step": 6820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.701720674294698e-05,
+      "loss": 0.889,
+      "step": 6830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.701283954930562e-05,
+      "loss": 1.0621,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7008472355664255e-05,
+      "loss": 1.0154,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.700410516202289e-05,
+      "loss": 0.9303,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.699973796838152e-05,
+      "loss": 0.9811,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6995370774740154e-05,
+      "loss": 0.9383,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.699100358109879e-05,
+      "loss": 0.8855,
+      "step": 6890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.698663638745742e-05,
+      "loss": 0.9079,
+      "step": 6900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.698226919381606e-05,
+      "loss": 0.9674,
+      "step": 6910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.697790200017469e-05,
+      "loss": 0.9785,
+      "step": 6920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.697353480653333e-05,
+      "loss": 1.0726,
+      "step": 6930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6969167612891953e-05,
+      "loss": 0.9426,
+      "step": 6940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6964800419250593e-05,
+      "loss": 0.9499,
+      "step": 6950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.696043322560923e-05,
+      "loss": 1.0167,
+      "step": 6960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.695606603196786e-05,
+      "loss": 0.9911,
+      "step": 6970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.695169883832649e-05,
+      "loss": 0.8555,
+      "step": 6980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6947331644685126e-05,
+      "loss": 0.9792,
+      "step": 6990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6942964451043766e-05,
+      "loss": 0.8869,
+      "step": 7000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.693859725740239e-05,
+      "loss": 0.9446,
+      "step": 7010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.693423006376103e-05,
+      "loss": 0.9355,
+      "step": 7020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.692986287011966e-05,
+      "loss": 0.9304,
+      "step": 7030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.69254956764783e-05,
+      "loss": 0.8644,
+      "step": 7040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.692112848283693e-05,
+      "loss": 0.9132,
+      "step": 7050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6916761289195565e-05,
+      "loss": 0.9018,
+      "step": 7060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69123940955542e-05,
+      "loss": 0.9422,
+      "step": 7070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.690802690191283e-05,
+      "loss": 1.0124,
+      "step": 7080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.690365970827147e-05,
+      "loss": 0.9288,
+      "step": 7090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.68992925146301e-05,
+      "loss": 0.944,
+      "step": 7100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.689492532098874e-05,
+      "loss": 0.9644,
+      "step": 7110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6890558127347365e-05,
+      "loss": 0.9294,
+      "step": 7120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6886190933706005e-05,
+      "loss": 0.8939,
+      "step": 7130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.688182374006463e-05,
+      "loss": 0.8937,
+      "step": 7140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.687745654642327e-05,
+      "loss": 0.9284,
+      "step": 7150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6873089352781904e-05,
+      "loss": 0.9391,
+      "step": 7160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.686872215914054e-05,
+      "loss": 1.059,
+      "step": 7170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.686435496549918e-05,
+      "loss": 0.9478,
+      "step": 7180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6859987771857804e-05,
+      "loss": 0.9316,
+      "step": 7190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6855620578216444e-05,
+      "loss": 0.8847,
+      "step": 7200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.685125338457507e-05,
+      "loss": 0.8741,
+      "step": 7210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.684688619093371e-05,
+      "loss": 0.9435,
+      "step": 7220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.684251899729234e-05,
+      "loss": 0.9314,
+      "step": 7230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.683815180365098e-05,
+      "loss": 0.9677,
+      "step": 7240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.683378461000961e-05,
+      "loss": 0.9047,
+      "step": 7250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.682941741636824e-05,
+      "loss": 0.89,
+      "step": 7260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6825050222726876e-05,
+      "loss": 0.8205,
+      "step": 7270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.682068302908551e-05,
+      "loss": 1.0917,
+      "step": 7280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.681631583544415e-05,
+      "loss": 0.9073,
+      "step": 7290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6811948641802776e-05,
+      "loss": 1.0848,
+      "step": 7300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6807581448161416e-05,
+      "loss": 0.9949,
+      "step": 7310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.680321425452005e-05,
+      "loss": 0.9402,
+      "step": 7320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.679884706087868e-05,
+      "loss": 0.8855,
+      "step": 7330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6794479867237315e-05,
+      "loss": 1.0907,
+      "step": 7340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.679011267359595e-05,
+      "loss": 1.0442,
+      "step": 7350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.678574547995458e-05,
+      "loss": 0.9652,
+      "step": 7360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6781378286313215e-05,
+      "loss": 0.9912,
+      "step": 7370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6777011092671855e-05,
+      "loss": 0.9111,
+      "step": 7380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.677264389903048e-05,
+      "loss": 0.9201,
+      "step": 7390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.676827670538912e-05,
+      "loss": 0.838,
+      "step": 7400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6763909511747755e-05,
+      "loss": 0.8992,
+      "step": 7410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.675954231810639e-05,
+      "loss": 0.9815,
+      "step": 7420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.675517512446502e-05,
+      "loss": 1.0163,
+      "step": 7430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6750807930823654e-05,
+      "loss": 0.9521,
+      "step": 7440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.674644073718229e-05,
+      "loss": 0.9802,
+      "step": 7450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.674207354354092e-05,
+      "loss": 0.9717,
+      "step": 7460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6737706349899554e-05,
+      "loss": 0.849,
+      "step": 7470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6733339156258194e-05,
+      "loss": 0.8842,
+      "step": 7480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.672897196261683e-05,
+      "loss": 0.9432,
+      "step": 7490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.672460476897546e-05,
+      "loss": 0.9463,
+      "step": 7500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6720237575334093e-05,
+      "loss": 0.9011,
+      "step": 7510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.671587038169273e-05,
+      "loss": 0.9069,
+      "step": 7520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.671150318805136e-05,
+      "loss": 0.8761,
+      "step": 7530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.670713599440999e-05,
+      "loss": 0.8911,
+      "step": 7540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6702768800768626e-05,
+      "loss": 0.891,
+      "step": 7550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.669840160712726e-05,
+      "loss": 1.0597,
+      "step": 7560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.66940344134859e-05,
+      "loss": 0.9485,
+      "step": 7570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.668966721984453e-05,
+      "loss": 0.9077,
+      "step": 7580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6685300026203166e-05,
+      "loss": 1.0007,
+      "step": 7590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.66809328325618e-05,
+      "loss": 0.9607,
+      "step": 7600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.667656563892043e-05,
+      "loss": 0.879,
+      "step": 7610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6672198445279065e-05,
+      "loss": 0.9697,
+      "step": 7620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.66678312516377e-05,
+      "loss": 0.9778,
+      "step": 7630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.666346405799633e-05,
+      "loss": 0.9063,
+      "step": 7640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6659096864354965e-05,
+      "loss": 0.9382,
+      "step": 7650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6654729670713605e-05,
+      "loss": 0.8835,
+      "step": 7660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.665036247707223e-05,
+      "loss": 0.9242,
+      "step": 7670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.664599528343087e-05,
+      "loss": 0.9049,
+      "step": 7680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6641628089789505e-05,
+      "loss": 0.9353,
+      "step": 7690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.663726089614814e-05,
+      "loss": 0.9879,
+      "step": 7700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.663289370250677e-05,
+      "loss": 0.9736,
+      "step": 7710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6628526508865404e-05,
+      "loss": 0.9421,
+      "step": 7720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6624159315224044e-05,
+      "loss": 0.951,
+      "step": 7730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.661979212158267e-05,
+      "loss": 1.0041,
+      "step": 7740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.661542492794131e-05,
+      "loss": 0.9923,
+      "step": 7750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.661105773429994e-05,
+      "loss": 1.0187,
+      "step": 7760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.660669054065858e-05,
+      "loss": 0.9704,
+      "step": 7770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.660232334701721e-05,
+      "loss": 0.9147,
+      "step": 7780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6597956153375843e-05,
+      "loss": 0.9098,
+      "step": 7790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.659358895973448e-05,
+      "loss": 0.8603,
+      "step": 7800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.658922176609311e-05,
+      "loss": 0.9084,
+      "step": 7810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.658485457245175e-05,
+      "loss": 0.943,
+      "step": 7820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6580487378810376e-05,
+      "loss": 1.0826,
+      "step": 7830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6576120185169016e-05,
+      "loss": 0.8726,
+      "step": 7840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657175299152764e-05,
+      "loss": 0.8969,
+      "step": 7850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.656738579788628e-05,
+      "loss": 0.9451,
+      "step": 7860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.656301860424491e-05,
+      "loss": 0.8773,
+      "step": 7870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655865141060355e-05,
+      "loss": 0.8431,
+      "step": 7880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655428421696218e-05,
+      "loss": 1.0478,
+      "step": 7890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6549917023320816e-05,
+      "loss": 1.0526,
+      "step": 7900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6545549829679455e-05,
+      "loss": 0.8925,
+      "step": 7910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654118263603808e-05,
+      "loss": 0.9562,
+      "step": 7920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653681544239672e-05,
+      "loss": 0.9244,
+      "step": 7930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653244824875535e-05,
+      "loss": 0.9023,
+      "step": 7940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.652808105511399e-05,
+      "loss": 0.899,
+      "step": 7950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6523713861472615e-05,
+      "loss": 0.8497,
+      "step": 7960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6519346667831255e-05,
+      "loss": 0.9133,
+      "step": 7970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.651497947418989e-05,
+      "loss": 0.9295,
+      "step": 7980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.651061228054852e-05,
+      "loss": 1.0113,
+      "step": 7990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6506245086907154e-05,
+      "loss": 0.9535,
+      "step": 8000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.650187789326579e-05,
+      "loss": 0.8836,
+      "step": 8010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.649751069962443e-05,
+      "loss": 0.9482,
+      "step": 8020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6493143505983054e-05,
+      "loss": 0.9013,
+      "step": 8030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6488776312341694e-05,
+      "loss": 1.0265,
+      "step": 8040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.648440911870032e-05,
+      "loss": 0.8934,
+      "step": 8050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.648004192505896e-05,
+      "loss": 1.0311,
+      "step": 8060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6475674731417594e-05,
+      "loss": 0.828,
+      "step": 8070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647130753777623e-05,
+      "loss": 0.9535,
+      "step": 8080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.646694034413486e-05,
+      "loss": 0.9586,
+      "step": 8090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.646257315049349e-05,
+      "loss": 0.8356,
+      "step": 8100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.645820595685213e-05,
+      "loss": 0.9981,
+      "step": 8110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.645383876321076e-05,
+      "loss": 0.9218,
+      "step": 8120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.64494715695694e-05,
+      "loss": 0.9265,
+      "step": 8130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.644510437592803e-05,
+      "loss": 0.9731,
+      "step": 8140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6440737182286666e-05,
+      "loss": 0.9636,
+      "step": 8150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.64363699886453e-05,
+      "loss": 0.9791,
+      "step": 8160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.643200279500393e-05,
+      "loss": 0.9366,
+      "step": 8170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6427635601362566e-05,
+      "loss": 0.9753,
+      "step": 8180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.64232684077212e-05,
+      "loss": 0.8684,
+      "step": 8190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.641890121407983e-05,
+      "loss": 1.0108,
+      "step": 8200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6414534020438465e-05,
+      "loss": 0.9217,
+      "step": 8210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6410166826797105e-05,
+      "loss": 0.9008,
+      "step": 8220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.640579963315574e-05,
+      "loss": 0.8441,
+      "step": 8230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.640143243951437e-05,
+      "loss": 0.8878,
+      "step": 8240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6397065245873005e-05,
+      "loss": 0.9021,
+      "step": 8250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.639269805223164e-05,
+      "loss": 0.877,
+      "step": 8260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.638833085859027e-05,
+      "loss": 0.9297,
+      "step": 8270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6383963664948904e-05,
+      "loss": 1.0486,
+      "step": 8280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.637959647130754e-05,
+      "loss": 0.8648,
+      "step": 8290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.637522927766617e-05,
+      "loss": 0.8761,
+      "step": 8300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.637086208402481e-05,
+      "loss": 0.88,
+      "step": 8310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6366494890383444e-05,
+      "loss": 0.8634,
+      "step": 8320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.636212769674208e-05,
+      "loss": 0.9119,
+      "step": 8330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.635776050310071e-05,
+      "loss": 0.8945,
+      "step": 8340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6353393309459344e-05,
+      "loss": 0.8123,
+      "step": 8350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.634902611581798e-05,
+      "loss": 0.9114,
+      "step": 8360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.634465892217661e-05,
+      "loss": 0.9065,
+      "step": 8370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.634029172853524e-05,
+      "loss": 0.7807,
+      "step": 8380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.633592453489388e-05,
+      "loss": 0.9842,
+      "step": 8390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.633155734125251e-05,
+      "loss": 0.9122,
+      "step": 8400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.632719014761115e-05,
+      "loss": 0.9702,
+      "step": 8410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.632282295396978e-05,
+      "loss": 1.0104,
+      "step": 8420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6318455760328416e-05,
+      "loss": 1.0367,
+      "step": 8430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.631408856668705e-05,
+      "loss": 0.9515,
+      "step": 8440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.630972137304568e-05,
+      "loss": 0.9196,
+      "step": 8450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6305354179404316e-05,
+      "loss": 0.9348,
+      "step": 8460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.630098698576295e-05,
+      "loss": 0.9261,
+      "step": 8470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.629661979212159e-05,
+      "loss": 0.8827,
+      "step": 8480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6292252598480215e-05,
+      "loss": 0.8363,
+      "step": 8490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6287885404838855e-05,
+      "loss": 0.9509,
+      "step": 8500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.628351821119749e-05,
+      "loss": 0.9425,
+      "step": 8510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.627915101755612e-05,
+      "loss": 1.0142,
+      "step": 8520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6274783823914755e-05,
+      "loss": 0.9711,
+      "step": 8530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.627041663027339e-05,
+      "loss": 0.9595,
+      "step": 8540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.626604943663203e-05,
+      "loss": 0.9476,
+      "step": 8550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6261682242990654e-05,
+      "loss": 0.9114,
+      "step": 8560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6257315049349294e-05,
+      "loss": 0.9543,
+      "step": 8570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.625294785570792e-05,
+      "loss": 0.9228,
+      "step": 8580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.624858066206656e-05,
+      "loss": 0.9868,
+      "step": 8590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.624421346842519e-05,
+      "loss": 0.8871,
+      "step": 8600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.623984627478383e-05,
+      "loss": 0.8485,
+      "step": 8610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.623547908114246e-05,
+      "loss": 0.9434,
+      "step": 8620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6231111887501094e-05,
+      "loss": 0.9924,
+      "step": 8630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6226744693859734e-05,
+      "loss": 0.8118,
+      "step": 8640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622237750021836e-05,
+      "loss": 0.9211,
+      "step": 8650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6218010306577e-05,
+      "loss": 0.9975,
+      "step": 8660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6213643112935626e-05,
+      "loss": 0.9707,
+      "step": 8670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6209275919294266e-05,
+      "loss": 0.8622,
+      "step": 8680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.620490872565289e-05,
+      "loss": 1.022,
+      "step": 8690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.620054153201153e-05,
+      "loss": 0.9881,
+      "step": 8700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6196174338370166e-05,
+      "loss": 0.8869,
+      "step": 8710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.61918071447288e-05,
+      "loss": 1.0154,
+      "step": 8720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.618743995108743e-05,
+      "loss": 0.9508,
+      "step": 8730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6183072757446066e-05,
+      "loss": 0.8722,
+      "step": 8740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6178705563804706e-05,
+      "loss": 0.9598,
+      "step": 8750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.617433837016333e-05,
+      "loss": 0.8668,
+      "step": 8760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.616997117652197e-05,
+      "loss": 0.9405,
+      "step": 8770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.61656039828806e-05,
+      "loss": 0.846,
+      "step": 8780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.616123678923924e-05,
+      "loss": 0.9587,
+      "step": 8790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.615686959559787e-05,
+      "loss": 0.8373,
+      "step": 8800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6152502401956505e-05,
+      "loss": 0.8456,
+      "step": 8810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.614813520831514e-05,
+      "loss": 0.9896,
+      "step": 8820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.614376801467377e-05,
+      "loss": 1.0811,
+      "step": 8830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.613940082103241e-05,
+      "loss": 0.997,
+      "step": 8840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.613503362739104e-05,
+      "loss": 0.982,
+      "step": 8850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.613066643374968e-05,
+      "loss": 0.9422,
+      "step": 8860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6126299240108304e-05,
+      "loss": 0.8499,
+      "step": 8870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6121932046466944e-05,
+      "loss": 0.9511,
+      "step": 8880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.611756485282558e-05,
+      "loss": 0.8959,
+      "step": 8890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.611319765918421e-05,
+      "loss": 0.8872,
+      "step": 8900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6108830465542844e-05,
+      "loss": 0.9708,
+      "step": 8910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.610446327190148e-05,
+      "loss": 0.8085,
+      "step": 8920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.610009607826011e-05,
+      "loss": 0.9264,
+      "step": 8930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.609572888461874e-05,
+      "loss": 0.8876,
+      "step": 8940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.609136169097738e-05,
+      "loss": 1.0197,
+      "step": 8950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6086994497336016e-05,
+      "loss": 1.0333,
+      "step": 8960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.608262730369465e-05,
+      "loss": 0.8699,
+      "step": 8970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.607826011005328e-05,
+      "loss": 0.9157,
+      "step": 8980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6073892916411916e-05,
+      "loss": 0.8753,
+      "step": 8990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.606952572277055e-05,
+      "loss": 0.9993,
+      "step": 9000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.606515852912918e-05,
+      "loss": 0.8956,
+      "step": 9010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6060791335487816e-05,
+      "loss": 0.9916,
+      "step": 9020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.605642414184645e-05,
+      "loss": 0.8797,
+      "step": 9030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.605205694820509e-05,
+      "loss": 0.9454,
+      "step": 9040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.604768975456372e-05,
+      "loss": 0.8976,
+      "step": 9050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6043322560922355e-05,
+      "loss": 0.8632,
+      "step": 9060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.603895536728099e-05,
+      "loss": 0.8919,
+      "step": 9070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.603458817363962e-05,
+      "loss": 0.791,
+      "step": 9080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6030220979998255e-05,
+      "loss": 0.824,
+      "step": 9090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.602585378635689e-05,
+      "loss": 1.0017,
+      "step": 9100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.602148659271552e-05,
+      "loss": 0.909,
+      "step": 9110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6017119399074154e-05,
+      "loss": 0.9327,
+      "step": 9120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.601275220543279e-05,
+      "loss": 0.9564,
+      "step": 9130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.600838501179143e-05,
+      "loss": 0.9171,
+      "step": 9140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.600401781815006e-05,
+      "loss": 1.0033,
+      "step": 9150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5999650624508694e-05,
+      "loss": 0.8151,
+      "step": 9160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.599528343086733e-05,
+      "loss": 0.9367,
+      "step": 9170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.599091623722596e-05,
+      "loss": 0.7902,
+      "step": 9180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5986549043584594e-05,
+      "loss": 0.8708,
+      "step": 9190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.598218184994323e-05,
+      "loss": 0.7969,
+      "step": 9200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.597781465630187e-05,
+      "loss": 1.0689,
+      "step": 9210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597344746266049e-05,
+      "loss": 1.0747,
+      "step": 9220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.596908026901913e-05,
+      "loss": 0.9492,
+      "step": 9230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5964713075377766e-05,
+      "loss": 0.9497,
+      "step": 9240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.59603458817364e-05,
+      "loss": 0.9191,
+      "step": 9250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.595597868809503e-05,
+      "loss": 0.8723,
+      "step": 9260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5951611494453666e-05,
+      "loss": 0.9379,
+      "step": 9270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.59472443008123e-05,
+      "loss": 0.9149,
+      "step": 9280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.594287710717093e-05,
+      "loss": 0.9203,
+      "step": 9290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.593850991352957e-05,
+      "loss": 0.8775,
+      "step": 9300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.59341427198882e-05,
+      "loss": 0.8603,
+      "step": 9310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.592977552624684e-05,
+      "loss": 0.978,
+      "step": 9320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5925408332605465e-05,
+      "loss": 0.8297,
+      "step": 9330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5921041138964105e-05,
+      "loss": 0.797,
+      "step": 9340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.591667394532274e-05,
+      "loss": 0.8891,
+      "step": 9350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.591230675168137e-05,
+      "loss": 0.9431,
+      "step": 9360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.590793955804001e-05,
+      "loss": 0.8584,
+      "step": 9370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.590357236439864e-05,
+      "loss": 0.9427,
+      "step": 9380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.589920517075728e-05,
+      "loss": 0.8603,
+      "step": 9390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5894837977115904e-05,
+      "loss": 0.8461,
+      "step": 9400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5890470783474544e-05,
+      "loss": 0.9328,
+      "step": 9410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.588610358983317e-05,
+      "loss": 0.9883,
+      "step": 9420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.588173639619181e-05,
+      "loss": 0.9322,
+      "step": 9430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5877369202550444e-05,
+      "loss": 0.8851,
+      "step": 9440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.587300200890908e-05,
+      "loss": 0.8376,
+      "step": 9450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.586863481526771e-05,
+      "loss": 0.8978,
+      "step": 9460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5864267621626344e-05,
+      "loss": 1.0406,
+      "step": 9470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5859900427984984e-05,
+      "loss": 0.9252,
+      "step": 9480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.585553323434361e-05,
+      "loss": 0.8589,
+      "step": 9490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.585116604070225e-05,
+      "loss": 0.9067,
+      "step": 9500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5846798847060876e-05,
+      "loss": 0.967,
+      "step": 9510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5842431653419516e-05,
+      "loss": 0.9374,
+      "step": 9520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.583806445977814e-05,
+      "loss": 0.8765,
+      "step": 9530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.583369726613678e-05,
+      "loss": 0.9199,
+      "step": 9540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5829330072495416e-05,
+      "loss": 0.8998,
+      "step": 9550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582496287885405e-05,
+      "loss": 0.8998,
+      "step": 9560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582059568521269e-05,
+      "loss": 0.8978,
+      "step": 9570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5816228491571316e-05,
+      "loss": 0.8726,
+      "step": 9580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5811861297929956e-05,
+      "loss": 0.7637,
+      "step": 9590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.580749410428858e-05,
+      "loss": 0.9338,
+      "step": 9600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.580312691064722e-05,
+      "loss": 0.9755,
+      "step": 9610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5798759717005855e-05,
+      "loss": 0.9454,
+      "step": 9620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.579439252336449e-05,
+      "loss": 0.9308,
+      "step": 9630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.579002532972312e-05,
+      "loss": 0.963,
+      "step": 9640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5785658136081755e-05,
+      "loss": 0.9523,
+      "step": 9650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.578129094244039e-05,
+      "loss": 0.8774,
+      "step": 9660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.577692374879902e-05,
+      "loss": 0.9609,
+      "step": 9670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577255655515766e-05,
+      "loss": 0.8705,
+      "step": 9680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.576818936151629e-05,
+      "loss": 0.917,
+      "step": 9690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.576382216787493e-05,
+      "loss": 0.8363,
+      "step": 9700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.575945497423356e-05,
+      "loss": 1.0622,
+      "step": 9710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5755087780592194e-05,
+      "loss": 0.8973,
+      "step": 9720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.575072058695083e-05,
+      "loss": 0.9617,
+      "step": 9730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.574635339330946e-05,
+      "loss": 0.8569,
+      "step": 9740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5741986199668094e-05,
+      "loss": 0.8894,
+      "step": 9750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.573761900602673e-05,
+      "loss": 0.978,
+      "step": 9760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.573325181238537e-05,
+      "loss": 0.8624,
+      "step": 9770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.572888461874399e-05,
+      "loss": 0.979,
+      "step": 9780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.572451742510263e-05,
+      "loss": 0.9463,
+      "step": 9790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5720150231461266e-05,
+      "loss": 1.0555,
+      "step": 9800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.57157830378199e-05,
+      "loss": 0.8618,
+      "step": 9810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.571141584417853e-05,
+      "loss": 0.8996,
+      "step": 9820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5707048650537166e-05,
+      "loss": 0.8343,
+      "step": 9830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.57026814568958e-05,
+      "loss": 0.9402,
+      "step": 9840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.569831426325443e-05,
+      "loss": 0.9295,
+      "step": 9850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5693947069613066e-05,
+      "loss": 1.0123,
+      "step": 9860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5689579875971706e-05,
+      "loss": 0.9938,
+      "step": 9870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.568521268233034e-05,
+      "loss": 0.8626,
+      "step": 9880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.568084548868897e-05,
+      "loss": 0.9466,
+      "step": 9890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5676478295047605e-05,
+      "loss": 0.858,
+      "step": 9900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567211110140624e-05,
+      "loss": 0.765,
+      "step": 9910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.566774390776487e-05,
+      "loss": 0.9197,
+      "step": 9920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5663376714123505e-05,
+      "loss": 0.8931,
+      "step": 9930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.565900952048214e-05,
+      "loss": 1.001,
+      "step": 9940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.565464232684077e-05,
+      "loss": 0.8776,
+      "step": 9950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.565027513319941e-05,
+      "loss": 0.8704,
+      "step": 9960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5645907939558044e-05,
+      "loss": 0.8803,
+      "step": 9970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.564154074591668e-05,
+      "loss": 0.9628,
+      "step": 9980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.563717355227531e-05,
+      "loss": 0.8756,
+      "step": 9990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5632806358633944e-05,
+      "loss": 0.9446,
+      "step": 10000
+    },
+    {
+      "epoch": 0.87,
+      "eval_accuracy": 0.5734499361304903,
+      "eval_loss": 0.9110677242279053,
+      "eval_runtime": 84.0697,
+      "eval_samples_per_second": 121.054,
+      "eval_steps_per_second": 15.142,
+      "step": 10000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.562843916499258e-05,
+      "loss": 0.9811,
+      "step": 10010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562407197135121e-05,
+      "loss": 0.831,
+      "step": 10020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.561970477770985e-05,
+      "loss": 0.759,
+      "step": 10030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.561533758406848e-05,
+      "loss": 1.0103,
+      "step": 10040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.561097039042712e-05,
+      "loss": 0.8994,
+      "step": 10050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.560660319678574e-05,
+      "loss": 0.8891,
+      "step": 10060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.560223600314438e-05,
+      "loss": 0.9094,
+      "step": 10070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5597868809503016e-05,
+      "loss": 0.9042,
+      "step": 10080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.559350161586165e-05,
+      "loss": 0.9048,
+      "step": 10090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.558913442222028e-05,
+      "loss": 0.879,
+      "step": 10100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5584767228578916e-05,
+      "loss": 0.9188,
+      "step": 10110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5580400034937556e-05,
+      "loss": 0.8391,
+      "step": 10120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.557603284129618e-05,
+      "loss": 0.7726,
+      "step": 10130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.557166564765482e-05,
+      "loss": 0.9201,
+      "step": 10140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.556729845401345e-05,
+      "loss": 1.0287,
+      "step": 10150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.556293126037209e-05,
+      "loss": 0.9263,
+      "step": 10160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.555856406673072e-05,
+      "loss": 0.8906,
+      "step": 10170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5554196873089355e-05,
+      "loss": 0.9107,
+      "step": 10180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.554982967944799e-05,
+      "loss": 0.9614,
+      "step": 10190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.554546248580662e-05,
+      "loss": 0.9123,
+      "step": 10200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.554109529216526e-05,
+      "loss": 0.8693,
+      "step": 10210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.553672809852389e-05,
+      "loss": 0.8381,
+      "step": 10220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.553236090488253e-05,
+      "loss": 0.9337,
+      "step": 10230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5527993711241154e-05,
+      "loss": 0.9651,
+      "step": 10240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5523626517599794e-05,
+      "loss": 0.9808,
+      "step": 10250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.551925932395842e-05,
+      "loss": 1.0326,
+      "step": 10260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.551489213031706e-05,
+      "loss": 0.8245,
+      "step": 10270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5510524936675694e-05,
+      "loss": 0.814,
+      "step": 10280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.550615774303433e-05,
+      "loss": 0.9873,
+      "step": 10290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.550179054939297e-05,
+      "loss": 0.9243,
+      "step": 10300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5497423355751594e-05,
+      "loss": 0.9284,
+      "step": 10310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5493056162110234e-05,
+      "loss": 1.0906,
+      "step": 10320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.548868896846886e-05,
+      "loss": 0.8104,
+      "step": 10330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.54843217748275e-05,
+      "loss": 0.9079,
+      "step": 10340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5479954581186126e-05,
+      "loss": 1.0131,
+      "step": 10350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5475587387544766e-05,
+      "loss": 0.7866,
+      "step": 10360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.54712201939034e-05,
+      "loss": 1.0117,
+      "step": 10370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.546685300026203e-05,
+      "loss": 0.8584,
+      "step": 10380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5462485806620666e-05,
+      "loss": 1.0829,
+      "step": 10390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.54581186129793e-05,
+      "loss": 0.9519,
+      "step": 10400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.545375141933794e-05,
+      "loss": 0.9963,
+      "step": 10410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5449384225696566e-05,
+      "loss": 1.0018,
+      "step": 10420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5445017032055206e-05,
+      "loss": 0.9403,
+      "step": 10430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.544064983841384e-05,
+      "loss": 0.8824,
+      "step": 10440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.543628264477247e-05,
+      "loss": 0.9186,
+      "step": 10450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5431915451131105e-05,
+      "loss": 0.8281,
+      "step": 10460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.542754825748974e-05,
+      "loss": 0.9951,
+      "step": 10470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.542318106384837e-05,
+      "loss": 0.9504,
+      "step": 10480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5418813870207005e-05,
+      "loss": 0.8944,
+      "step": 10490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5414446676565645e-05,
+      "loss": 0.8705,
+      "step": 10500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.541007948292427e-05,
+      "loss": 0.8399,
+      "step": 10510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.540571228928291e-05,
+      "loss": 0.9143,
+      "step": 10520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5401345095641544e-05,
+      "loss": 0.9148,
+      "step": 10530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.539697790200018e-05,
+      "loss": 0.8801,
+      "step": 10540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.539261070835881e-05,
+      "loss": 0.9361,
+      "step": 10550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5388243514717444e-05,
+      "loss": 0.9047,
+      "step": 10560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.538387632107608e-05,
+      "loss": 0.8674,
+      "step": 10570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.537950912743471e-05,
+      "loss": 0.8867,
+      "step": 10580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5375141933793344e-05,
+      "loss": 0.9204,
+      "step": 10590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.537077474015198e-05,
+      "loss": 0.9896,
+      "step": 10600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.536640754651062e-05,
+      "loss": 0.9175,
+      "step": 10610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.536204035286925e-05,
+      "loss": 0.8352,
+      "step": 10620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.535767315922788e-05,
+      "loss": 0.9144,
+      "step": 10630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5353305965586516e-05,
+      "loss": 0.9773,
+      "step": 10640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.534893877194515e-05,
+      "loss": 0.9161,
+      "step": 10650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.534457157830378e-05,
+      "loss": 0.8555,
+      "step": 10660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5340204384662416e-05,
+      "loss": 0.8726,
+      "step": 10670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.533583719102105e-05,
+      "loss": 0.9121,
+      "step": 10680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.533146999737969e-05,
+      "loss": 0.8925,
+      "step": 10690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.532710280373832e-05,
+      "loss": 0.978,
+      "step": 10700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5322735610096956e-05,
+      "loss": 0.8737,
+      "step": 10710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.531836841645559e-05,
+      "loss": 0.8594,
+      "step": 10720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.531400122281422e-05,
+      "loss": 0.836,
+      "step": 10730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5309634029172855e-05,
+      "loss": 0.9575,
+      "step": 10740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.530526683553149e-05,
+      "loss": 0.8756,
+      "step": 10750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.530089964189012e-05,
+      "loss": 0.9887,
+      "step": 10760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5296532448248755e-05,
+      "loss": 0.8885,
+      "step": 10770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5292165254607395e-05,
+      "loss": 0.9223,
+      "step": 10780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.528779806096603e-05,
+      "loss": 0.9529,
+      "step": 10790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.528343086732466e-05,
+      "loss": 0.8108,
+      "step": 10800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5279063673683294e-05,
+      "loss": 0.9745,
+      "step": 10810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.527469648004193e-05,
+      "loss": 0.8772,
+      "step": 10820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.527032928640056e-05,
+      "loss": 0.875,
+      "step": 10830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5265962092759194e-05,
+      "loss": 0.9734,
+      "step": 10840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5261594899117834e-05,
+      "loss": 0.9213,
+      "step": 10850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.525722770547646e-05,
+      "loss": 0.9138,
+      "step": 10860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.52528605118351e-05,
+      "loss": 0.9242,
+      "step": 10870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.524849331819373e-05,
+      "loss": 0.9694,
+      "step": 10880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.524412612455237e-05,
+      "loss": 0.9103,
+      "step": 10890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5239758930911e-05,
+      "loss": 0.8588,
+      "step": 10900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.523539173726963e-05,
+      "loss": 0.9302,
+      "step": 10910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5231024543628266e-05,
+      "loss": 0.9684,
+      "step": 10920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.52266573499869e-05,
+      "loss": 0.8293,
+      "step": 10930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.522229015634554e-05,
+      "loss": 0.9172,
+      "step": 10940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5217922962704166e-05,
+      "loss": 0.9415,
+      "step": 10950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5213555769062806e-05,
+      "loss": 0.9137,
+      "step": 10960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.520918857542143e-05,
+      "loss": 0.9424,
+      "step": 10970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.520482138178007e-05,
+      "loss": 0.8267,
+      "step": 10980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5200454188138706e-05,
+      "loss": 0.8788,
+      "step": 10990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.519608699449734e-05,
+      "loss": 0.9256,
+      "step": 11000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.519171980085597e-05,
+      "loss": 1.0088,
+      "step": 11010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5187352607214605e-05,
+      "loss": 0.9595,
+      "step": 11020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5182985413573245e-05,
+      "loss": 0.9277,
+      "step": 11030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.517861821993187e-05,
+      "loss": 0.9578,
+      "step": 11040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517425102629051e-05,
+      "loss": 0.9886,
+      "step": 11050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.516988383264914e-05,
+      "loss": 1.0115,
+      "step": 11060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.516551663900778e-05,
+      "loss": 1.0093,
+      "step": 11070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5161149445366404e-05,
+      "loss": 0.889,
+      "step": 11080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5156782251725044e-05,
+      "loss": 0.8705,
+      "step": 11090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.515241505808368e-05,
+      "loss": 0.9113,
+      "step": 11100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.514804786444231e-05,
+      "loss": 0.8333,
+      "step": 11110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.514368067080095e-05,
+      "loss": 0.8226,
+      "step": 11120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.513931347715958e-05,
+      "loss": 0.9597,
+      "step": 11130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.513494628351822e-05,
+      "loss": 0.9097,
+      "step": 11140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5130579089876844e-05,
+      "loss": 0.8062,
+      "step": 11150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5126211896235484e-05,
+      "loss": 0.8212,
+      "step": 11160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512184470259411e-05,
+      "loss": 0.887,
+      "step": 11170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.511747750895275e-05,
+      "loss": 1.0557,
+      "step": 11180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.511311031531138e-05,
+      "loss": 0.9703,
+      "step": 11190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5108743121670016e-05,
+      "loss": 0.9554,
+      "step": 11200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.510437592802865e-05,
+      "loss": 1.0286,
+      "step": 11210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.510000873438728e-05,
+      "loss": 1.0108,
+      "step": 11220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.509564154074592e-05,
+      "loss": 0.8789,
+      "step": 11230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.509127434710455e-05,
+      "loss": 0.9179,
+      "step": 11240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.508690715346319e-05,
+      "loss": 0.9582,
+      "step": 11250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.508253995982182e-05,
+      "loss": 0.9739,
+      "step": 11260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5078172766180456e-05,
+      "loss": 0.8952,
+      "step": 11270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.507380557253909e-05,
+      "loss": 0.8655,
+      "step": 11280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.506943837889772e-05,
+      "loss": 0.9757,
+      "step": 11290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5065071185256355e-05,
+      "loss": 0.9027,
+      "step": 11300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.506070399161499e-05,
+      "loss": 0.931,
+      "step": 11310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.505633679797363e-05,
+      "loss": 0.9226,
+      "step": 11320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5051969604332255e-05,
+      "loss": 0.9397,
+      "step": 11330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5047602410690895e-05,
+      "loss": 0.9259,
+      "step": 11340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.504323521704953e-05,
+      "loss": 1.0322,
+      "step": 11350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.503886802340816e-05,
+      "loss": 0.8922,
+      "step": 11360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5034500829766794e-05,
+      "loss": 0.8602,
+      "step": 11370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.503013363612543e-05,
+      "loss": 0.9701,
+      "step": 11380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.502576644248406e-05,
+      "loss": 0.96,
+      "step": 11390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5021399248842694e-05,
+      "loss": 0.8228,
+      "step": 11400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.501703205520133e-05,
+      "loss": 1.0048,
+      "step": 11410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.501266486155996e-05,
+      "loss": 0.9371,
+      "step": 11420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.50082976679186e-05,
+      "loss": 0.9576,
+      "step": 11430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5003930474277234e-05,
+      "loss": 0.9371,
+      "step": 11440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.499956328063587e-05,
+      "loss": 0.9208,
+      "step": 11450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.49951960869945e-05,
+      "loss": 0.8842,
+      "step": 11460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.499082889335313e-05,
+      "loss": 0.9082,
+      "step": 11470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4986461699711766e-05,
+      "loss": 0.9201,
+      "step": 11480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.49820945060704e-05,
+      "loss": 0.9304,
+      "step": 11490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.497772731242903e-05,
+      "loss": 0.826,
+      "step": 11500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497336011878767e-05,
+      "loss": 0.958,
+      "step": 11510
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4968992925146306e-05,
+      "loss": 0.9367,
+      "step": 11520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.496462573150494e-05,
+      "loss": 0.8309,
+      "step": 11530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.496025853786357e-05,
+      "loss": 0.8894,
+      "step": 11540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4955891344222206e-05,
+      "loss": 0.882,
+      "step": 11550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.495152415058084e-05,
+      "loss": 1.0092,
+      "step": 11560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.494715695693947e-05,
+      "loss": 0.8797,
+      "step": 11570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4942789763298105e-05,
+      "loss": 0.9061,
+      "step": 11580
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.493842256965674e-05,
+      "loss": 0.9299,
+      "step": 11590
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.493405537601538e-05,
+      "loss": 0.8721,
+      "step": 11600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4929688182374005e-05,
+      "loss": 0.8272,
+      "step": 11610
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4925320988732645e-05,
+      "loss": 0.9526,
+      "step": 11620
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.492095379509128e-05,
+      "loss": 0.9967,
+      "step": 11630
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.491658660144991e-05,
+      "loss": 0.9281,
+      "step": 11640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4912219407808544e-05,
+      "loss": 0.8985,
+      "step": 11650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.490785221416718e-05,
+      "loss": 1.0259,
+      "step": 11660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.490348502052581e-05,
+      "loss": 0.9732,
+      "step": 11670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4899117826884444e-05,
+      "loss": 0.9151,
+      "step": 11680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4894750633243084e-05,
+      "loss": 0.9595,
+      "step": 11690
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.489038343960171e-05,
+      "loss": 0.9329,
+      "step": 11700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.488601624596035e-05,
+      "loss": 0.9071,
+      "step": 11710
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4881649052318984e-05,
+      "loss": 0.9468,
+      "step": 11720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.487728185867762e-05,
+      "loss": 0.9363,
+      "step": 11730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.487291466503625e-05,
+      "loss": 0.8816,
+      "step": 11740
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.486854747139488e-05,
+      "loss": 0.7856,
+      "step": 11750
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.486418027775352e-05,
+      "loss": 0.8097,
+      "step": 11760
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.485981308411215e-05,
+      "loss": 0.8906,
+      "step": 11770
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.485544589047079e-05,
+      "loss": 0.9801,
+      "step": 11780
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4851078696829416e-05,
+      "loss": 0.9034,
+      "step": 11790
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4846711503188056e-05,
+      "loss": 0.8384,
+      "step": 11800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.484234430954668e-05,
+      "loss": 0.8314,
+      "step": 11810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.483797711590532e-05,
+      "loss": 0.9133,
+      "step": 11820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4833609922263956e-05,
+      "loss": 1.032,
+      "step": 11830
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.482924272862259e-05,
+      "loss": 0.8315,
+      "step": 11840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.482487553498123e-05,
+      "loss": 0.9895,
+      "step": 11850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4820508341339855e-05,
+      "loss": 0.9437,
+      "step": 11860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4816141147698495e-05,
+      "loss": 0.8944,
+      "step": 11870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.481177395405712e-05,
+      "loss": 0.9036,
+      "step": 11880
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.480740676041576e-05,
+      "loss": 0.8344,
+      "step": 11890
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.480303956677439e-05,
+      "loss": 1.0339,
+      "step": 11900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.479867237313303e-05,
+      "loss": 0.8336,
+      "step": 11910
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.479430517949166e-05,
+      "loss": 0.9073,
+      "step": 11920
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4789937985850294e-05,
+      "loss": 0.9492,
+      "step": 11930
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.478557079220893e-05,
+      "loss": 0.956,
+      "step": 11940
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.478120359856756e-05,
+      "loss": 0.9705,
+      "step": 11950
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.47768364049262e-05,
+      "loss": 0.9705,
+      "step": 11960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.477246921128483e-05,
+      "loss": 0.9774,
+      "step": 11970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.476810201764347e-05,
+      "loss": 0.9664,
+      "step": 11980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4763734824002094e-05,
+      "loss": 0.8447,
+      "step": 11990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4759367630360734e-05,
+      "loss": 0.9623,
+      "step": 12000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.475500043671937e-05,
+      "loss": 0.8843,
+      "step": 12010
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4750633243078e-05,
+      "loss": 0.9471,
+      "step": 12020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.474626604943663e-05,
+      "loss": 0.8741,
+      "step": 12030
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4741898855795266e-05,
+      "loss": 0.8642,
+      "step": 12040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4737531662153906e-05,
+      "loss": 0.8563,
+      "step": 12050
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.473316446851253e-05,
+      "loss": 0.9213,
+      "step": 12060
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.472879727487117e-05,
+      "loss": 0.9986,
+      "step": 12070
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.47244300812298e-05,
+      "loss": 0.91,
+      "step": 12080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.472006288758844e-05,
+      "loss": 0.9023,
+      "step": 12090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.471569569394707e-05,
+      "loss": 0.976,
+      "step": 12100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4711328500305706e-05,
+      "loss": 0.8676,
+      "step": 12110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.470696130666434e-05,
+      "loss": 0.9281,
+      "step": 12120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.470259411302297e-05,
+      "loss": 0.9033,
+      "step": 12130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4698226919381605e-05,
+      "loss": 0.8466,
+      "step": 12140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.469385972574024e-05,
+      "loss": 0.8466,
+      "step": 12150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.468949253209888e-05,
+      "loss": 0.9056,
+      "step": 12160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.468512533845751e-05,
+      "loss": 0.9986,
+      "step": 12170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4680758144816145e-05,
+      "loss": 0.9672,
+      "step": 12180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.467639095117478e-05,
+      "loss": 1.0553,
+      "step": 12190
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467202375753341e-05,
+      "loss": 0.9605,
+      "step": 12200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4667656563892044e-05,
+      "loss": 0.8074,
+      "step": 12210
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.466328937025068e-05,
+      "loss": 0.9541,
+      "step": 12220
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.465892217660931e-05,
+      "loss": 0.9729,
+      "step": 12230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4654554982967944e-05,
+      "loss": 0.9304,
+      "step": 12240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4650187789326584e-05,
+      "loss": 0.9546,
+      "step": 12250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.464582059568522e-05,
+      "loss": 0.9757,
+      "step": 12260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.464145340204385e-05,
+      "loss": 1.065,
+      "step": 12270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4637086208402484e-05,
+      "loss": 1.0735,
+      "step": 12280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.463271901476112e-05,
+      "loss": 1.0065,
+      "step": 12290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.462835182111975e-05,
+      "loss": 0.8726,
+      "step": 12300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462398462747838e-05,
+      "loss": 0.9266,
+      "step": 12310
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4619617433837016e-05,
+      "loss": 0.8507,
+      "step": 12320
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4615250240195656e-05,
+      "loss": 0.8866,
+      "step": 12330
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.461088304655428e-05,
+      "loss": 0.9401,
+      "step": 12340
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.460651585291292e-05,
+      "loss": 0.8221,
+      "step": 12350
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4602148659271556e-05,
+      "loss": 0.9033,
+      "step": 12360
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.459778146563019e-05,
+      "loss": 0.9285,
+      "step": 12370
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.459341427198882e-05,
+      "loss": 0.9238,
+      "step": 12380
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4589047078347456e-05,
+      "loss": 0.9655,
+      "step": 12390
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.458467988470609e-05,
+      "loss": 0.9495,
+      "step": 12400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.458031269106472e-05,
+      "loss": 0.9983,
+      "step": 12410
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.457594549742336e-05,
+      "loss": 0.9928,
+      "step": 12420
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.457157830378199e-05,
+      "loss": 0.9189,
+      "step": 12430
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.456721111014063e-05,
+      "loss": 0.9271,
+      "step": 12440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.456284391649926e-05,
+      "loss": 0.8451,
+      "step": 12450
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4558476722857895e-05,
+      "loss": 0.893,
+      "step": 12460
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.455410952921653e-05,
+      "loss": 0.8496,
+      "step": 12470
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.454974233557516e-05,
+      "loss": 0.8163,
+      "step": 12480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4545375141933794e-05,
+      "loss": 0.9072,
+      "step": 12490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.454100794829243e-05,
+      "loss": 0.8713,
+      "step": 12500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.453664075465107e-05,
+      "loss": 0.982,
+      "step": 12510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4532273561009694e-05,
+      "loss": 0.9044,
+      "step": 12520
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4527906367368334e-05,
+      "loss": 0.9728,
+      "step": 12530
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.452353917372696e-05,
+      "loss": 0.936,
+      "step": 12540
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.45191719800856e-05,
+      "loss": 0.851,
+      "step": 12550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4514804786444234e-05,
+      "loss": 0.8592,
+      "step": 12560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.451043759280287e-05,
+      "loss": 0.9521,
+      "step": 12570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.450607039916151e-05,
+      "loss": 0.8226,
+      "step": 12580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.450170320552013e-05,
+      "loss": 0.8988,
+      "step": 12590
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.449733601187877e-05,
+      "loss": 1.1016,
+      "step": 12600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.44929688182374e-05,
+      "loss": 0.8615,
+      "step": 12610
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.448860162459604e-05,
+      "loss": 0.8843,
+      "step": 12620
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4484234430954666e-05,
+      "loss": 0.9107,
+      "step": 12630
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4479867237313306e-05,
+      "loss": 0.8771,
+      "step": 12640
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.447550004367194e-05,
+      "loss": 0.8495,
+      "step": 12650
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.447113285003057e-05,
+      "loss": 0.8912,
+      "step": 12660
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4466765656389206e-05,
+      "loss": 0.9469,
+      "step": 12670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.446239846274784e-05,
+      "loss": 0.9458,
+      "step": 12680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.445803126910648e-05,
+      "loss": 1.0344,
+      "step": 12690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4453664075465105e-05,
+      "loss": 0.9375,
+      "step": 12700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4449296881823745e-05,
+      "loss": 0.9414,
+      "step": 12710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.444492968818237e-05,
+      "loss": 0.9428,
+      "step": 12720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.444056249454101e-05,
+      "loss": 0.9601,
+      "step": 12730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4436195300899645e-05,
+      "loss": 0.9152,
+      "step": 12740
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.443182810725828e-05,
+      "loss": 0.8413,
+      "step": 12750
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.442746091361691e-05,
+      "loss": 0.9359,
+      "step": 12760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4423093719975544e-05,
+      "loss": 0.9442,
+      "step": 12770
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4418726526334184e-05,
+      "loss": 0.8774,
+      "step": 12780
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.441435933269281e-05,
+      "loss": 0.9719,
+      "step": 12790
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.440999213905145e-05,
+      "loss": 0.855,
+      "step": 12800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.440562494541008e-05,
+      "loss": 1.0001,
+      "step": 12810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.440125775176872e-05,
+      "loss": 0.9675,
+      "step": 12820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.439689055812735e-05,
+      "loss": 0.8379,
+      "step": 12830
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4392523364485984e-05,
+      "loss": 0.8493,
+      "step": 12840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.438815617084462e-05,
+      "loss": 0.9467,
+      "step": 12850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.438378897720325e-05,
+      "loss": 0.8218,
+      "step": 12860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.437942178356188e-05,
+      "loss": 1.018,
+      "step": 12870
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4375054589920517e-05,
+      "loss": 0.8891,
+      "step": 12880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4370687396279156e-05,
+      "loss": 0.859,
+      "step": 12890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.436632020263778e-05,
+      "loss": 0.9007,
+      "step": 12900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.436195300899642e-05,
+      "loss": 0.9974,
+      "step": 12910
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4357585815355056e-05,
+      "loss": 0.7639,
+      "step": 12920
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.435321862171369e-05,
+      "loss": 0.7397,
+      "step": 12930
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.434885142807232e-05,
+      "loss": 0.9167,
+      "step": 12940
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4344484234430956e-05,
+      "loss": 0.9047,
+      "step": 12950
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.434011704078959e-05,
+      "loss": 0.8814,
+      "step": 12960
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.433574984714822e-05,
+      "loss": 0.8639,
+      "step": 12970
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.433138265350686e-05,
+      "loss": 0.7749,
+      "step": 12980
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4327015459865495e-05,
+      "loss": 0.9552,
+      "step": 12990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.432264826622413e-05,
+      "loss": 0.985,
+      "step": 13000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.431828107258276e-05,
+      "loss": 1.0176,
+      "step": 13010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4313913878941395e-05,
+      "loss": 0.9972,
+      "step": 13020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.430954668530003e-05,
+      "loss": 0.9697,
+      "step": 13030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.430517949165866e-05,
+      "loss": 0.9851,
+      "step": 13040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4300812298017295e-05,
+      "loss": 0.8945,
+      "step": 13050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.429644510437593e-05,
+      "loss": 0.8953,
+      "step": 13060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.429207791073456e-05,
+      "loss": 0.8801,
+      "step": 13070
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.42877107170932e-05,
+      "loss": 0.9492,
+      "step": 13080
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4283343523451834e-05,
+      "loss": 0.9237,
+      "step": 13090
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.427897632981047e-05,
+      "loss": 0.8587,
+      "step": 13100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.42746091361691e-05,
+      "loss": 1.0227,
+      "step": 13110
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4270241942527734e-05,
+      "loss": 0.9383,
+      "step": 13120
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.426587474888637e-05,
+      "loss": 0.902,
+      "step": 13130
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4261507555245e-05,
+      "loss": 0.8372,
+      "step": 13140
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.425714036160363e-05,
+      "loss": 0.9497,
+      "step": 13150
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4252773167962267e-05,
+      "loss": 0.9441,
+      "step": 13160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4248405974320906e-05,
+      "loss": 0.9118,
+      "step": 13170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.424403878067954e-05,
+      "loss": 0.8783,
+      "step": 13180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.423967158703817e-05,
+      "loss": 0.9718,
+      "step": 13190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4235304393396806e-05,
+      "loss": 0.9306,
+      "step": 13200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.423093719975544e-05,
+      "loss": 0.9114,
+      "step": 13210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.422657000611407e-05,
+      "loss": 0.8427,
+      "step": 13220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4222202812472706e-05,
+      "loss": 1.0131,
+      "step": 13230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4217835618831346e-05,
+      "loss": 0.8891,
+      "step": 13240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.421346842518997e-05,
+      "loss": 0.8747,
+      "step": 13250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.420910123154861e-05,
+      "loss": 0.8912,
+      "step": 13260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.420473403790724e-05,
+      "loss": 0.9641,
+      "step": 13270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.420036684426588e-05,
+      "loss": 0.9093,
+      "step": 13280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.419599965062451e-05,
+      "loss": 0.9304,
+      "step": 13290
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4191632456983145e-05,
+      "loss": 0.8958,
+      "step": 13300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.418726526334178e-05,
+      "loss": 0.7827,
+      "step": 13310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.418289806970041e-05,
+      "loss": 0.9512,
+      "step": 13320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.417853087605905e-05,
+      "loss": 0.8458,
+      "step": 13330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.417416368241768e-05,
+      "loss": 0.9175,
+      "step": 13340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.416979648877632e-05,
+      "loss": 0.8291,
+      "step": 13350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4165429295134944e-05,
+      "loss": 0.831,
+      "step": 13360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4161062101493584e-05,
+      "loss": 0.8486,
+      "step": 13370
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.415669490785222e-05,
+      "loss": 0.9529,
+      "step": 13380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.415232771421085e-05,
+      "loss": 0.8852,
+      "step": 13390
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4147960520569484e-05,
+      "loss": 0.9277,
+      "step": 13400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.414359332692812e-05,
+      "loss": 0.8877,
+      "step": 13410
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.413922613328676e-05,
+      "loss": 0.9588,
+      "step": 13420
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.413485893964538e-05,
+      "loss": 0.8317,
+      "step": 13430
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.413049174600402e-05,
+      "loss": 0.8091,
+      "step": 13440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.412612455236265e-05,
+      "loss": 0.9759,
+      "step": 13450
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412175735872129e-05,
+      "loss": 0.8775,
+      "step": 13460
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4117390165079916e-05,
+      "loss": 1.0209,
+      "step": 13470
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4113022971438556e-05,
+      "loss": 0.8956,
+      "step": 13480
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.410865577779719e-05,
+      "loss": 0.8969,
+      "step": 13490
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.410428858415582e-05,
+      "loss": 0.9865,
+      "step": 13500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.409992139051446e-05,
+      "loss": 0.7589,
+      "step": 13510
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.409555419687309e-05,
+      "loss": 0.9255,
+      "step": 13520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.409118700323173e-05,
+      "loss": 1.09,
+      "step": 13530
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4086819809590355e-05,
+      "loss": 1.0017,
+      "step": 13540
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4082452615948995e-05,
+      "loss": 1.0634,
+      "step": 13550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.407808542230762e-05,
+      "loss": 0.8959,
+      "step": 13560
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.407371822866626e-05,
+      "loss": 0.8688,
+      "step": 13570
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4069351035024895e-05,
+      "loss": 0.9134,
+      "step": 13580
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.406498384138353e-05,
+      "loss": 0.9179,
+      "step": 13590
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.406061664774216e-05,
+      "loss": 0.9349,
+      "step": 13600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4056249454100795e-05,
+      "loss": 0.962,
+      "step": 13610
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4051882260459435e-05,
+      "loss": 0.8952,
+      "step": 13620
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.404751506681806e-05,
+      "loss": 0.9066,
+      "step": 13630
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.40431478731767e-05,
+      "loss": 0.9379,
+      "step": 13640
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4038780679535334e-05,
+      "loss": 0.9543,
+      "step": 13650
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.403441348589397e-05,
+      "loss": 0.8335,
+      "step": 13660
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.40300462922526e-05,
+      "loss": 0.8901,
+      "step": 13670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4025679098611234e-05,
+      "loss": 1.0229,
+      "step": 13680
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.402131190496987e-05,
+      "loss": 0.9643,
+      "step": 13690
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.40169447113285e-05,
+      "loss": 0.8058,
+      "step": 13700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.401257751768714e-05,
+      "loss": 1.1178,
+      "step": 13710
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4008210324045767e-05,
+      "loss": 0.934,
+      "step": 13720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4003843130404407e-05,
+      "loss": 0.8609,
+      "step": 13730
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.399947593676304e-05,
+      "loss": 0.9171,
+      "step": 13740
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.399510874312167e-05,
+      "loss": 0.9381,
+      "step": 13750
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.3990741549480306e-05,
+      "loss": 0.9924,
+      "step": 13760
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.398637435583894e-05,
+      "loss": 0.9531,
+      "step": 13770
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.398200716219757e-05,
+      "loss": 0.8536,
+      "step": 13780
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.3977639968556206e-05,
+      "loss": 0.7362,
+      "step": 13790
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.397327277491484e-05,
+      "loss": 0.8746,
+      "step": 13800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.396890558127348e-05,
+      "loss": 0.8384,
+      "step": 13810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.396453838763211e-05,
+      "loss": 1.0544,
+      "step": 13820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3960171193990745e-05,
+      "loss": 0.7399,
+      "step": 13830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.395580400034938e-05,
+      "loss": 0.9517,
+      "step": 13840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.395143680670801e-05,
+      "loss": 0.7535,
+      "step": 13850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3947069613066645e-05,
+      "loss": 0.9315,
+      "step": 13860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.394270241942528e-05,
+      "loss": 0.8628,
+      "step": 13870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.393833522578391e-05,
+      "loss": 0.9181,
+      "step": 13880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3933968032142545e-05,
+      "loss": 0.9031,
+      "step": 13890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3929600838501185e-05,
+      "loss": 0.9423,
+      "step": 13900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.392523364485982e-05,
+      "loss": 0.8268,
+      "step": 13910
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.392086645121845e-05,
+      "loss": 0.9842,
+      "step": 13920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3916499257577084e-05,
+      "loss": 0.8407,
+      "step": 13930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.391213206393572e-05,
+      "loss": 0.9953,
+      "step": 13940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.390776487029435e-05,
+      "loss": 0.9934,
+      "step": 13950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3903397676652984e-05,
+      "loss": 0.9091,
+      "step": 13960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.389903048301162e-05,
+      "loss": 0.9382,
+      "step": 13970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.389466328937025e-05,
+      "loss": 0.9491,
+      "step": 13980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.389029609572889e-05,
+      "loss": 0.9556,
+      "step": 13990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3885928902087517e-05,
+      "loss": 0.9202,
+      "step": 14000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3881561708446157e-05,
+      "loss": 0.8045,
+      "step": 14010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.387719451480479e-05,
+      "loss": 0.9044,
+      "step": 14020
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.387282732116342e-05,
+      "loss": 0.9394,
+      "step": 14030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3868460127522056e-05,
+      "loss": 0.885,
+      "step": 14040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.386409293388069e-05,
+      "loss": 0.9719,
+      "step": 14050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.385972574023933e-05,
+      "loss": 0.9355,
+      "step": 14060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3855358546597956e-05,
+      "loss": 0.9881,
+      "step": 14070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3850991352956596e-05,
+      "loss": 0.9022,
+      "step": 14080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.384662415931522e-05,
+      "loss": 0.9913,
+      "step": 14090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.384225696567386e-05,
+      "loss": 0.7881,
+      "step": 14100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3837889772032495e-05,
+      "loss": 0.9161,
+      "step": 14110
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.383352257839113e-05,
+      "loss": 0.8111,
+      "step": 14120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.382915538474976e-05,
+      "loss": 0.8397,
+      "step": 14130
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3824788191108395e-05,
+      "loss": 0.9204,
+      "step": 14140
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3820420997467035e-05,
+      "loss": 0.8167,
+      "step": 14150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.381605380382566e-05,
+      "loss": 0.982,
+      "step": 14160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.38116866101843e-05,
+      "loss": 0.9961,
+      "step": 14170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.380731941654293e-05,
+      "loss": 0.9445,
+      "step": 14180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.380295222290157e-05,
+      "loss": 0.936,
+      "step": 14190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3798585029260194e-05,
+      "loss": 0.89,
+      "step": 14200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3794217835618834e-05,
+      "loss": 0.9726,
+      "step": 14210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.378985064197747e-05,
+      "loss": 1.0089,
+      "step": 14220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.37854834483361e-05,
+      "loss": 0.9562,
+      "step": 14230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.378111625469474e-05,
+      "loss": 0.9268,
+      "step": 14240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.377674906105337e-05,
+      "loss": 0.9123,
+      "step": 14250
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.377238186741201e-05,
+      "loss": 0.8134,
+      "step": 14260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.376801467377063e-05,
+      "loss": 0.8873,
+      "step": 14270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.376364748012927e-05,
+      "loss": 0.9168,
+      "step": 14280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.37592802864879e-05,
+      "loss": 0.8936,
+      "step": 14290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.375491309284654e-05,
+      "loss": 0.8572,
+      "step": 14300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.375054589920517e-05,
+      "loss": 0.7955,
+      "step": 14310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3746178705563806e-05,
+      "loss": 0.9254,
+      "step": 14320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.374181151192244e-05,
+      "loss": 0.9034,
+      "step": 14330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.373744431828107e-05,
+      "loss": 0.9295,
+      "step": 14340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.373307712463971e-05,
+      "loss": 0.8735,
+      "step": 14350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.372870993099834e-05,
+      "loss": 1.0061,
+      "step": 14360
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.372434273735698e-05,
+      "loss": 0.8904,
+      "step": 14370
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3719975543715605e-05,
+      "loss": 0.9275,
+      "step": 14380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3715608350074245e-05,
+      "loss": 0.8618,
+      "step": 14390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.371124115643288e-05,
+      "loss": 0.8542,
+      "step": 14400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.370687396279151e-05,
+      "loss": 0.8597,
+      "step": 14410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3702506769150145e-05,
+      "loss": 0.8854,
+      "step": 14420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.369813957550878e-05,
+      "loss": 0.8318,
+      "step": 14430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.369377238186742e-05,
+      "loss": 0.9554,
+      "step": 14440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3689405188226045e-05,
+      "loss": 0.8539,
+      "step": 14450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3685037994584685e-05,
+      "loss": 0.9091,
+      "step": 14460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.368067080094332e-05,
+      "loss": 0.8744,
+      "step": 14470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.367630360730195e-05,
+      "loss": 0.8784,
+      "step": 14480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3671936413660584e-05,
+      "loss": 0.9389,
+      "step": 14490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366756922001922e-05,
+      "loss": 0.9308,
+      "step": 14500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366320202637785e-05,
+      "loss": 0.7971,
+      "step": 14510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3658834832736484e-05,
+      "loss": 0.9154,
+      "step": 14520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.365446763909512e-05,
+      "loss": 0.9362,
+      "step": 14530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.365010044545375e-05,
+      "loss": 0.9265,
+      "step": 14540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.364573325181239e-05,
+      "loss": 0.9781,
+      "step": 14550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.364136605817102e-05,
+      "loss": 0.8626,
+      "step": 14560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3636998864529657e-05,
+      "loss": 0.9103,
+      "step": 14570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.363263167088829e-05,
+      "loss": 0.9261,
+      "step": 14580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.362826447724692e-05,
+      "loss": 0.9449,
+      "step": 14590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3623897283605556e-05,
+      "loss": 0.9019,
+      "step": 14600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361953008996419e-05,
+      "loss": 0.9538,
+      "step": 14610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361516289632282e-05,
+      "loss": 0.9009,
+      "step": 14620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361079570268146e-05,
+      "loss": 0.8667,
+      "step": 14630
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3606428509040096e-05,
+      "loss": 0.86,
+      "step": 14640
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.360206131539873e-05,
+      "loss": 0.8449,
+      "step": 14650
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.359769412175736e-05,
+      "loss": 0.9632,
+      "step": 14660
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3593326928115995e-05,
+      "loss": 0.8477,
+      "step": 14670
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358895973447463e-05,
+      "loss": 0.8059,
+      "step": 14680
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358459254083326e-05,
+      "loss": 0.9049,
+      "step": 14690
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3580225347191895e-05,
+      "loss": 0.9158,
+      "step": 14700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.357585815355053e-05,
+      "loss": 0.9335,
+      "step": 14710
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357149095990917e-05,
+      "loss": 0.8919,
+      "step": 14720
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3567123766267795e-05,
+      "loss": 0.9624,
+      "step": 14730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3562756572626435e-05,
+      "loss": 0.9487,
+      "step": 14740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.355838937898507e-05,
+      "loss": 0.7973,
+      "step": 14750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.35540221853437e-05,
+      "loss": 0.8796,
+      "step": 14760
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3549654991702334e-05,
+      "loss": 0.9861,
+      "step": 14770
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.354528779806097e-05,
+      "loss": 0.8296,
+      "step": 14780
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.35409206044196e-05,
+      "loss": 0.8536,
+      "step": 14790
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3536553410778234e-05,
+      "loss": 0.935,
+      "step": 14800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3532186217136874e-05,
+      "loss": 0.8927,
+      "step": 14810
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.35278190234955e-05,
+      "loss": 1.0031,
+      "step": 14820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.352345182985414e-05,
+      "loss": 0.9235,
+      "step": 14830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.351908463621277e-05,
+      "loss": 0.9993,
+      "step": 14840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3514717442571407e-05,
+      "loss": 0.817,
+      "step": 14850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.351035024893004e-05,
+      "loss": 0.9763,
+      "step": 14860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.350598305528867e-05,
+      "loss": 0.8549,
+      "step": 14870
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.350161586164731e-05,
+      "loss": 0.6822,
+      "step": 14880
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.349724866800594e-05,
+      "loss": 0.9731,
+      "step": 14890
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.349288147436458e-05,
+      "loss": 0.8859,
+      "step": 14900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3488514280723206e-05,
+      "loss": 0.999,
+      "step": 14910
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3484147087081846e-05,
+      "loss": 0.8618,
+      "step": 14920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.347977989344047e-05,
+      "loss": 0.8044,
+      "step": 14930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.347541269979911e-05,
+      "loss": 0.9209,
+      "step": 14940
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3471045506157745e-05,
+      "loss": 0.8291,
+      "step": 14950
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.346667831251638e-05,
+      "loss": 0.8276,
+      "step": 14960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.346231111887502e-05,
+      "loss": 1.0486,
+      "step": 14970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3457943925233645e-05,
+      "loss": 1.0739,
+      "step": 14980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3453576731592285e-05,
+      "loss": 0.9206,
+      "step": 14990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.344920953795091e-05,
+      "loss": 0.9701,
+      "step": 15000
+    },
+    {
+      "epoch": 1.31,
+      "eval_accuracy": 0.572762110641643,
+      "eval_loss": 0.9019526243209839,
+      "eval_runtime": 84.1256,
+      "eval_samples_per_second": 120.974,
+      "eval_steps_per_second": 15.132,
+      "step": 15000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.344484234430955e-05,
+      "loss": 0.919,
+      "step": 15010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.344047515066818e-05,
+      "loss": 0.9125,
+      "step": 15020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.343610795702682e-05,
+      "loss": 0.8528,
+      "step": 15030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.343174076338545e-05,
+      "loss": 0.8604,
+      "step": 15040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3427373569744084e-05,
+      "loss": 0.7717,
+      "step": 15050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.342300637610272e-05,
+      "loss": 0.9932,
+      "step": 15060
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.341863918246135e-05,
+      "loss": 0.936,
+      "step": 15070
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.341427198881999e-05,
+      "loss": 0.8852,
+      "step": 15080
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.340990479517862e-05,
+      "loss": 0.8235,
+      "step": 15090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.340553760153726e-05,
+      "loss": 0.941,
+      "step": 15100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3401170407895883e-05,
+      "loss": 0.9002,
+      "step": 15110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.339680321425452e-05,
+      "loss": 0.8642,
+      "step": 15120
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3392436020613157e-05,
+      "loss": 0.9763,
+      "step": 15130
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.338806882697179e-05,
+      "loss": 0.9184,
+      "step": 15140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.338370163333042e-05,
+      "loss": 0.9508,
+      "step": 15150
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3379334439689056e-05,
+      "loss": 0.8666,
+      "step": 15160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3374967246047696e-05,
+      "loss": 0.7512,
+      "step": 15170
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337060005240632e-05,
+      "loss": 0.8854,
+      "step": 15180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.336623285876496e-05,
+      "loss": 0.8542,
+      "step": 15190
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.336186566512359e-05,
+      "loss": 0.9862,
+      "step": 15200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.335749847148223e-05,
+      "loss": 0.9491,
+      "step": 15210
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.335313127784086e-05,
+      "loss": 0.8883,
+      "step": 15220
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3348764084199495e-05,
+      "loss": 0.9286,
+      "step": 15230
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.334439689055813e-05,
+      "loss": 0.8865,
+      "step": 15240
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.334002969691676e-05,
+      "loss": 0.9284,
+      "step": 15250
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3335662503275395e-05,
+      "loss": 0.9085,
+      "step": 15260
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.333129530963403e-05,
+      "loss": 0.8477,
+      "step": 15270
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.332692811599267e-05,
+      "loss": 0.9808,
+      "step": 15280
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.33225609223513e-05,
+      "loss": 1.0297,
+      "step": 15290
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3318193728709935e-05,
+      "loss": 0.8064,
+      "step": 15300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.331382653506857e-05,
+      "loss": 0.7985,
+      "step": 15310
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.33094593414272e-05,
+      "loss": 0.9677,
+      "step": 15320
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3305092147785834e-05,
+      "loss": 0.7872,
+      "step": 15330
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.330072495414447e-05,
+      "loss": 0.7886,
+      "step": 15340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.32963577605031e-05,
+      "loss": 0.8831,
+      "step": 15350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3291990566861734e-05,
+      "loss": 0.9471,
+      "step": 15360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3287623373220374e-05,
+      "loss": 0.8748,
+      "step": 15370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.328325617957901e-05,
+      "loss": 0.8131,
+      "step": 15380
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.327888898593764e-05,
+      "loss": 0.8532,
+      "step": 15390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.327452179229627e-05,
+      "loss": 0.9196,
+      "step": 15400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3270154598654907e-05,
+      "loss": 1.0177,
+      "step": 15410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.326578740501354e-05,
+      "loss": 0.9363,
+      "step": 15420
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.326142021137217e-05,
+      "loss": 0.8901,
+      "step": 15430
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3257053017730806e-05,
+      "loss": 0.9239,
+      "step": 15440
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.325268582408944e-05,
+      "loss": 0.8278,
+      "step": 15450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.324831863044807e-05,
+      "loss": 0.9067,
+      "step": 15460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.324395143680671e-05,
+      "loss": 0.934,
+      "step": 15470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3239584243165346e-05,
+      "loss": 0.8488,
+      "step": 15480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.323521704952398e-05,
+      "loss": 0.9234,
+      "step": 15490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.323084985588261e-05,
+      "loss": 1.0279,
+      "step": 15500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3226482662241245e-05,
+      "loss": 0.8982,
+      "step": 15510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.322211546859988e-05,
+      "loss": 0.9552,
+      "step": 15520
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.321774827495851e-05,
+      "loss": 0.9862,
+      "step": 15530
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.321338108131715e-05,
+      "loss": 0.9822,
+      "step": 15540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.320901388767578e-05,
+      "loss": 0.8063,
+      "step": 15550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.320464669403442e-05,
+      "loss": 1.012,
+      "step": 15560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.320027950039305e-05,
+      "loss": 0.8776,
+      "step": 15570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3195912306751685e-05,
+      "loss": 0.9512,
+      "step": 15580
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.319154511311032e-05,
+      "loss": 0.9187,
+      "step": 15590
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.318717791946895e-05,
+      "loss": 0.8636,
+      "step": 15600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3182810725827584e-05,
+      "loss": 0.8431,
+      "step": 15610
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.317844353218622e-05,
+      "loss": 0.8475,
+      "step": 15620
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.317407633854486e-05,
+      "loss": 1.0569,
+      "step": 15630
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3169709144903484e-05,
+      "loss": 0.9097,
+      "step": 15640
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3165341951262124e-05,
+      "loss": 0.9391,
+      "step": 15650
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.316097475762075e-05,
+      "loss": 1.0363,
+      "step": 15660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.315660756397939e-05,
+      "loss": 0.8641,
+      "step": 15670
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3152240370338023e-05,
+      "loss": 0.8073,
+      "step": 15680
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3147873176696657e-05,
+      "loss": 0.8456,
+      "step": 15690
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3143505983055297e-05,
+      "loss": 0.9166,
+      "step": 15700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.313913878941392e-05,
+      "loss": 0.9334,
+      "step": 15710
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.313477159577256e-05,
+      "loss": 0.974,
+      "step": 15720
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.313040440213119e-05,
+      "loss": 0.887,
+      "step": 15730
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.312603720848983e-05,
+      "loss": 0.8785,
+      "step": 15740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3121670014848456e-05,
+      "loss": 0.959,
+      "step": 15750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3117302821207096e-05,
+      "loss": 0.9278,
+      "step": 15760
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.311293562756573e-05,
+      "loss": 0.9141,
+      "step": 15770
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.310856843392436e-05,
+      "loss": 0.9412,
+      "step": 15780
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3104201240282995e-05,
+      "loss": 0.9377,
+      "step": 15790
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.309983404664163e-05,
+      "loss": 0.9439,
+      "step": 15800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.309546685300027e-05,
+      "loss": 0.9453,
+      "step": 15810
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3091099659358895e-05,
+      "loss": 0.9007,
+      "step": 15820
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3086732465717535e-05,
+      "loss": 0.8418,
+      "step": 15830
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.308236527207616e-05,
+      "loss": 0.8998,
+      "step": 15840
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.30779980784348e-05,
+      "loss": 0.8341,
+      "step": 15850
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.307363088479343e-05,
+      "loss": 0.8769,
+      "step": 15860
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.306926369115207e-05,
+      "loss": 0.8236,
+      "step": 15870
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.30648964975107e-05,
+      "loss": 0.8204,
+      "step": 15880
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3060529303869334e-05,
+      "loss": 0.8364,
+      "step": 15890
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3056162110227974e-05,
+      "loss": 0.9732,
+      "step": 15900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.30517949165866e-05,
+      "loss": 0.8895,
+      "step": 15910
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.304742772294524e-05,
+      "loss": 0.9245,
+      "step": 15920
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.304306052930387e-05,
+      "loss": 0.9548,
+      "step": 15930
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.303869333566251e-05,
+      "loss": 0.9749,
+      "step": 15940
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.303432614202114e-05,
+      "loss": 0.9181,
+      "step": 15950
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3029958948379773e-05,
+      "loss": 0.8772,
+      "step": 15960
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.302559175473841e-05,
+      "loss": 0.8677,
+      "step": 15970
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302122456109704e-05,
+      "loss": 0.8469,
+      "step": 15980
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.301685736745567e-05,
+      "loss": 0.9965,
+      "step": 15990
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3012490173814306e-05,
+      "loss": 0.8987,
+      "step": 16000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3008122980172946e-05,
+      "loss": 1.0017,
+      "step": 16010
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.300375578653157e-05,
+      "loss": 0.8191,
+      "step": 16020
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.299938859289021e-05,
+      "loss": 0.9799,
+      "step": 16030
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2995021399248846e-05,
+      "loss": 0.8877,
+      "step": 16040
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.299065420560748e-05,
+      "loss": 1.0045,
+      "step": 16050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.298628701196611e-05,
+      "loss": 0.9374,
+      "step": 16060
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2981919818324745e-05,
+      "loss": 0.8568,
+      "step": 16070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.297755262468338e-05,
+      "loss": 0.9129,
+      "step": 16080
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.297318543104201e-05,
+      "loss": 0.8271,
+      "step": 16090
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.296881823740065e-05,
+      "loss": 1.0483,
+      "step": 16100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2964451043759285e-05,
+      "loss": 0.8711,
+      "step": 16110
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.296008385011792e-05,
+      "loss": 0.8456,
+      "step": 16120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.295571665647655e-05,
+      "loss": 0.9166,
+      "step": 16130
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2951349462835185e-05,
+      "loss": 0.8962,
+      "step": 16140
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.294698226919382e-05,
+      "loss": 0.8397,
+      "step": 16150
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.294261507555245e-05,
+      "loss": 0.8275,
+      "step": 16160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2938247881911084e-05,
+      "loss": 0.788,
+      "step": 16170
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.293388068826972e-05,
+      "loss": 0.8464,
+      "step": 16180
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.292951349462835e-05,
+      "loss": 0.889,
+      "step": 16190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.292514630098699e-05,
+      "loss": 0.9127,
+      "step": 16200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2920779107345624e-05,
+      "loss": 0.8973,
+      "step": 16210
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.291641191370426e-05,
+      "loss": 0.9193,
+      "step": 16220
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.291204472006289e-05,
+      "loss": 0.8628,
+      "step": 16230
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2907677526421523e-05,
+      "loss": 0.9211,
+      "step": 16240
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.290331033278016e-05,
+      "loss": 0.8169,
+      "step": 16250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.289894313913879e-05,
+      "loss": 0.875,
+      "step": 16260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.289457594549742e-05,
+      "loss": 0.8632,
+      "step": 16270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2890208751856056e-05,
+      "loss": 0.9422,
+      "step": 16280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2885841558214696e-05,
+      "loss": 0.8801,
+      "step": 16290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.288147436457333e-05,
+      "loss": 0.9036,
+      "step": 16300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.287710717093196e-05,
+      "loss": 0.9846,
+      "step": 16310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2872739977290596e-05,
+      "loss": 0.8923,
+      "step": 16320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.286837278364923e-05,
+      "loss": 0.9101,
+      "step": 16330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.286400559000786e-05,
+      "loss": 0.9685,
+      "step": 16340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2859638396366495e-05,
+      "loss": 0.9544,
+      "step": 16350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2855271202725135e-05,
+      "loss": 0.9659,
+      "step": 16360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.285090400908376e-05,
+      "loss": 0.9084,
+      "step": 16370
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.28465368154424e-05,
+      "loss": 0.8866,
+      "step": 16380
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.284216962180103e-05,
+      "loss": 0.9668,
+      "step": 16390
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.283780242815967e-05,
+      "loss": 0.8922,
+      "step": 16400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.28334352345183e-05,
+      "loss": 0.8499,
+      "step": 16410
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2829068040876935e-05,
+      "loss": 0.8611,
+      "step": 16420
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282470084723557e-05,
+      "loss": 0.8986,
+      "step": 16430
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.28203336535942e-05,
+      "loss": 0.8972,
+      "step": 16440
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.281596645995284e-05,
+      "loss": 0.8521,
+      "step": 16450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.281159926631147e-05,
+      "loss": 0.8603,
+      "step": 16460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.280723207267011e-05,
+      "loss": 0.8369,
+      "step": 16470
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2802864879028734e-05,
+      "loss": 0.8814,
+      "step": 16480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2798497685387374e-05,
+      "loss": 0.9379,
+      "step": 16490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.279413049174601e-05,
+      "loss": 0.8773,
+      "step": 16500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.278976329810464e-05,
+      "loss": 0.9398,
+      "step": 16510
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2785396104463273e-05,
+      "loss": 0.9155,
+      "step": 16520
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.278102891082191e-05,
+      "loss": 0.8884,
+      "step": 16530
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.277666171718055e-05,
+      "loss": 0.9837,
+      "step": 16540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.277229452353917e-05,
+      "loss": 0.9552,
+      "step": 16550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.276792732989781e-05,
+      "loss": 0.8481,
+      "step": 16560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.276356013625644e-05,
+      "loss": 0.8376,
+      "step": 16570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.275919294261508e-05,
+      "loss": 1.0289,
+      "step": 16580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2754825748973706e-05,
+      "loss": 0.9598,
+      "step": 16590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2750458555332346e-05,
+      "loss": 0.7642,
+      "step": 16600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.274609136169098e-05,
+      "loss": 0.9944,
+      "step": 16610
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.274172416804961e-05,
+      "loss": 0.8965,
+      "step": 16620
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.273735697440825e-05,
+      "loss": 0.8526,
+      "step": 16630
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.273298978076688e-05,
+      "loss": 0.9839,
+      "step": 16640
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.272862258712552e-05,
+      "loss": 0.9198,
+      "step": 16650
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2724255393484145e-05,
+      "loss": 0.9558,
+      "step": 16660
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2719888199842785e-05,
+      "loss": 0.9056,
+      "step": 16670
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.271552100620141e-05,
+      "loss": 0.9045,
+      "step": 16680
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.271115381256005e-05,
+      "loss": 0.9077,
+      "step": 16690
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2706786618918685e-05,
+      "loss": 0.9907,
+      "step": 16700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.270241942527732e-05,
+      "loss": 0.7794,
+      "step": 16710
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.269805223163595e-05,
+      "loss": 0.8631,
+      "step": 16720
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2693685037994584e-05,
+      "loss": 0.8454,
+      "step": 16730
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2689317844353224e-05,
+      "loss": 0.7876,
+      "step": 16740
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.268495065071185e-05,
+      "loss": 0.8867,
+      "step": 16750
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.268058345707049e-05,
+      "loss": 0.8463,
+      "step": 16760
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2676216263429124e-05,
+      "loss": 0.9008,
+      "step": 16770
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.267184906978776e-05,
+      "loss": 0.8707,
+      "step": 16780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.266748187614639e-05,
+      "loss": 1.0162,
+      "step": 16790
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2663114682505023e-05,
+      "loss": 0.8805,
+      "step": 16800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.265874748886366e-05,
+      "loss": 0.8647,
+      "step": 16810
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.265438029522229e-05,
+      "loss": 0.8551,
+      "step": 16820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.265001310158093e-05,
+      "loss": 0.8118,
+      "step": 16830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2645645907939556e-05,
+      "loss": 0.9197,
+      "step": 16840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2641278714298196e-05,
+      "loss": 0.9326,
+      "step": 16850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.263691152065683e-05,
+      "loss": 0.9563,
+      "step": 16860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.263254432701546e-05,
+      "loss": 0.8319,
+      "step": 16870
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2628177133374096e-05,
+      "loss": 0.8515,
+      "step": 16880
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.262380993973273e-05,
+      "loss": 0.9814,
+      "step": 16890
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.261944274609136e-05,
+      "loss": 0.9255,
+      "step": 16900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2615075552449995e-05,
+      "loss": 0.974,
+      "step": 16910
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.261070835880863e-05,
+      "loss": 0.8692,
+      "step": 16920
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.260634116516726e-05,
+      "loss": 0.8521,
+      "step": 16930
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.26019739715259e-05,
+      "loss": 0.8625,
+      "step": 16940
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2597606777884535e-05,
+      "loss": 0.9537,
+      "step": 16950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.259323958424317e-05,
+      "loss": 0.8031,
+      "step": 16960
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.25888723906018e-05,
+      "loss": 0.8392,
+      "step": 16970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2584505196960435e-05,
+      "loss": 0.9705,
+      "step": 16980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.258013800331907e-05,
+      "loss": 0.8809,
+      "step": 16990
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.25757708096777e-05,
+      "loss": 0.9855,
+      "step": 17000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2571403616036334e-05,
+      "loss": 0.8754,
+      "step": 17010
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2567036422394974e-05,
+      "loss": 0.8633,
+      "step": 17020
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.256266922875361e-05,
+      "loss": 0.9082,
+      "step": 17030
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.255830203511224e-05,
+      "loss": 0.942,
+      "step": 17040
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2553934841470874e-05,
+      "loss": 0.8263,
+      "step": 17050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.254956764782951e-05,
+      "loss": 1.0218,
+      "step": 17060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.254520045418814e-05,
+      "loss": 0.9534,
+      "step": 17070
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2540833260546773e-05,
+      "loss": 0.8879,
+      "step": 17080
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.253646606690541e-05,
+      "loss": 0.8308,
+      "step": 17090
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.253209887326404e-05,
+      "loss": 0.9103,
+      "step": 17100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.252773167962268e-05,
+      "loss": 0.8881,
+      "step": 17110
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2523364485981306e-05,
+      "loss": 0.9397,
+      "step": 17120
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2518997292339946e-05,
+      "loss": 0.8297,
+      "step": 17130
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.251463009869858e-05,
+      "loss": 0.8779,
+      "step": 17140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.251026290505721e-05,
+      "loss": 0.9187,
+      "step": 17150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2505895711415846e-05,
+      "loss": 0.9123,
+      "step": 17160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.250152851777448e-05,
+      "loss": 0.8494,
+      "step": 17170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.249716132413312e-05,
+      "loss": 0.9272,
+      "step": 17180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2492794130491745e-05,
+      "loss": 1.0028,
+      "step": 17190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2488426936850385e-05,
+      "loss": 0.8532,
+      "step": 17200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.248405974320901e-05,
+      "loss": 0.9382,
+      "step": 17210
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.247969254956765e-05,
+      "loss": 1.0165,
+      "step": 17220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2475325355926285e-05,
+      "loss": 0.9373,
+      "step": 17230
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247095816228492e-05,
+      "loss": 0.7885,
+      "step": 17240
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.246659096864355e-05,
+      "loss": 0.873,
+      "step": 17250
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2462223775002185e-05,
+      "loss": 1.0064,
+      "step": 17260
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2457856581360825e-05,
+      "loss": 0.9196,
+      "step": 17270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.245348938771945e-05,
+      "loss": 0.9122,
+      "step": 17280
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.244912219407809e-05,
+      "loss": 0.8801,
+      "step": 17290
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.244475500043672e-05,
+      "loss": 1.0196,
+      "step": 17300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.244038780679536e-05,
+      "loss": 0.9083,
+      "step": 17310
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2436020613153984e-05,
+      "loss": 0.8368,
+      "step": 17320
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2431653419512624e-05,
+      "loss": 0.9514,
+      "step": 17330
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.242728622587126e-05,
+      "loss": 0.9155,
+      "step": 17340
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.242291903222989e-05,
+      "loss": 0.9076,
+      "step": 17350
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.241855183858853e-05,
+      "loss": 0.9341,
+      "step": 17360
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.241418464494716e-05,
+      "loss": 1.0474,
+      "step": 17370
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.24098174513058e-05,
+      "loss": 0.8937,
+      "step": 17380
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.240545025766442e-05,
+      "loss": 0.8946,
+      "step": 17390
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.240108306402306e-05,
+      "loss": 0.8847,
+      "step": 17400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.239671587038169e-05,
+      "loss": 0.943,
+      "step": 17410
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.239234867674033e-05,
+      "loss": 0.9523,
+      "step": 17420
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.238798148309896e-05,
+      "loss": 0.8895,
+      "step": 17430
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2383614289457596e-05,
+      "loss": 0.9144,
+      "step": 17440
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.237924709581623e-05,
+      "loss": 0.877,
+      "step": 17450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.237487990217486e-05,
+      "loss": 0.8782,
+      "step": 17460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.23705127085335e-05,
+      "loss": 0.9803,
+      "step": 17470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.236614551489213e-05,
+      "loss": 0.9307,
+      "step": 17480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.236177832125077e-05,
+      "loss": 0.8361,
+      "step": 17490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2357411127609395e-05,
+      "loss": 0.884,
+      "step": 17500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2353043933968035e-05,
+      "loss": 0.909,
+      "step": 17510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.234867674032667e-05,
+      "loss": 0.857,
+      "step": 17520
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.23443095466853e-05,
+      "loss": 0.8743,
+      "step": 17530
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2339942353043935e-05,
+      "loss": 0.8617,
+      "step": 17540
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.233557515940257e-05,
+      "loss": 0.9176,
+      "step": 17550
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.233120796576121e-05,
+      "loss": 1.0353,
+      "step": 17560
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2326840772119834e-05,
+      "loss": 0.8611,
+      "step": 17570
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2322473578478474e-05,
+      "loss": 0.9295,
+      "step": 17580
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.231810638483711e-05,
+      "loss": 0.8975,
+      "step": 17590
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.231373919119574e-05,
+      "loss": 0.9232,
+      "step": 17600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2309371997554374e-05,
+      "loss": 0.9378,
+      "step": 17610
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.230500480391301e-05,
+      "loss": 0.8603,
+      "step": 17620
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.230063761027164e-05,
+      "loss": 0.8299,
+      "step": 17630
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2296270416630273e-05,
+      "loss": 0.9498,
+      "step": 17640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.229190322298891e-05,
+      "loss": 0.9334,
+      "step": 17650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.228753602934754e-05,
+      "loss": 0.922,
+      "step": 17660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.228316883570618e-05,
+      "loss": 0.8291,
+      "step": 17670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.227880164206481e-05,
+      "loss": 0.9326,
+      "step": 17680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2274434448423446e-05,
+      "loss": 0.9288,
+      "step": 17690
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.227006725478208e-05,
+      "loss": 0.9133,
+      "step": 17700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.226570006114071e-05,
+      "loss": 1.0129,
+      "step": 17710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2261332867499346e-05,
+      "loss": 0.9741,
+      "step": 17720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.225696567385798e-05,
+      "loss": 0.8641,
+      "step": 17730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.225259848021661e-05,
+      "loss": 0.9729,
+      "step": 17740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2248231286575245e-05,
+      "loss": 0.8942,
+      "step": 17750
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2243864092933885e-05,
+      "loss": 0.8583,
+      "step": 17760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.223949689929252e-05,
+      "loss": 0.83,
+      "step": 17770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.223512970565115e-05,
+      "loss": 1.019,
+      "step": 17780
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2230762512009785e-05,
+      "loss": 0.9381,
+      "step": 17790
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.222639531836842e-05,
+      "loss": 0.8856,
+      "step": 17800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.222202812472705e-05,
+      "loss": 0.9126,
+      "step": 17810
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2217660931085685e-05,
+      "loss": 0.8096,
+      "step": 17820
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.221329373744432e-05,
+      "loss": 0.8637,
+      "step": 17830
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.220892654380296e-05,
+      "loss": 0.7979,
+      "step": 17840
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.220455935016159e-05,
+      "loss": 0.8867,
+      "step": 17850
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2200192156520224e-05,
+      "loss": 0.9575,
+      "step": 17860
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.219582496287886e-05,
+      "loss": 0.8894,
+      "step": 17870
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.219145776923749e-05,
+      "loss": 0.8141,
+      "step": 17880
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2187090575596124e-05,
+      "loss": 0.9643,
+      "step": 17890
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.218272338195476e-05,
+      "loss": 0.8852,
+      "step": 17900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.217835618831339e-05,
+      "loss": 0.9744,
+      "step": 17910
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2173988994672023e-05,
+      "loss": 0.87,
+      "step": 17920
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2169621801030663e-05,
+      "loss": 0.8538,
+      "step": 17930
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.216525460738929e-05,
+      "loss": 0.9016,
+      "step": 17940
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.216088741374793e-05,
+      "loss": 0.8854,
+      "step": 17950
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.215652022010656e-05,
+      "loss": 0.925,
+      "step": 17960
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2152153026465196e-05,
+      "loss": 0.8494,
+      "step": 17970
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.214778583282383e-05,
+      "loss": 0.8484,
+      "step": 17980
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.214341863918246e-05,
+      "loss": 0.9416,
+      "step": 17990
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.21390514455411e-05,
+      "loss": 0.8182,
+      "step": 18000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.213468425189973e-05,
+      "loss": 0.8701,
+      "step": 18010
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.213031705825837e-05,
+      "loss": 0.9982,
+      "step": 18020
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2125949864616996e-05,
+      "loss": 0.8275,
+      "step": 18030
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2121582670975635e-05,
+      "loss": 0.9125,
+      "step": 18040
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.211721547733427e-05,
+      "loss": 0.845,
+      "step": 18050
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.21128482836929e-05,
+      "loss": 0.8504,
+      "step": 18060
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2108481090051535e-05,
+      "loss": 0.8587,
+      "step": 18070
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.210411389641017e-05,
+      "loss": 0.9246,
+      "step": 18080
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.209974670276881e-05,
+      "loss": 1.0316,
+      "step": 18090
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2095379509127435e-05,
+      "loss": 0.8776,
+      "step": 18100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2091012315486075e-05,
+      "loss": 0.8518,
+      "step": 18110
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.20866451218447e-05,
+      "loss": 0.9769,
+      "step": 18120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.208227792820334e-05,
+      "loss": 0.8358,
+      "step": 18130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.207791073456197e-05,
+      "loss": 0.9775,
+      "step": 18140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.207354354092061e-05,
+      "loss": 0.9359,
+      "step": 18150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.206917634727924e-05,
+      "loss": 0.9132,
+      "step": 18160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2064809153637874e-05,
+      "loss": 0.9181,
+      "step": 18170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2060441959996514e-05,
+      "loss": 0.8262,
+      "step": 18180
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.205607476635514e-05,
+      "loss": 0.8896,
+      "step": 18190
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.205170757271378e-05,
+      "loss": 0.9324,
+      "step": 18200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.204734037907241e-05,
+      "loss": 0.9263,
+      "step": 18210
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.204297318543105e-05,
+      "loss": 0.9152,
+      "step": 18220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.203860599178967e-05,
+      "loss": 0.9595,
+      "step": 18230
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.203423879814831e-05,
+      "loss": 1.0747,
+      "step": 18240
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2029871604506946e-05,
+      "loss": 0.9535,
+      "step": 18250
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.202550441086558e-05,
+      "loss": 0.8782,
+      "step": 18260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202113721722421e-05,
+      "loss": 0.904,
+      "step": 18270
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2016770023582846e-05,
+      "loss": 0.857,
+      "step": 18280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2012402829941486e-05,
+      "loss": 1.0048,
+      "step": 18290
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.200803563630011e-05,
+      "loss": 0.8859,
+      "step": 18300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.200366844265875e-05,
+      "loss": 0.8749,
+      "step": 18310
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.199930124901738e-05,
+      "loss": 0.88,
+      "step": 18320
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.199493405537602e-05,
+      "loss": 0.8392,
+      "step": 18330
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.199056686173465e-05,
+      "loss": 0.98,
+      "step": 18340
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.1986199668093285e-05,
+      "loss": 0.8203,
+      "step": 18350
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.198183247445192e-05,
+      "loss": 1.013,
+      "step": 18360
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.197746528081055e-05,
+      "loss": 0.8189,
+      "step": 18370
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.197309808716919e-05,
+      "loss": 0.8422,
+      "step": 18380
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.196873089352782e-05,
+      "loss": 0.9924,
+      "step": 18390
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.196436369988646e-05,
+      "loss": 0.8456,
+      "step": 18400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1959996506245084e-05,
+      "loss": 0.9858,
+      "step": 18410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1955629312603724e-05,
+      "loss": 0.8816,
+      "step": 18420
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.195126211896236e-05,
+      "loss": 0.9604,
+      "step": 18430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.194689492532099e-05,
+      "loss": 0.9552,
+      "step": 18440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1942527731679624e-05,
+      "loss": 0.876,
+      "step": 18450
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.193816053803826e-05,
+      "loss": 0.8493,
+      "step": 18460
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.193379334439689e-05,
+      "loss": 0.9565,
+      "step": 18470
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1929426150755524e-05,
+      "loss": 0.845,
+      "step": 18480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1925058957114163e-05,
+      "loss": 0.8699,
+      "step": 18490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.19206917634728e-05,
+      "loss": 0.8708,
+      "step": 18500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.191632456983143e-05,
+      "loss": 0.8965,
+      "step": 18510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.191195737619006e-05,
+      "loss": 0.8135,
+      "step": 18520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1907590182548696e-05,
+      "loss": 0.8992,
+      "step": 18530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.190322298890733e-05,
+      "loss": 0.9404,
+      "step": 18540
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.189885579526596e-05,
+      "loss": 0.7823,
+      "step": 18550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1894488601624596e-05,
+      "loss": 0.9082,
+      "step": 18560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.189012140798323e-05,
+      "loss": 0.9906,
+      "step": 18570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.188575421434187e-05,
+      "loss": 0.9024,
+      "step": 18580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.18813870207005e-05,
+      "loss": 0.836,
+      "step": 18590
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1877019827059136e-05,
+      "loss": 0.7802,
+      "step": 18600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.187265263341777e-05,
+      "loss": 0.8276,
+      "step": 18610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.18682854397764e-05,
+      "loss": 0.9457,
+      "step": 18620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1863918246135035e-05,
+      "loss": 0.9752,
+      "step": 18630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.185955105249367e-05,
+      "loss": 0.9981,
+      "step": 18640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.18551838588523e-05,
+      "loss": 0.9319,
+      "step": 18650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.185081666521094e-05,
+      "loss": 0.9295,
+      "step": 18660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.184644947156957e-05,
+      "loss": 0.9695,
+      "step": 18670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.184208227792821e-05,
+      "loss": 0.9095,
+      "step": 18680
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.183771508428684e-05,
+      "loss": 0.9705,
+      "step": 18690
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1833347890645474e-05,
+      "loss": 0.901,
+      "step": 18700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.182898069700411e-05,
+      "loss": 0.8911,
+      "step": 18710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182461350336274e-05,
+      "loss": 0.8377,
+      "step": 18720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1820246309721374e-05,
+      "loss": 0.9186,
+      "step": 18730
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.181587911608001e-05,
+      "loss": 0.8603,
+      "step": 18740
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.181151192243865e-05,
+      "loss": 0.8852,
+      "step": 18750
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1807144728797274e-05,
+      "loss": 1.0386,
+      "step": 18760
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1802777535155914e-05,
+      "loss": 0.9363,
+      "step": 18770
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.179841034151455e-05,
+      "loss": 0.9045,
+      "step": 18780
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.179404314787318e-05,
+      "loss": 0.9825,
+      "step": 18790
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.178967595423181e-05,
+      "loss": 0.8761,
+      "step": 18800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1785308760590446e-05,
+      "loss": 0.8271,
+      "step": 18810
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.178094156694908e-05,
+      "loss": 0.8955,
+      "step": 18820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.177657437330771e-05,
+      "loss": 0.9342,
+      "step": 18830
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.177220717966635e-05,
+      "loss": 0.8671,
+      "step": 18840
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.176783998602498e-05,
+      "loss": 0.8368,
+      "step": 18850
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.176347279238362e-05,
+      "loss": 1.0078,
+      "step": 18860
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1759105598742246e-05,
+      "loss": 0.8837,
+      "step": 18870
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1754738405100886e-05,
+      "loss": 0.9034,
+      "step": 18880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.175037121145952e-05,
+      "loss": 0.8497,
+      "step": 18890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.174600401781815e-05,
+      "loss": 0.8728,
+      "step": 18900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.174163682417679e-05,
+      "loss": 0.9399,
+      "step": 18910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.173726963053542e-05,
+      "loss": 1.0085,
+      "step": 18920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.173290243689406e-05,
+      "loss": 0.9641,
+      "step": 18930
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1728535243252685e-05,
+      "loss": 0.8724,
+      "step": 18940
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1724168049611325e-05,
+      "loss": 0.885,
+      "step": 18950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.171980085596995e-05,
+      "loss": 0.8666,
+      "step": 18960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.171543366232859e-05,
+      "loss": 0.885,
+      "step": 18970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1711066468687224e-05,
+      "loss": 0.8987,
+      "step": 18980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.170669927504586e-05,
+      "loss": 0.8221,
+      "step": 18990
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.170233208140449e-05,
+      "loss": 0.9286,
+      "step": 19000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1697964887763124e-05,
+      "loss": 1.017,
+      "step": 19010
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1693597694121764e-05,
+      "loss": 0.8988,
+      "step": 19020
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.168923050048039e-05,
+      "loss": 0.8287,
+      "step": 19030
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.168486330683903e-05,
+      "loss": 0.8693,
+      "step": 19040
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.168049611319766e-05,
+      "loss": 0.9094,
+      "step": 19050
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.16761289195563e-05,
+      "loss": 0.8359,
+      "step": 19060
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.167176172591493e-05,
+      "loss": 0.9595,
+      "step": 19070
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.166739453227356e-05,
+      "loss": 0.882,
+      "step": 19080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1663027338632196e-05,
+      "loss": 1.0107,
+      "step": 19090
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.165866014499083e-05,
+      "loss": 0.8231,
+      "step": 19100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.165429295134947e-05,
+      "loss": 0.7482,
+      "step": 19110
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1649925757708096e-05,
+      "loss": 0.84,
+      "step": 19120
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1645558564066736e-05,
+      "loss": 1.0007,
+      "step": 19130
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.164119137042536e-05,
+      "loss": 0.8836,
+      "step": 19140
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1636824176784e-05,
+      "loss": 0.8767,
+      "step": 19150
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1632456983142636e-05,
+      "loss": 0.8825,
+      "step": 19160
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.162808978950127e-05,
+      "loss": 0.8629,
+      "step": 19170
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.16237225958599e-05,
+      "loss": 0.8175,
+      "step": 19180
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1619355402218535e-05,
+      "loss": 0.9335,
+      "step": 19190
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.161498820857717e-05,
+      "loss": 0.9198,
+      "step": 19200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.16106210149358e-05,
+      "loss": 0.8916,
+      "step": 19210
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.160625382129444e-05,
+      "loss": 0.9295,
+      "step": 19220
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.160188662765307e-05,
+      "loss": 0.9003,
+      "step": 19230
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.159751943401171e-05,
+      "loss": 0.7715,
+      "step": 19240
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.159315224037034e-05,
+      "loss": 0.9378,
+      "step": 19250
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1588785046728974e-05,
+      "loss": 0.847,
+      "step": 19260
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.158441785308761e-05,
+      "loss": 0.8368,
+      "step": 19270
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.158005065944624e-05,
+      "loss": 0.8627,
+      "step": 19280
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1575683465804874e-05,
+      "loss": 0.907,
+      "step": 19290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157131627216351e-05,
+      "loss": 0.8004,
+      "step": 19300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.156694907852215e-05,
+      "loss": 0.9652,
+      "step": 19310
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.156258188488078e-05,
+      "loss": 0.8974,
+      "step": 19320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1558214691239414e-05,
+      "loss": 0.9365,
+      "step": 19330
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.155384749759805e-05,
+      "loss": 0.8668,
+      "step": 19340
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.154948030395668e-05,
+      "loss": 0.9372,
+      "step": 19350
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.154511311031531e-05,
+      "loss": 0.9489,
+      "step": 19360
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1540745916673946e-05,
+      "loss": 0.8398,
+      "step": 19370
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.153637872303258e-05,
+      "loss": 0.8952,
+      "step": 19380
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.153201152939121e-05,
+      "loss": 0.927,
+      "step": 19390
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1527644335749846e-05,
+      "loss": 1.0132,
+      "step": 19400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1523277142108486e-05,
+      "loss": 0.8372,
+      "step": 19410
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.151890994846712e-05,
+      "loss": 0.7357,
+      "step": 19420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.151454275482575e-05,
+      "loss": 0.9532,
+      "step": 19430
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1510175561184386e-05,
+      "loss": 0.933,
+      "step": 19440
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.150580836754302e-05,
+      "loss": 0.881,
+      "step": 19450
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.150144117390165e-05,
+      "loss": 0.8622,
+      "step": 19460
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1497073980260285e-05,
+      "loss": 0.7896,
+      "step": 19470
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1492706786618925e-05,
+      "loss": 0.9075,
+      "step": 19480
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.148833959297755e-05,
+      "loss": 0.927,
+      "step": 19490
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.148397239933619e-05,
+      "loss": 0.7989,
+      "step": 19500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1479605205694825e-05,
+      "loss": 0.8322,
+      "step": 19510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.147523801205346e-05,
+      "loss": 0.8631,
+      "step": 19520
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.147087081841209e-05,
+      "loss": 0.951,
+      "step": 19530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1466503624770724e-05,
+      "loss": 0.9259,
+      "step": 19540
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.146213643112936e-05,
+      "loss": 0.8473,
+      "step": 19550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.145776923748799e-05,
+      "loss": 0.8849,
+      "step": 19560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.145340204384663e-05,
+      "loss": 0.9134,
+      "step": 19570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.144903485020526e-05,
+      "loss": 0.8946,
+      "step": 19580
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.14446676565639e-05,
+      "loss": 0.8402,
+      "step": 19590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1440300462922524e-05,
+      "loss": 0.8189,
+      "step": 19600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1435933269281164e-05,
+      "loss": 0.9458,
+      "step": 19610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.14315660756398e-05,
+      "loss": 0.8323,
+      "step": 19620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.142719888199843e-05,
+      "loss": 0.8898,
+      "step": 19630
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.142283168835706e-05,
+      "loss": 0.9271,
+      "step": 19640
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1418464494715696e-05,
+      "loss": 0.815,
+      "step": 19650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1414097301074336e-05,
+      "loss": 0.8342,
+      "step": 19660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.140973010743296e-05,
+      "loss": 0.801,
+      "step": 19670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.14053629137916e-05,
+      "loss": 0.9814,
+      "step": 19680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.140099572015023e-05,
+      "loss": 0.8655,
+      "step": 19690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.139662852650887e-05,
+      "loss": 0.9116,
+      "step": 19700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.13922613328675e-05,
+      "loss": 0.9432,
+      "step": 19710
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1387894139226136e-05,
+      "loss": 0.9369,
+      "step": 19720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.138352694558477e-05,
+      "loss": 0.8762,
+      "step": 19730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.13791597519434e-05,
+      "loss": 0.9253,
+      "step": 19740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137479255830204e-05,
+      "loss": 0.7886,
+      "step": 19750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137042536466067e-05,
+      "loss": 0.9707,
+      "step": 19760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.136605817101931e-05,
+      "loss": 0.8852,
+      "step": 19770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1361690977377935e-05,
+      "loss": 0.8879,
+      "step": 19780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1357323783736575e-05,
+      "loss": 0.9292,
+      "step": 19790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.13529565900952e-05,
+      "loss": 0.823,
+      "step": 19800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.134858939645384e-05,
+      "loss": 0.9564,
+      "step": 19810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1344222202812474e-05,
+      "loss": 0.9664,
+      "step": 19820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.133985500917111e-05,
+      "loss": 0.8088,
+      "step": 19830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.133548781552975e-05,
+      "loss": 0.8775,
+      "step": 19840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1331120621888374e-05,
+      "loss": 0.8847,
+      "step": 19850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1326753428247014e-05,
+      "loss": 0.9577,
+      "step": 19860
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.132238623460564e-05,
+      "loss": 0.8617,
+      "step": 19870
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.131801904096428e-05,
+      "loss": 0.9295,
+      "step": 19880
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1313651847322914e-05,
+      "loss": 0.8586,
+      "step": 19890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.130928465368155e-05,
+      "loss": 0.9147,
+      "step": 19900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.130491746004018e-05,
+      "loss": 0.859,
+      "step": 19910
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.130055026639881e-05,
+      "loss": 0.9751,
+      "step": 19920
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1296183072757446e-05,
+      "loss": 1.0088,
+      "step": 19930
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.129181587911608e-05,
+      "loss": 0.9525,
+      "step": 19940
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.128744868547472e-05,
+      "loss": 0.8735,
+      "step": 19950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1283081491833346e-05,
+      "loss": 1.0307,
+      "step": 19960
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1278714298191986e-05,
+      "loss": 0.8379,
+      "step": 19970
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127434710455062e-05,
+      "loss": 0.899,
+      "step": 19980
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.126997991090925e-05,
+      "loss": 0.8703,
+      "step": 19990
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1265612717267886e-05,
+      "loss": 1.0364,
+      "step": 20000
+    },
+    {
+      "epoch": 1.75,
+      "eval_accuracy": 0.574629065539943,
+      "eval_loss": 0.9053159356117249,
+      "eval_runtime": 84.1188,
+      "eval_samples_per_second": 120.984,
+      "eval_steps_per_second": 15.133,
+      "step": 20000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.126124552362652e-05,
+      "loss": 0.8943,
+      "step": 20010
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.125687832998515e-05,
+      "loss": 0.8525,
+      "step": 20020
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1252511136343785e-05,
+      "loss": 0.9,
+      "step": 20030
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1248143942702425e-05,
+      "loss": 0.8541,
+      "step": 20040
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.124377674906105e-05,
+      "loss": 0.821,
+      "step": 20050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.123940955541969e-05,
+      "loss": 0.8161,
+      "step": 20060
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1235042361778325e-05,
+      "loss": 0.8613,
+      "step": 20070
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.123067516813696e-05,
+      "loss": 0.8843,
+      "step": 20080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.122630797449559e-05,
+      "loss": 0.8546,
+      "step": 20090
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1221940780854224e-05,
+      "loss": 0.8802,
+      "step": 20100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.121757358721286e-05,
+      "loss": 0.8566,
+      "step": 20110
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.121320639357149e-05,
+      "loss": 0.8757,
+      "step": 20120
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1208839199930124e-05,
+      "loss": 0.9832,
+      "step": 20130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1204472006288764e-05,
+      "loss": 0.8816,
+      "step": 20140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.12001048126474e-05,
+      "loss": 0.7858,
+      "step": 20150
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.119573761900603e-05,
+      "loss": 0.9879,
+      "step": 20160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1191370425364664e-05,
+      "loss": 1.0506,
+      "step": 20170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.11870032317233e-05,
+      "loss": 0.9166,
+      "step": 20180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.118263603808193e-05,
+      "loss": 0.8969,
+      "step": 20190
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.117826884444056e-05,
+      "loss": 0.7902,
+      "step": 20200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1173901650799196e-05,
+      "loss": 0.9282,
+      "step": 20210
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.116953445715783e-05,
+      "loss": 0.8329,
+      "step": 20220
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.116516726351647e-05,
+      "loss": 1.0167,
+      "step": 20230
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.11608000698751e-05,
+      "loss": 0.9792,
+      "step": 20240
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1156432876233736e-05,
+      "loss": 0.9326,
+      "step": 20250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.115206568259237e-05,
+      "loss": 0.9197,
+      "step": 20260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1147698488951e-05,
+      "loss": 0.8049,
+      "step": 20270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1143331295309636e-05,
+      "loss": 0.9345,
+      "step": 20280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.113896410166827e-05,
+      "loss": 0.8331,
+      "step": 20290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.11345969080269e-05,
+      "loss": 1.0869,
+      "step": 20300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1130229714385535e-05,
+      "loss": 0.8144,
+      "step": 20310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1125862520744175e-05,
+      "loss": 1.0061,
+      "step": 20320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.11214953271028e-05,
+      "loss": 0.9571,
+      "step": 20330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.111712813346144e-05,
+      "loss": 0.8798,
+      "step": 20340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1112760939820075e-05,
+      "loss": 0.804,
+      "step": 20350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.110839374617871e-05,
+      "loss": 0.8891,
+      "step": 20360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.110402655253734e-05,
+      "loss": 0.8569,
+      "step": 20370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1099659358895974e-05,
+      "loss": 0.9317,
+      "step": 20380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1095292165254614e-05,
+      "loss": 0.8127,
+      "step": 20390
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.109092497161324e-05,
+      "loss": 0.8602,
+      "step": 20400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.108655777797188e-05,
+      "loss": 0.8579,
+      "step": 20410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.108219058433051e-05,
+      "loss": 0.8826,
+      "step": 20420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.107782339068915e-05,
+      "loss": 0.8966,
+      "step": 20430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.107345619704778e-05,
+      "loss": 0.9043,
+      "step": 20440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1069089003406414e-05,
+      "loss": 1.0149,
+      "step": 20450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.106472180976505e-05,
+      "loss": 0.9105,
+      "step": 20460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.106035461612368e-05,
+      "loss": 0.7942,
+      "step": 20470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.105598742248232e-05,
+      "loss": 0.8985,
+      "step": 20480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1051620228840946e-05,
+      "loss": 0.892,
+      "step": 20490
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1047253035199586e-05,
+      "loss": 0.8625,
+      "step": 20500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.104288584155821e-05,
+      "loss": 0.9714,
+      "step": 20510
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.103851864791685e-05,
+      "loss": 1.0135,
+      "step": 20520
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.103415145427548e-05,
+      "loss": 0.8904,
+      "step": 20530
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.102978426063412e-05,
+      "loss": 0.8925,
+      "step": 20540
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.102541706699275e-05,
+      "loss": 0.9535,
+      "step": 20550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1021049873351386e-05,
+      "loss": 0.9517,
+      "step": 20560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1016682679710026e-05,
+      "loss": 0.9011,
+      "step": 20570
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.101231548606865e-05,
+      "loss": 0.9281,
+      "step": 20580
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.100794829242729e-05,
+      "loss": 0.9168,
+      "step": 20590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.100358109878592e-05,
+      "loss": 0.911,
+      "step": 20600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.099921390514456e-05,
+      "loss": 0.891,
+      "step": 20610
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.0994846711503185e-05,
+      "loss": 0.857,
+      "step": 20620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.0990479517861825e-05,
+      "loss": 0.794,
+      "step": 20630
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.098611232422046e-05,
+      "loss": 0.8943,
+      "step": 20640
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.098174513057909e-05,
+      "loss": 0.839,
+      "step": 20650
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.0977377936937724e-05,
+      "loss": 0.8686,
+      "step": 20660
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.097301074329636e-05,
+      "loss": 0.8671,
+      "step": 20670
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0968643549655e-05,
+      "loss": 0.8398,
+      "step": 20680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0964276356013624e-05,
+      "loss": 0.8708,
+      "step": 20690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0959909162372264e-05,
+      "loss": 0.9895,
+      "step": 20700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.095554196873089e-05,
+      "loss": 0.9162,
+      "step": 20710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.095117477508953e-05,
+      "loss": 0.8487,
+      "step": 20720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0946807581448164e-05,
+      "loss": 0.9237,
+      "step": 20730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.09424403878068e-05,
+      "loss": 0.8511,
+      "step": 20740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.093807319416543e-05,
+      "loss": 0.9194,
+      "step": 20750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.093370600052406e-05,
+      "loss": 0.8543,
+      "step": 20760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.09293388068827e-05,
+      "loss": 0.8915,
+      "step": 20770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092497161324133e-05,
+      "loss": 0.8599,
+      "step": 20780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092060441959997e-05,
+      "loss": 0.9343,
+      "step": 20790
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.09162372259586e-05,
+      "loss": 0.8082,
+      "step": 20800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0911870032317236e-05,
+      "loss": 0.7932,
+      "step": 20810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.090750283867587e-05,
+      "loss": 0.9403,
+      "step": 20820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.09031356450345e-05,
+      "loss": 0.8183,
+      "step": 20830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0898768451393136e-05,
+      "loss": 0.8393,
+      "step": 20840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.089440125775177e-05,
+      "loss": 0.7841,
+      "step": 20850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.08900340641104e-05,
+      "loss": 0.8817,
+      "step": 20860
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0885666870469035e-05,
+      "loss": 0.9287,
+      "step": 20870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0881299676827675e-05,
+      "loss": 0.7928,
+      "step": 20880
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.087693248318631e-05,
+      "loss": 0.9636,
+      "step": 20890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.087256528954494e-05,
+      "loss": 0.9103,
+      "step": 20900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0868198095903575e-05,
+      "loss": 0.8192,
+      "step": 20910
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.086383090226221e-05,
+      "loss": 0.9612,
+      "step": 20920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.085946370862084e-05,
+      "loss": 0.8878,
+      "step": 20930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0855096514979474e-05,
+      "loss": 0.8935,
+      "step": 20940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.085072932133811e-05,
+      "loss": 0.9094,
+      "step": 20950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.084636212769675e-05,
+      "loss": 0.9467,
+      "step": 20960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.084199493405538e-05,
+      "loss": 0.9798,
+      "step": 20970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0837627740414014e-05,
+      "loss": 0.7955,
+      "step": 20980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.083326054677265e-05,
+      "loss": 0.8728,
+      "step": 20990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.082889335313128e-05,
+      "loss": 0.9996,
+      "step": 21000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0824526159489914e-05,
+      "loss": 0.998,
+      "step": 21010
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.082015896584855e-05,
+      "loss": 0.861,
+      "step": 21020
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.081579177220718e-05,
+      "loss": 0.8528,
+      "step": 21030
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.081142457856581e-05,
+      "loss": 0.8457,
+      "step": 21040
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.080705738492445e-05,
+      "loss": 0.8819,
+      "step": 21050
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.080269019128308e-05,
+      "loss": 0.9683,
+      "step": 21060
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.079832299764172e-05,
+      "loss": 0.957,
+      "step": 21070
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.079395580400035e-05,
+      "loss": 0.9263,
+      "step": 21080
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0789588610358986e-05,
+      "loss": 0.7987,
+      "step": 21090
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.078522141671762e-05,
+      "loss": 0.9648,
+      "step": 21100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.078085422307625e-05,
+      "loss": 0.907,
+      "step": 21110
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0776487029434886e-05,
+      "loss": 0.874,
+      "step": 21120
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.077211983579352e-05,
+      "loss": 0.9533,
+      "step": 21130
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.076775264215216e-05,
+      "loss": 0.973,
+      "step": 21140
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0763385448510785e-05,
+      "loss": 0.8306,
+      "step": 21150
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0759018254869425e-05,
+      "loss": 0.7781,
+      "step": 21160
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.075465106122806e-05,
+      "loss": 0.9331,
+      "step": 21170
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.075028386758669e-05,
+      "loss": 0.8894,
+      "step": 21180
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0745916673945325e-05,
+      "loss": 0.8544,
+      "step": 21190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.074154948030396e-05,
+      "loss": 0.9052,
+      "step": 21200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.07371822866626e-05,
+      "loss": 0.7815,
+      "step": 21210
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0732815093021224e-05,
+      "loss": 0.7964,
+      "step": 21220
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0728447899379864e-05,
+      "loss": 0.877,
+      "step": 21230
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.072408070573849e-05,
+      "loss": 0.8209,
+      "step": 21240
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.071971351209713e-05,
+      "loss": 0.9611,
+      "step": 21250
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.071534631845576e-05,
+      "loss": 0.8529,
+      "step": 21260
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.07109791248144e-05,
+      "loss": 0.8635,
+      "step": 21270
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.070661193117303e-05,
+      "loss": 0.94,
+      "step": 21280
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0702244737531664e-05,
+      "loss": 0.9409,
+      "step": 21290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0697877543890304e-05,
+      "loss": 0.9844,
+      "step": 21300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.069351035024893e-05,
+      "loss": 0.9141,
+      "step": 21310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.068914315660757e-05,
+      "loss": 0.9606,
+      "step": 21320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0684775962966196e-05,
+      "loss": 0.8466,
+      "step": 21330
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0680408769324836e-05,
+      "loss": 0.8608,
+      "step": 21340
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.067604157568346e-05,
+      "loss": 0.9744,
+      "step": 21350
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.06716743820421e-05,
+      "loss": 0.9645,
+      "step": 21360
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0667307188400736e-05,
+      "loss": 0.8962,
+      "step": 21370
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.066293999475937e-05,
+      "loss": 0.9371,
+      "step": 21380
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0658572801118e-05,
+      "loss": 0.975,
+      "step": 21390
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0654205607476636e-05,
+      "loss": 0.8093,
+      "step": 21400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0649838413835276e-05,
+      "loss": 0.9707,
+      "step": 21410
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.06454712201939e-05,
+      "loss": 0.8995,
+      "step": 21420
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.064110402655254e-05,
+      "loss": 0.8861,
+      "step": 21430
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.063673683291117e-05,
+      "loss": 0.9399,
+      "step": 21440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.063236963926981e-05,
+      "loss": 0.9286,
+      "step": 21450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.062800244562844e-05,
+      "loss": 0.9884,
+      "step": 21460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0623635251987075e-05,
+      "loss": 0.9003,
+      "step": 21470
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.061926805834571e-05,
+      "loss": 0.8981,
+      "step": 21480
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.061490086470434e-05,
+      "loss": 0.924,
+      "step": 21490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.061053367106298e-05,
+      "loss": 0.9664,
+      "step": 21500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.060616647742161e-05,
+      "loss": 1.0078,
+      "step": 21510
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.060179928378025e-05,
+      "loss": 0.9344,
+      "step": 21520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0597432090138874e-05,
+      "loss": 0.9192,
+      "step": 21530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0593064896497514e-05,
+      "loss": 0.9847,
+      "step": 21540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.058869770285615e-05,
+      "loss": 0.9889,
+      "step": 21550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.058433050921478e-05,
+      "loss": 0.8552,
+      "step": 21560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0579963315573414e-05,
+      "loss": 0.8667,
+      "step": 21570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.057559612193205e-05,
+      "loss": 0.9019,
+      "step": 21580
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.057122892829068e-05,
+      "loss": 0.8803,
+      "step": 21590
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.056686173464931e-05,
+      "loss": 0.9645,
+      "step": 21600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.056249454100795e-05,
+      "loss": 0.8544,
+      "step": 21610
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0558127347366586e-05,
+      "loss": 0.8383,
+      "step": 21620
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.055376015372522e-05,
+      "loss": 0.9774,
+      "step": 21630
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.054939296008385e-05,
+      "loss": 0.8571,
+      "step": 21640
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0545025766442486e-05,
+      "loss": 0.7946,
+      "step": 21650
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.054065857280112e-05,
+      "loss": 0.9535,
+      "step": 21660
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.053629137915975e-05,
+      "loss": 0.9974,
+      "step": 21670
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0531924185518386e-05,
+      "loss": 0.8984,
+      "step": 21680
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.052755699187702e-05,
+      "loss": 0.9009,
+      "step": 21690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052318979823566e-05,
+      "loss": 0.8197,
+      "step": 21700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.051882260459429e-05,
+      "loss": 0.8758,
+      "step": 21710
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0514455410952925e-05,
+      "loss": 0.9092,
+      "step": 21720
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.051008821731156e-05,
+      "loss": 0.8336,
+      "step": 21730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.050572102367019e-05,
+      "loss": 0.9101,
+      "step": 21740
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0501353830028825e-05,
+      "loss": 1.0601,
+      "step": 21750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.049698663638746e-05,
+      "loss": 0.9331,
+      "step": 21760
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.049261944274609e-05,
+      "loss": 0.9243,
+      "step": 21770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0488252249104724e-05,
+      "loss": 0.7971,
+      "step": 21780
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.048388505546336e-05,
+      "loss": 0.909,
+      "step": 21790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0479517861822e-05,
+      "loss": 0.8771,
+      "step": 21800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.047515066818063e-05,
+      "loss": 0.9831,
+      "step": 21810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0470783474539264e-05,
+      "loss": 0.8827,
+      "step": 21820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.04664162808979e-05,
+      "loss": 0.8779,
+      "step": 21830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.046204908725653e-05,
+      "loss": 0.7628,
+      "step": 21840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0457681893615164e-05,
+      "loss": 0.8307,
+      "step": 21850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.04533146999738e-05,
+      "loss": 0.9691,
+      "step": 21860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.044894750633244e-05,
+      "loss": 1.0388,
+      "step": 21870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.044458031269106e-05,
+      "loss": 0.9314,
+      "step": 21880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.04402131190497e-05,
+      "loss": 0.9269,
+      "step": 21890
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0435845925408336e-05,
+      "loss": 0.8417,
+      "step": 21900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.043147873176697e-05,
+      "loss": 0.8343,
+      "step": 21910
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.04271115381256e-05,
+      "loss": 0.9138,
+      "step": 21920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0422744344484236e-05,
+      "loss": 0.8187,
+      "step": 21930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.041837715084287e-05,
+      "loss": 0.9805,
+      "step": 21940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.04140099572015e-05,
+      "loss": 0.8165,
+      "step": 21950
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.040964276356014e-05,
+      "loss": 1.0771,
+      "step": 21960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.040527556991877e-05,
+      "loss": 0.725,
+      "step": 21970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.040090837627741e-05,
+      "loss": 0.7819,
+      "step": 21980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0396541182636035e-05,
+      "loss": 1.0414,
+      "step": 21990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0392173988994675e-05,
+      "loss": 0.8359,
+      "step": 22000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.038780679535331e-05,
+      "loss": 0.8497,
+      "step": 22010
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.038343960171194e-05,
+      "loss": 0.919,
+      "step": 22020
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.037907240807058e-05,
+      "loss": 0.812,
+      "step": 22030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.037470521442921e-05,
+      "loss": 0.9045,
+      "step": 22040
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.037033802078785e-05,
+      "loss": 0.8461,
+      "step": 22050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0365970827146474e-05,
+      "loss": 0.9765,
+      "step": 22060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0361603633505114e-05,
+      "loss": 0.9953,
+      "step": 22070
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.035723643986374e-05,
+      "loss": 0.8718,
+      "step": 22080
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.035286924622238e-05,
+      "loss": 0.9663,
+      "step": 22090
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0348502052581014e-05,
+      "loss": 0.9163,
+      "step": 22100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.034413485893965e-05,
+      "loss": 0.8368,
+      "step": 22110
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.033976766529828e-05,
+      "loss": 0.8405,
+      "step": 22120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0335400471656914e-05,
+      "loss": 0.9603,
+      "step": 22130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0331033278015554e-05,
+      "loss": 0.9125,
+      "step": 22140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.032666608437418e-05,
+      "loss": 0.9466,
+      "step": 22150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032229889073282e-05,
+      "loss": 0.8659,
+      "step": 22160
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0317931697091446e-05,
+      "loss": 0.8547,
+      "step": 22170
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0313564503450086e-05,
+      "loss": 0.9856,
+      "step": 22180
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.030919730980871e-05,
+      "loss": 0.8859,
+      "step": 22190
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.030483011616735e-05,
+      "loss": 0.7921,
+      "step": 22200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0300462922525986e-05,
+      "loss": 0.9071,
+      "step": 22210
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.029609572888462e-05,
+      "loss": 1.0703,
+      "step": 22220
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.029172853524326e-05,
+      "loss": 0.9665,
+      "step": 22230
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0287361341601886e-05,
+      "loss": 1.0264,
+      "step": 22240
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0282994147960526e-05,
+      "loss": 0.9011,
+      "step": 22250
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.027862695431915e-05,
+      "loss": 0.8626,
+      "step": 22260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.027425976067779e-05,
+      "loss": 0.8657,
+      "step": 22270
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0269892567036425e-05,
+      "loss": 0.712,
+      "step": 22280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.026552537339506e-05,
+      "loss": 0.9101,
+      "step": 22290
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.026115817975369e-05,
+      "loss": 0.9167,
+      "step": 22300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0256790986112325e-05,
+      "loss": 0.9731,
+      "step": 22310
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.025242379247096e-05,
+      "loss": 0.8731,
+      "step": 22320
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.024805659882959e-05,
+      "loss": 0.8458,
+      "step": 22330
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.024368940518823e-05,
+      "loss": 0.9359,
+      "step": 22340
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.023932221154686e-05,
+      "loss": 0.8079,
+      "step": 22350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.02349550179055e-05,
+      "loss": 0.9221,
+      "step": 22360
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.023058782426413e-05,
+      "loss": 0.8865,
+      "step": 22370
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0226220630622764e-05,
+      "loss": 0.827,
+      "step": 22380
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.02218534369814e-05,
+      "loss": 0.8529,
+      "step": 22390
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.021748624334003e-05,
+      "loss": 0.8286,
+      "step": 22400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0213119049698664e-05,
+      "loss": 0.9091,
+      "step": 22410
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.02087518560573e-05,
+      "loss": 0.8375,
+      "step": 22420
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.020438466241594e-05,
+      "loss": 0.8778,
+      "step": 22430
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.020001746877457e-05,
+      "loss": 0.916,
+      "step": 22440
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.01956502751332e-05,
+      "loss": 0.7706,
+      "step": 22450
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0191283081491836e-05,
+      "loss": 1.0034,
+      "step": 22460
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.018691588785047e-05,
+      "loss": 0.9329,
+      "step": 22470
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.01825486942091e-05,
+      "loss": 0.9582,
+      "step": 22480
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0178181500567736e-05,
+      "loss": 0.8396,
+      "step": 22490
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.017381430692637e-05,
+      "loss": 0.9684,
+      "step": 22500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0169447113285e-05,
+      "loss": 0.8473,
+      "step": 22510
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0165079919643636e-05,
+      "loss": 0.8606,
+      "step": 22520
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0160712726002276e-05,
+      "loss": 0.8702,
+      "step": 22530
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.015634553236091e-05,
+      "loss": 0.9901,
+      "step": 22540
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.015197833871954e-05,
+      "loss": 0.8046,
+      "step": 22550
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0147611145078175e-05,
+      "loss": 0.8869,
+      "step": 22560
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.014324395143681e-05,
+      "loss": 0.9871,
+      "step": 22570
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.013887675779544e-05,
+      "loss": 1.0331,
+      "step": 22580
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0134509564154075e-05,
+      "loss": 0.8722,
+      "step": 22590
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.013014237051271e-05,
+      "loss": 0.9293,
+      "step": 22600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.012577517687134e-05,
+      "loss": 0.9316,
+      "step": 22610
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.012140798322998e-05,
+      "loss": 0.8034,
+      "step": 22620
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0117040789588614e-05,
+      "loss": 0.8186,
+      "step": 22630
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.011267359594725e-05,
+      "loss": 0.8013,
+      "step": 22640
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.010830640230588e-05,
+      "loss": 0.9372,
+      "step": 22650
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0103939208664514e-05,
+      "loss": 0.9258,
+      "step": 22660
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.009957201502315e-05,
+      "loss": 0.9477,
+      "step": 22670
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.009520482138178e-05,
+      "loss": 0.8514,
+      "step": 22680
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.009083762774042e-05,
+      "loss": 0.9309,
+      "step": 22690
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.008647043409905e-05,
+      "loss": 0.8842,
+      "step": 22700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.008210324045769e-05,
+      "loss": 0.9121,
+      "step": 22710
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.007773604681631e-05,
+      "loss": 0.8357,
+      "step": 22720
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.007336885317495e-05,
+      "loss": 0.957,
+      "step": 22730
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0069001659533586e-05,
+      "loss": 0.9514,
+      "step": 22740
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.006463446589222e-05,
+      "loss": 0.9317,
+      "step": 22750
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.006026727225085e-05,
+      "loss": 1.0788,
+      "step": 22760
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0055900078609486e-05,
+      "loss": 0.7874,
+      "step": 22770
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0051532884968126e-05,
+      "loss": 0.8283,
+      "step": 22780
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.004716569132675e-05,
+      "loss": 0.8525,
+      "step": 22790
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.004279849768539e-05,
+      "loss": 0.8768,
+      "step": 22800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.003843130404402e-05,
+      "loss": 0.8875,
+      "step": 22810
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.003406411040266e-05,
+      "loss": 0.9899,
+      "step": 22820
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.002969691676129e-05,
+      "loss": 0.9096,
+      "step": 22830
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0025329723119925e-05,
+      "loss": 0.9522,
+      "step": 22840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002096252947856e-05,
+      "loss": 0.8351,
+      "step": 22850
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.001659533583719e-05,
+      "loss": 0.967,
+      "step": 22860
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.001222814219583e-05,
+      "loss": 0.9432,
+      "step": 22870
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.000786094855446e-05,
+      "loss": 0.8261,
+      "step": 22880
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.00034937549131e-05,
+      "loss": 0.9316,
+      "step": 22890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.9999126561271724e-05,
+      "loss": 0.914,
+      "step": 22900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.9994759367630364e-05,
+      "loss": 0.7597,
+      "step": 22910
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.999039217398899e-05,
+      "loss": 0.9356,
+      "step": 22920
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.998602498034763e-05,
+      "loss": 1.0288,
+      "step": 22930
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.9981657786706264e-05,
+      "loss": 0.9847,
+      "step": 22940
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.99772905930649e-05,
+      "loss": 0.9289,
+      "step": 22950
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.997292339942354e-05,
+      "loss": 0.8002,
+      "step": 22960
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9968556205782164e-05,
+      "loss": 0.8894,
+      "step": 22970
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9964189012140804e-05,
+      "loss": 0.8822,
+      "step": 22980
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.995982181849943e-05,
+      "loss": 1.0376,
+      "step": 22990
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.995545462485807e-05,
+      "loss": 0.7726,
+      "step": 23000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9951087431216697e-05,
+      "loss": 0.8633,
+      "step": 23010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9946720237575336e-05,
+      "loss": 0.9781,
+      "step": 23020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.994235304393397e-05,
+      "loss": 0.9212,
+      "step": 23030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.99379858502926e-05,
+      "loss": 0.8477,
+      "step": 23040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9933618656651236e-05,
+      "loss": 0.9468,
+      "step": 23050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.992925146300987e-05,
+      "loss": 0.8713,
+      "step": 23060
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.992488426936851e-05,
+      "loss": 0.9488,
+      "step": 23070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9920517075727136e-05,
+      "loss": 0.889,
+      "step": 23080
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9916149882085776e-05,
+      "loss": 0.9408,
+      "step": 23090
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.991178268844441e-05,
+      "loss": 0.8618,
+      "step": 23100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.990741549480304e-05,
+      "loss": 0.8909,
+      "step": 23110
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9903048301161675e-05,
+      "loss": 0.773,
+      "step": 23120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.989868110752031e-05,
+      "loss": 1.0007,
+      "step": 23130
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.989431391387894e-05,
+      "loss": 0.852,
+      "step": 23140
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9889946720237575e-05,
+      "loss": 0.9066,
+      "step": 23150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9885579526596215e-05,
+      "loss": 0.8275,
+      "step": 23160
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.988121233295484e-05,
+      "loss": 0.8664,
+      "step": 23170
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.987684513931348e-05,
+      "loss": 0.7777,
+      "step": 23180
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9872477945672114e-05,
+      "loss": 0.9574,
+      "step": 23190
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.986811075203075e-05,
+      "loss": 0.8532,
+      "step": 23200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.986374355838938e-05,
+      "loss": 0.8032,
+      "step": 23210
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9859376364748014e-05,
+      "loss": 0.985,
+      "step": 23220
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.985500917110665e-05,
+      "loss": 0.9106,
+      "step": 23230
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.985064197746528e-05,
+      "loss": 0.8986,
+      "step": 23240
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9846274783823914e-05,
+      "loss": 0.9993,
+      "step": 23250
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9841907590182554e-05,
+      "loss": 0.8737,
+      "step": 23260
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.983754039654119e-05,
+      "loss": 0.9188,
+      "step": 23270
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.983317320289982e-05,
+      "loss": 0.823,
+      "step": 23280
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.982880600925845e-05,
+      "loss": 0.9166,
+      "step": 23290
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9824438815617086e-05,
+      "loss": 0.897,
+      "step": 23300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982007162197572e-05,
+      "loss": 0.8725,
+      "step": 23310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.981570442833435e-05,
+      "loss": 0.9018,
+      "step": 23320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9811337234692986e-05,
+      "loss": 0.8738,
+      "step": 23330
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.980697004105162e-05,
+      "loss": 0.9419,
+      "step": 23340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.980260284741026e-05,
+      "loss": 0.8153,
+      "step": 23350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.979823565376889e-05,
+      "loss": 0.8514,
+      "step": 23360
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9793868460127526e-05,
+      "loss": 0.909,
+      "step": 23370
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.978950126648616e-05,
+      "loss": 0.998,
+      "step": 23380
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.978513407284479e-05,
+      "loss": 0.8643,
+      "step": 23390
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9780766879203425e-05,
+      "loss": 0.919,
+      "step": 23400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.977639968556206e-05,
+      "loss": 0.9264,
+      "step": 23410
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977203249192069e-05,
+      "loss": 0.88,
+      "step": 23420
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9767665298279325e-05,
+      "loss": 0.9223,
+      "step": 23430
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9763298104637965e-05,
+      "loss": 0.8655,
+      "step": 23440
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.975893091099659e-05,
+      "loss": 0.8948,
+      "step": 23450
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.975456371735523e-05,
+      "loss": 0.899,
+      "step": 23460
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9750196523713864e-05,
+      "loss": 0.8429,
+      "step": 23470
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.97458293300725e-05,
+      "loss": 0.7953,
+      "step": 23480
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.974146213643113e-05,
+      "loss": 0.8256,
+      "step": 23490
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9737094942789764e-05,
+      "loss": 0.7756,
+      "step": 23500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9732727749148404e-05,
+      "loss": 0.958,
+      "step": 23510
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.972836055550703e-05,
+      "loss": 0.93,
+      "step": 23520
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.972399336186567e-05,
+      "loss": 0.865,
+      "step": 23530
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.97196261682243e-05,
+      "loss": 0.8069,
+      "step": 23540
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971525897458294e-05,
+      "loss": 0.9169,
+      "step": 23550
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971089178094157e-05,
+      "loss": 0.8527,
+      "step": 23560
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.97065245873002e-05,
+      "loss": 0.9844,
+      "step": 23570
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9702157393658837e-05,
+      "loss": 0.8899,
+      "step": 23580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.969779020001747e-05,
+      "loss": 0.8415,
+      "step": 23590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.969342300637611e-05,
+      "loss": 0.8295,
+      "step": 23600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9689055812734736e-05,
+      "loss": 0.939,
+      "step": 23610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9684688619093376e-05,
+      "loss": 0.8631,
+      "step": 23620
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9680321425452e-05,
+      "loss": 0.8957,
+      "step": 23630
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.967595423181064e-05,
+      "loss": 0.8927,
+      "step": 23640
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.967158703816927e-05,
+      "loss": 0.8832,
+      "step": 23650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.966721984452791e-05,
+      "loss": 0.8359,
+      "step": 23660
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.966285265088654e-05,
+      "loss": 0.9299,
+      "step": 23670
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9658485457245175e-05,
+      "loss": 0.8928,
+      "step": 23680
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9654118263603815e-05,
+      "loss": 1.0467,
+      "step": 23690
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.964975106996244e-05,
+      "loss": 0.9758,
+      "step": 23700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.964538387632108e-05,
+      "loss": 0.9251,
+      "step": 23710
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.964101668267971e-05,
+      "loss": 0.8069,
+      "step": 23720
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.963664948903835e-05,
+      "loss": 0.9514,
+      "step": 23730
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9632282295396975e-05,
+      "loss": 0.8373,
+      "step": 23740
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9627915101755615e-05,
+      "loss": 0.8144,
+      "step": 23750
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962354790811425e-05,
+      "loss": 0.917,
+      "step": 23760
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.961918071447288e-05,
+      "loss": 0.9509,
+      "step": 23770
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9614813520831514e-05,
+      "loss": 0.8924,
+      "step": 23780
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.961044632719015e-05,
+      "loss": 0.9203,
+      "step": 23790
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.960607913354879e-05,
+      "loss": 0.9308,
+      "step": 23800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9601711939907414e-05,
+      "loss": 0.8257,
+      "step": 23810
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9597344746266054e-05,
+      "loss": 0.9576,
+      "step": 23820
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.959297755262468e-05,
+      "loss": 0.938,
+      "step": 23830
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.958861035898332e-05,
+      "loss": 0.9299,
+      "step": 23840
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.958424316534195e-05,
+      "loss": 0.8273,
+      "step": 23850
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9579875971700587e-05,
+      "loss": 1.036,
+      "step": 23860
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.957550877805922e-05,
+      "loss": 0.8539,
+      "step": 23870
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957114158441785e-05,
+      "loss": 0.8875,
+      "step": 23880
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.956677439077649e-05,
+      "loss": 0.879,
+      "step": 23890
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.956240719713512e-05,
+      "loss": 0.9686,
+      "step": 23900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.955804000349376e-05,
+      "loss": 0.955,
+      "step": 23910
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.955367280985239e-05,
+      "loss": 0.9249,
+      "step": 23920
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9549305616211026e-05,
+      "loss": 0.905,
+      "step": 23930
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.954493842256966e-05,
+      "loss": 0.9022,
+      "step": 23940
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.954057122892829e-05,
+      "loss": 0.9259,
+      "step": 23950
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9536204035286925e-05,
+      "loss": 0.9778,
+      "step": 23960
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.953183684164556e-05,
+      "loss": 0.7921,
+      "step": 23970
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.952746964800419e-05,
+      "loss": 0.8971,
+      "step": 23980
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9523102454362825e-05,
+      "loss": 0.9093,
+      "step": 23990
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9518735260721465e-05,
+      "loss": 0.9566,
+      "step": 24000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.95143680670801e-05,
+      "loss": 0.8507,
+      "step": 24010
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.951000087343873e-05,
+      "loss": 0.7779,
+      "step": 24020
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9505633679797365e-05,
+      "loss": 0.8863,
+      "step": 24030
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9501266486156e-05,
+      "loss": 0.931,
+      "step": 24040
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.949689929251463e-05,
+      "loss": 0.9493,
+      "step": 24050
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9492532098873264e-05,
+      "loss": 0.8711,
+      "step": 24060
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.94881649052319e-05,
+      "loss": 1.0326,
+      "step": 24070
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.948379771159053e-05,
+      "loss": 0.8192,
+      "step": 24080
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.947943051794917e-05,
+      "loss": 0.888,
+      "step": 24090
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9475063324307804e-05,
+      "loss": 0.8611,
+      "step": 24100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.947069613066644e-05,
+      "loss": 0.9298,
+      "step": 24110
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.946632893702507e-05,
+      "loss": 0.9856,
+      "step": 24120
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.94619617433837e-05,
+      "loss": 0.8074,
+      "step": 24130
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9457594549742337e-05,
+      "loss": 0.8909,
+      "step": 24140
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.945322735610097e-05,
+      "loss": 0.9486,
+      "step": 24150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.94488601624596e-05,
+      "loss": 0.8203,
+      "step": 24160
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.944449296881824e-05,
+      "loss": 0.7625,
+      "step": 24170
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.944012577517687e-05,
+      "loss": 0.8786,
+      "step": 24180
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.943575858153551e-05,
+      "loss": 0.9183,
+      "step": 24190
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.943139138789414e-05,
+      "loss": 0.8982,
+      "step": 24200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9427024194252776e-05,
+      "loss": 0.8998,
+      "step": 24210
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.942265700061141e-05,
+      "loss": 1.0057,
+      "step": 24220
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.941828980697004e-05,
+      "loss": 0.9163,
+      "step": 24230
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9413922613328675e-05,
+      "loss": 0.9484,
+      "step": 24240
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.940955541968731e-05,
+      "loss": 0.838,
+      "step": 24250
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.940518822604595e-05,
+      "loss": 0.8994,
+      "step": 24260
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9400821032404575e-05,
+      "loss": 0.8689,
+      "step": 24270
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9396453838763215e-05,
+      "loss": 0.9182,
+      "step": 24280
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.939208664512185e-05,
+      "loss": 0.9549,
+      "step": 24290
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.938771945148048e-05,
+      "loss": 0.8033,
+      "step": 24300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9383352257839115e-05,
+      "loss": 0.7709,
+      "step": 24310
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.937898506419775e-05,
+      "loss": 0.8909,
+      "step": 24320
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937461787055639e-05,
+      "loss": 0.8917,
+      "step": 24330
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9370250676915014e-05,
+      "loss": 0.8926,
+      "step": 24340
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9365883483273654e-05,
+      "loss": 0.8432,
+      "step": 24350
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.936151628963228e-05,
+      "loss": 0.8272,
+      "step": 24360
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.935714909599092e-05,
+      "loss": 0.9043,
+      "step": 24370
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.935278190234955e-05,
+      "loss": 0.7532,
+      "step": 24380
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.934841470870819e-05,
+      "loss": 0.8199,
+      "step": 24390
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.934404751506682e-05,
+      "loss": 1.0203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.933968032142545e-05,
+      "loss": 0.8517,
+      "step": 24410
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.933531312778409e-05,
+      "loss": 1.021,
+      "step": 24420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.933094593414272e-05,
+      "loss": 0.8812,
+      "step": 24430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.932657874050136e-05,
+      "loss": 0.9152,
+      "step": 24440
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9322211546859986e-05,
+      "loss": 0.9862,
+      "step": 24450
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9317844353218626e-05,
+      "loss": 1.0007,
+      "step": 24460
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.931347715957725e-05,
+      "loss": 0.9025,
+      "step": 24470
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.930910996593589e-05,
+      "loss": 0.8218,
+      "step": 24480
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9304742772294526e-05,
+      "loss": 1.0014,
+      "step": 24490
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.930037557865316e-05,
+      "loss": 0.9155,
+      "step": 24500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.929600838501179e-05,
+      "loss": 0.9372,
+      "step": 24510
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9291641191370425e-05,
+      "loss": 0.8334,
+      "step": 24520
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9287273997729065e-05,
+      "loss": 0.8021,
+      "step": 24530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.928290680408769e-05,
+      "loss": 0.96,
+      "step": 24540
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.927853961044633e-05,
+      "loss": 0.8882,
+      "step": 24550
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927417241680496e-05,
+      "loss": 0.7146,
+      "step": 24560
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92698052231636e-05,
+      "loss": 0.9438,
+      "step": 24570
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926543802952223e-05,
+      "loss": 0.8961,
+      "step": 24580
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9261070835880865e-05,
+      "loss": 0.8391,
+      "step": 24590
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92567036422395e-05,
+      "loss": 0.8944,
+      "step": 24600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.925233644859813e-05,
+      "loss": 0.832,
+      "step": 24610
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.924796925495677e-05,
+      "loss": 0.9138,
+      "step": 24620
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92436020613154e-05,
+      "loss": 0.8235,
+      "step": 24630
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.923923486767404e-05,
+      "loss": 0.848,
+      "step": 24640
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9234867674032664e-05,
+      "loss": 0.8115,
+      "step": 24650
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9230500480391304e-05,
+      "loss": 1.0193,
+      "step": 24660
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.922613328674994e-05,
+      "loss": 0.9207,
+      "step": 24670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.922176609310857e-05,
+      "loss": 0.9943,
+      "step": 24680
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.92173988994672e-05,
+      "loss": 0.9637,
+      "step": 24690
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9213031705825837e-05,
+      "loss": 0.8413,
+      "step": 24700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.920866451218447e-05,
+      "loss": 0.9529,
+      "step": 24710
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.92042973185431e-05,
+      "loss": 1.0203,
+      "step": 24720
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.919993012490174e-05,
+      "loss": 0.8928,
+      "step": 24730
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9195562931260376e-05,
+      "loss": 0.8102,
+      "step": 24740
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.919119573761901e-05,
+      "loss": 0.8577,
+      "step": 24750
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.918682854397764e-05,
+      "loss": 0.8211,
+      "step": 24760
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9182461350336276e-05,
+      "loss": 0.8001,
+      "step": 24770
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.917809415669491e-05,
+      "loss": 1.0066,
+      "step": 24780
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917372696305354e-05,
+      "loss": 0.8596,
+      "step": 24790
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9169359769412175e-05,
+      "loss": 0.9382,
+      "step": 24800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.916499257577081e-05,
+      "loss": 0.872,
+      "step": 24810
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.916062538212945e-05,
+      "loss": 0.7562,
+      "step": 24820
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.915625818848808e-05,
+      "loss": 0.7974,
+      "step": 24830
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9151890994846715e-05,
+      "loss": 0.8388,
+      "step": 24840
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.914752380120535e-05,
+      "loss": 0.8599,
+      "step": 24850
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.914315660756398e-05,
+      "loss": 0.8674,
+      "step": 24860
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9138789413922615e-05,
+      "loss": 0.8596,
+      "step": 24870
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.913442222028125e-05,
+      "loss": 0.9782,
+      "step": 24880
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.913005502663988e-05,
+      "loss": 0.8244,
+      "step": 24890
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9125687832998514e-05,
+      "loss": 0.903,
+      "step": 24900
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912132063935715e-05,
+      "loss": 0.9247,
+      "step": 24910
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.911695344571579e-05,
+      "loss": 0.8638,
+      "step": 24920
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.911258625207442e-05,
+      "loss": 0.8429,
+      "step": 24930
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9108219058433054e-05,
+      "loss": 0.8551,
+      "step": 24940
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.910385186479169e-05,
+      "loss": 0.8449,
+      "step": 24950
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.909948467115032e-05,
+      "loss": 0.7967,
+      "step": 24960
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.909511747750895e-05,
+      "loss": 0.9304,
+      "step": 24970
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9090750283867587e-05,
+      "loss": 0.9562,
+      "step": 24980
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9086383090226227e-05,
+      "loss": 0.9337,
+      "step": 24990
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.908201589658485e-05,
+      "loss": 1.0566,
+      "step": 25000
+    },
+    {
+      "epoch": 2.18,
+      "eval_accuracy": 0.5722708067210376,
+      "eval_loss": 0.893390417098999,
+      "eval_runtime": 84.0906,
+      "eval_samples_per_second": 121.024,
+      "eval_steps_per_second": 15.138,
+      "step": 25000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.907764870294349e-05,
+      "loss": 0.9265,
+      "step": 25010
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9073281509302126e-05,
+      "loss": 0.854,
+      "step": 25020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.906891431566076e-05,
+      "loss": 0.8346,
+      "step": 25030
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.906454712201939e-05,
+      "loss": 0.9734,
+      "step": 25040
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9060179928378026e-05,
+      "loss": 0.8484,
+      "step": 25050
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.905581273473666e-05,
+      "loss": 0.8578,
+      "step": 25060
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.905144554109529e-05,
+      "loss": 0.9587,
+      "step": 25070
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.904707834745393e-05,
+      "loss": 0.8706,
+      "step": 25080
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.904271115381256e-05,
+      "loss": 0.8656,
+      "step": 25090
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.90383439601712e-05,
+      "loss": 0.9157,
+      "step": 25100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.903397676652983e-05,
+      "loss": 0.8974,
+      "step": 25110
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9029609572888465e-05,
+      "loss": 0.9538,
+      "step": 25120
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.90252423792471e-05,
+      "loss": 0.8911,
+      "step": 25130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.902087518560573e-05,
+      "loss": 0.9247,
+      "step": 25140
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.901650799196437e-05,
+      "loss": 0.9344,
+      "step": 25150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9012140798323e-05,
+      "loss": 0.8821,
+      "step": 25160
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.900777360468164e-05,
+      "loss": 0.8694,
+      "step": 25170
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9003406411040264e-05,
+      "loss": 0.7757,
+      "step": 25180
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.8999039217398904e-05,
+      "loss": 0.9059,
+      "step": 25190
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.899467202375753e-05,
+      "loss": 0.9391,
+      "step": 25200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.899030483011617e-05,
+      "loss": 0.9832,
+      "step": 25210
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.8985937636474804e-05,
+      "loss": 0.876,
+      "step": 25220
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.898157044283344e-05,
+      "loss": 0.907,
+      "step": 25230
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.897720324919207e-05,
+      "loss": 0.8353,
+      "step": 25240
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.89728360555507e-05,
+      "loss": 0.9534,
+      "step": 25250
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.896846886190934e-05,
+      "loss": 0.7413,
+      "step": 25260
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.896410166826797e-05,
+      "loss": 0.9761,
+      "step": 25270
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.895973447462661e-05,
+      "loss": 0.9094,
+      "step": 25280
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8955367280985236e-05,
+      "loss": 0.8208,
+      "step": 25290
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8951000087343876e-05,
+      "loss": 0.879,
+      "step": 25300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.894663289370251e-05,
+      "loss": 0.8391,
+      "step": 25310
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.894226570006114e-05,
+      "loss": 0.9105,
+      "step": 25320
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8937898506419776e-05,
+      "loss": 1.027,
+      "step": 25330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.893353131277841e-05,
+      "loss": 0.8306,
+      "step": 25340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.892916411913705e-05,
+      "loss": 0.9332,
+      "step": 25350
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8924796925495675e-05,
+      "loss": 0.8374,
+      "step": 25360
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8920429731854315e-05,
+      "loss": 0.8523,
+      "step": 25370
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.891606253821294e-05,
+      "loss": 0.9511,
+      "step": 25380
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.891169534457158e-05,
+      "loss": 0.7786,
+      "step": 25390
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8907328150930215e-05,
+      "loss": 0.9563,
+      "step": 25400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.890296095728885e-05,
+      "loss": 0.9415,
+      "step": 25410
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.889859376364748e-05,
+      "loss": 0.865,
+      "step": 25420
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8894226570006115e-05,
+      "loss": 0.927,
+      "step": 25430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8889859376364755e-05,
+      "loss": 0.9167,
+      "step": 25440
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.888549218272338e-05,
+      "loss": 0.7889,
+      "step": 25450
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.888112498908202e-05,
+      "loss": 0.8907,
+      "step": 25460
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.887675779544065e-05,
+      "loss": 0.9217,
+      "step": 25470
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.887239060179929e-05,
+      "loss": 0.9181,
+      "step": 25480
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886802340815792e-05,
+      "loss": 0.9643,
+      "step": 25490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8863656214516554e-05,
+      "loss": 0.9515,
+      "step": 25500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.885928902087519e-05,
+      "loss": 0.8572,
+      "step": 25510
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.885492182723382e-05,
+      "loss": 0.9012,
+      "step": 25520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.885055463359245e-05,
+      "loss": 0.7615,
+      "step": 25530
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8846187439951087e-05,
+      "loss": 0.7784,
+      "step": 25540
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8841820246309727e-05,
+      "loss": 0.9518,
+      "step": 25550
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.883745305266835e-05,
+      "loss": 0.9307,
+      "step": 25560
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.883308585902699e-05,
+      "loss": 0.9266,
+      "step": 25570
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8828718665385626e-05,
+      "loss": 1.0072,
+      "step": 25580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.882435147174426e-05,
+      "loss": 0.8514,
+      "step": 25590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881998427810289e-05,
+      "loss": 0.8797,
+      "step": 25600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8815617084461526e-05,
+      "loss": 0.9151,
+      "step": 25610
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881124989082016e-05,
+      "loss": 0.7862,
+      "step": 25620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.880688269717879e-05,
+      "loss": 1.0018,
+      "step": 25630
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.880251550353743e-05,
+      "loss": 0.8088,
+      "step": 25640
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8798148309896065e-05,
+      "loss": 0.9481,
+      "step": 25650
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.87937811162547e-05,
+      "loss": 0.854,
+      "step": 25660
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.878941392261333e-05,
+      "loss": 0.8865,
+      "step": 25670
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8785046728971965e-05,
+      "loss": 0.8719,
+      "step": 25680
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.87806795353306e-05,
+      "loss": 0.783,
+      "step": 25690
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.877631234168923e-05,
+      "loss": 0.9409,
+      "step": 25700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8771945148047865e-05,
+      "loss": 0.9009,
+      "step": 25710
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.87675779544065e-05,
+      "loss": 0.9481,
+      "step": 25720
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.876321076076513e-05,
+      "loss": 0.783,
+      "step": 25730
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.875884356712377e-05,
+      "loss": 0.8429,
+      "step": 25740
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8754476373482404e-05,
+      "loss": 0.9001,
+      "step": 25750
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.875010917984104e-05,
+      "loss": 0.8734,
+      "step": 25760
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.874574198619967e-05,
+      "loss": 0.9672,
+      "step": 25770
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8741374792558304e-05,
+      "loss": 0.8796,
+      "step": 25780
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.873700759891694e-05,
+      "loss": 0.8603,
+      "step": 25790
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.873264040527557e-05,
+      "loss": 0.8875,
+      "step": 25800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.872827321163421e-05,
+      "loss": 0.9671,
+      "step": 25810
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8723906017992837e-05,
+      "loss": 0.9109,
+      "step": 25820
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8719538824351477e-05,
+      "loss": 0.8276,
+      "step": 25830
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.871517163071011e-05,
+      "loss": 1.024,
+      "step": 25840
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.871080443706874e-05,
+      "loss": 0.892,
+      "step": 25850
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8706437243427376e-05,
+      "loss": 0.917,
+      "step": 25860
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.870207004978601e-05,
+      "loss": 0.8997,
+      "step": 25870
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.869770285614464e-05,
+      "loss": 0.8367,
+      "step": 25880
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8693335662503276e-05,
+      "loss": 0.8792,
+      "step": 25890
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8688968468861916e-05,
+      "loss": 0.9476,
+      "step": 25900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.868460127522054e-05,
+      "loss": 0.9277,
+      "step": 25910
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.868023408157918e-05,
+      "loss": 0.8288,
+      "step": 25920
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.867586688793781e-05,
+      "loss": 0.9404,
+      "step": 25930
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.867149969429645e-05,
+      "loss": 0.9071,
+      "step": 25940
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.866713250065508e-05,
+      "loss": 0.856,
+      "step": 25950
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8662765307013715e-05,
+      "loss": 0.8676,
+      "step": 25960
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.865839811337235e-05,
+      "loss": 0.8512,
+      "step": 25970
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.865403091973098e-05,
+      "loss": 0.8674,
+      "step": 25980
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.864966372608962e-05,
+      "loss": 0.9658,
+      "step": 25990
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.864529653244825e-05,
+      "loss": 0.9211,
+      "step": 26000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.864092933880689e-05,
+      "loss": 0.9245,
+      "step": 26010
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8636562145165514e-05,
+      "loss": 0.8983,
+      "step": 26020
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8632194951524154e-05,
+      "loss": 0.8051,
+      "step": 26030
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.862782775788279e-05,
+      "loss": 0.8324,
+      "step": 26040
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.862346056424142e-05,
+      "loss": 0.9257,
+      "step": 26050
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8619093370600054e-05,
+      "loss": 0.9909,
+      "step": 26060
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.861472617695869e-05,
+      "loss": 0.8704,
+      "step": 26070
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.861035898331733e-05,
+      "loss": 0.7843,
+      "step": 26080
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8605991789675953e-05,
+      "loss": 0.9616,
+      "step": 26090
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8601624596034593e-05,
+      "loss": 0.798,
+      "step": 26100
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.859725740239322e-05,
+      "loss": 0.9102,
+      "step": 26110
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.859289020875186e-05,
+      "loss": 0.9212,
+      "step": 26120
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8588523015110486e-05,
+      "loss": 0.8784,
+      "step": 26130
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8584155821469126e-05,
+      "loss": 0.8685,
+      "step": 26140
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.857978862782776e-05,
+      "loss": 0.9278,
+      "step": 26150
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.857542143418639e-05,
+      "loss": 0.836,
+      "step": 26160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.857105424054503e-05,
+      "loss": 0.9116,
+      "step": 26170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.856668704690366e-05,
+      "loss": 0.8643,
+      "step": 26180
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.85623198532623e-05,
+      "loss": 1.0034,
+      "step": 26190
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8557952659620925e-05,
+      "loss": 0.9029,
+      "step": 26200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8553585465979565e-05,
+      "loss": 0.9276,
+      "step": 26210
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.85492182723382e-05,
+      "loss": 0.8415,
+      "step": 26220
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.854485107869683e-05,
+      "loss": 0.9597,
+      "step": 26230
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8540483885055465e-05,
+      "loss": 0.8595,
+      "step": 26240
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.85361166914141e-05,
+      "loss": 0.8665,
+      "step": 26250
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.853174949777273e-05,
+      "loss": 0.8617,
+      "step": 26260
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8527382304131365e-05,
+      "loss": 0.8992,
+      "step": 26270
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8523015110490005e-05,
+      "loss": 0.8361,
+      "step": 26280
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.851864791684863e-05,
+      "loss": 0.8541,
+      "step": 26290
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.851428072320727e-05,
+      "loss": 0.9196,
+      "step": 26300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8509913529565904e-05,
+      "loss": 0.9852,
+      "step": 26310
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.850554633592454e-05,
+      "loss": 0.8888,
+      "step": 26320
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.850117914228317e-05,
+      "loss": 1.0016,
+      "step": 26330
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8496811948641804e-05,
+      "loss": 0.9295,
+      "step": 26340
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.849244475500044e-05,
+      "loss": 0.9878,
+      "step": 26350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.848807756135907e-05,
+      "loss": 0.9318,
+      "step": 26360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.848371036771771e-05,
+      "loss": 0.8953,
+      "step": 26370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.847934317407634e-05,
+      "loss": 0.9807,
+      "step": 26380
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8474975980434977e-05,
+      "loss": 0.867,
+      "step": 26390
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.847060878679361e-05,
+      "loss": 0.9032,
+      "step": 26400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.846624159315224e-05,
+      "loss": 0.7552,
+      "step": 26410
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8461874399510876e-05,
+      "loss": 0.9564,
+      "step": 26420
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.845750720586951e-05,
+      "loss": 0.9544,
+      "step": 26430
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.845314001222814e-05,
+      "loss": 0.8537,
+      "step": 26440
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8448772818586776e-05,
+      "loss": 0.7866,
+      "step": 26450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.844440562494541e-05,
+      "loss": 0.7853,
+      "step": 26460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.844003843130405e-05,
+      "loss": 0.8332,
+      "step": 26470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.843567123766268e-05,
+      "loss": 0.7433,
+      "step": 26480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8431304044021315e-05,
+      "loss": 0.873,
+      "step": 26490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.842693685037995e-05,
+      "loss": 0.907,
+      "step": 26500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.842256965673858e-05,
+      "loss": 0.8519,
+      "step": 26510
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8418202463097215e-05,
+      "loss": 0.9365,
+      "step": 26520
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.841383526945585e-05,
+      "loss": 0.9625,
+      "step": 26530
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.840946807581448e-05,
+      "loss": 0.9123,
+      "step": 26540
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8405100882173115e-05,
+      "loss": 0.883,
+      "step": 26550
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8400733688531755e-05,
+      "loss": 1.0301,
+      "step": 26560
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.839636649489039e-05,
+      "loss": 0.8476,
+      "step": 26570
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.839199930124902e-05,
+      "loss": 0.8638,
+      "step": 26580
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8387632107607654e-05,
+      "loss": 0.8378,
+      "step": 26590
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.838326491396629e-05,
+      "loss": 0.8649,
+      "step": 26600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.837889772032492e-05,
+      "loss": 0.869,
+      "step": 26610
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8374530526683554e-05,
+      "loss": 0.9053,
+      "step": 26620
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8370163333042194e-05,
+      "loss": 0.9383,
+      "step": 26630
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.836579613940082e-05,
+      "loss": 0.7857,
+      "step": 26640
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.836142894575946e-05,
+      "loss": 0.8657,
+      "step": 26650
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.835706175211809e-05,
+      "loss": 0.9213,
+      "step": 26660
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.835269455847673e-05,
+      "loss": 0.9264,
+      "step": 26670
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.834832736483536e-05,
+      "loss": 1.0045,
+      "step": 26680
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.834396017119399e-05,
+      "loss": 0.9877,
+      "step": 26690
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8339592977552626e-05,
+      "loss": 0.9406,
+      "step": 26700
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.833522578391126e-05,
+      "loss": 0.7996,
+      "step": 26710
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.83308585902699e-05,
+      "loss": 0.9746,
+      "step": 26720
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8326491396628526e-05,
+      "loss": 0.8804,
+      "step": 26730
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8322124202987166e-05,
+      "loss": 0.8101,
+      "step": 26740
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.831775700934579e-05,
+      "loss": 0.9151,
+      "step": 26750
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.831338981570443e-05,
+      "loss": 0.9393,
+      "step": 26760
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8309022622063065e-05,
+      "loss": 0.9434,
+      "step": 26770
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.83046554284217e-05,
+      "loss": 0.9038,
+      "step": 26780
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.830028823478033e-05,
+      "loss": 0.939,
+      "step": 26790
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8295921041138965e-05,
+      "loss": 0.7909,
+      "step": 26800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8291553847497605e-05,
+      "loss": 0.8198,
+      "step": 26810
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.828718665385623e-05,
+      "loss": 0.9151,
+      "step": 26820
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.828281946021487e-05,
+      "loss": 0.8811,
+      "step": 26830
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.82784522665735e-05,
+      "loss": 0.9721,
+      "step": 26840
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.827408507293214e-05,
+      "loss": 0.824,
+      "step": 26850
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8269717879290764e-05,
+      "loss": 0.9274,
+      "step": 26860
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8265350685649404e-05,
+      "loss": 0.9757,
+      "step": 26870
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.826098349200804e-05,
+      "loss": 0.9253,
+      "step": 26880
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.825661629836667e-05,
+      "loss": 0.8558,
+      "step": 26890
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.825224910472531e-05,
+      "loss": 0.9289,
+      "step": 26900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.824788191108394e-05,
+      "loss": 0.9817,
+      "step": 26910
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.824351471744258e-05,
+      "loss": 0.8305,
+      "step": 26920
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8239147523801203e-05,
+      "loss": 0.8045,
+      "step": 26930
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8234780330159843e-05,
+      "loss": 0.8044,
+      "step": 26940
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.823041313651847e-05,
+      "loss": 0.874,
+      "step": 26950
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.822604594287711e-05,
+      "loss": 0.8977,
+      "step": 26960
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.822167874923574e-05,
+      "loss": 0.8725,
+      "step": 26970
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8217311555594376e-05,
+      "loss": 1.0004,
+      "step": 26980
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821294436195301e-05,
+      "loss": 0.9293,
+      "step": 26990
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.820857716831164e-05,
+      "loss": 0.7762,
+      "step": 27000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.820420997467028e-05,
+      "loss": 0.8381,
+      "step": 27010
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.819984278102891e-05,
+      "loss": 0.9129,
+      "step": 27020
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.819547558738755e-05,
+      "loss": 0.9258,
+      "step": 27030
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8191108393746175e-05,
+      "loss": 0.9834,
+      "step": 27040
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8186741200104815e-05,
+      "loss": 0.9089,
+      "step": 27050
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.818237400646345e-05,
+      "loss": 0.9071,
+      "step": 27060
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.817800681282208e-05,
+      "loss": 0.8775,
+      "step": 27070
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8173639619180715e-05,
+      "loss": 0.8611,
+      "step": 27080
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816927242553935e-05,
+      "loss": 0.9264,
+      "step": 27090
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816490523189799e-05,
+      "loss": 0.8804,
+      "step": 27100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8160538038256615e-05,
+      "loss": 0.8369,
+      "step": 27110
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8156170844615255e-05,
+      "loss": 0.8099,
+      "step": 27120
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.815180365097389e-05,
+      "loss": 0.8842,
+      "step": 27130
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.814743645733252e-05,
+      "loss": 0.908,
+      "step": 27140
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8143069263691154e-05,
+      "loss": 0.9624,
+      "step": 27150
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.813870207004979e-05,
+      "loss": 0.8502,
+      "step": 27160
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.813433487640842e-05,
+      "loss": 0.8926,
+      "step": 27170
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8129967682767054e-05,
+      "loss": 0.9042,
+      "step": 27180
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.812560048912569e-05,
+      "loss": 0.8802,
+      "step": 27190
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.812123329548432e-05,
+      "loss": 0.9184,
+      "step": 27200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.811686610184296e-05,
+      "loss": 0.9333,
+      "step": 27210
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8112498908201593e-05,
+      "loss": 0.9385,
+      "step": 27220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.810813171456023e-05,
+      "loss": 0.8899,
+      "step": 27230
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.810376452091886e-05,
+      "loss": 0.9678,
+      "step": 27240
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.809939732727749e-05,
+      "loss": 0.852,
+      "step": 27250
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8095030133636126e-05,
+      "loss": 0.841,
+      "step": 27260
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.809066293999476e-05,
+      "loss": 0.9297,
+      "step": 27270
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.808629574635339e-05,
+      "loss": 0.9052,
+      "step": 27280
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.808192855271203e-05,
+      "loss": 0.9502,
+      "step": 27290
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8077561359070666e-05,
+      "loss": 0.7993,
+      "step": 27300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.80731941654293e-05,
+      "loss": 0.8459,
+      "step": 27310
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.806882697178793e-05,
+      "loss": 0.9762,
+      "step": 27320
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8064459778146565e-05,
+      "loss": 0.9187,
+      "step": 27330
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.80600925845052e-05,
+      "loss": 0.9693,
+      "step": 27340
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.805572539086383e-05,
+      "loss": 0.8865,
+      "step": 27350
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8051358197222465e-05,
+      "loss": 0.8245,
+      "step": 27360
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.80469910035811e-05,
+      "loss": 0.8544,
+      "step": 27370
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.804262380993974e-05,
+      "loss": 0.9236,
+      "step": 27380
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8038256616298365e-05,
+      "loss": 0.9529,
+      "step": 27390
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8033889422657005e-05,
+      "loss": 0.9051,
+      "step": 27400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.802952222901564e-05,
+      "loss": 0.8557,
+      "step": 27410
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.802515503537427e-05,
+      "loss": 0.9562,
+      "step": 27420
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8020787841732904e-05,
+      "loss": 0.9046,
+      "step": 27430
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.801642064809154e-05,
+      "loss": 0.9238,
+      "step": 27440
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.801205345445017e-05,
+      "loss": 0.9644,
+      "step": 27450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8007686260808804e-05,
+      "loss": 0.954,
+      "step": 27460
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8003319067167444e-05,
+      "loss": 0.8797,
+      "step": 27470
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.799895187352607e-05,
+      "loss": 0.9005,
+      "step": 27480
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.799458467988471e-05,
+      "loss": 0.8653,
+      "step": 27490
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.7990217486243343e-05,
+      "loss": 0.8788,
+      "step": 27500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.798585029260198e-05,
+      "loss": 0.9159,
+      "step": 27510
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.798148309896061e-05,
+      "loss": 0.881,
+      "step": 27520
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.797711590531924e-05,
+      "loss": 0.9706,
+      "step": 27530
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.797274871167788e-05,
+      "loss": 0.9096,
+      "step": 27540
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.796838151803651e-05,
+      "loss": 0.8772,
+      "step": 27550
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.796401432439515e-05,
+      "loss": 0.8083,
+      "step": 27560
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7959647130753776e-05,
+      "loss": 0.9359,
+      "step": 27570
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7955279937112416e-05,
+      "loss": 0.8518,
+      "step": 27580
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.795091274347104e-05,
+      "loss": 0.9001,
+      "step": 27590
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.794654554982968e-05,
+      "loss": 0.8307,
+      "step": 27600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7942178356188315e-05,
+      "loss": 0.8333,
+      "step": 27610
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.793781116254695e-05,
+      "loss": 0.8095,
+      "step": 27620
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.793344396890559e-05,
+      "loss": 0.9528,
+      "step": 27630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7929076775264215e-05,
+      "loss": 0.9253,
+      "step": 27640
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7924709581622855e-05,
+      "loss": 0.8031,
+      "step": 27650
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.792034238798148e-05,
+      "loss": 1.0076,
+      "step": 27660
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.791597519434012e-05,
+      "loss": 0.9623,
+      "step": 27670
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.791160800069875e-05,
+      "loss": 0.8274,
+      "step": 27680
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.790724080705739e-05,
+      "loss": 0.9221,
+      "step": 27690
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.790287361341602e-05,
+      "loss": 0.8816,
+      "step": 27700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7898506419774654e-05,
+      "loss": 0.9172,
+      "step": 27710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.789413922613329e-05,
+      "loss": 0.8602,
+      "step": 27720
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.788977203249192e-05,
+      "loss": 0.8672,
+      "step": 27730
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.788540483885056e-05,
+      "loss": 0.8402,
+      "step": 27740
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.788103764520919e-05,
+      "loss": 0.9284,
+      "step": 27750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.787667045156783e-05,
+      "loss": 0.9028,
+      "step": 27760
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7872303257926453e-05,
+      "loss": 0.9364,
+      "step": 27770
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7867936064285093e-05,
+      "loss": 0.8405,
+      "step": 27780
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.786356887064373e-05,
+      "loss": 0.8595,
+      "step": 27790
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.785920167700236e-05,
+      "loss": 0.944,
+      "step": 27800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.785483448336099e-05,
+      "loss": 0.9186,
+      "step": 27810
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7850467289719626e-05,
+      "loss": 0.8419,
+      "step": 27820
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7846100096078266e-05,
+      "loss": 0.9227,
+      "step": 27830
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.784173290243689e-05,
+      "loss": 0.9303,
+      "step": 27840
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.783736570879553e-05,
+      "loss": 0.78,
+      "step": 27850
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.783299851515416e-05,
+      "loss": 0.8949,
+      "step": 27860
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.78286313215128e-05,
+      "loss": 0.76,
+      "step": 27870
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.782426412787143e-05,
+      "loss": 0.8809,
+      "step": 27880
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7819896934230065e-05,
+      "loss": 0.8449,
+      "step": 27890
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.78155297405887e-05,
+      "loss": 1.0643,
+      "step": 27900
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.781116254694733e-05,
+      "loss": 1.0307,
+      "step": 27910
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7806795353305965e-05,
+      "loss": 0.8784,
+      "step": 27920
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.78024281596646e-05,
+      "loss": 0.8797,
+      "step": 27930
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.779806096602324e-05,
+      "loss": 0.8203,
+      "step": 27940
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.779369377238187e-05,
+      "loss": 0.8096,
+      "step": 27950
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7789326578740505e-05,
+      "loss": 0.8641,
+      "step": 27960
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.778495938509914e-05,
+      "loss": 0.832,
+      "step": 27970
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.778059219145777e-05,
+      "loss": 0.9069,
+      "step": 27980
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7776224997816404e-05,
+      "loss": 0.8991,
+      "step": 27990
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.777185780417504e-05,
+      "loss": 0.9697,
+      "step": 28000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.776749061053367e-05,
+      "loss": 0.9659,
+      "step": 28010
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7763123416892304e-05,
+      "loss": 0.8802,
+      "step": 28020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7758756223250944e-05,
+      "loss": 0.8412,
+      "step": 28030
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.775438902960958e-05,
+      "loss": 0.8759,
+      "step": 28040
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.775002183596821e-05,
+      "loss": 1.0513,
+      "step": 28050
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7745654642326843e-05,
+      "loss": 0.8851,
+      "step": 28060
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.774128744868548e-05,
+      "loss": 0.965,
+      "step": 28070
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.773692025504411e-05,
+      "loss": 0.8704,
+      "step": 28080
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.773255306140274e-05,
+      "loss": 0.85,
+      "step": 28090
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7728185867761376e-05,
+      "loss": 0.8991,
+      "step": 28100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7723818674120016e-05,
+      "loss": 1.0349,
+      "step": 28110
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.771945148047864e-05,
+      "loss": 0.8779,
+      "step": 28120
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.771508428683728e-05,
+      "loss": 0.9126,
+      "step": 28130
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7710717093195916e-05,
+      "loss": 0.9072,
+      "step": 28140
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.770634989955455e-05,
+      "loss": 0.8645,
+      "step": 28150
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.770198270591318e-05,
+      "loss": 0.8625,
+      "step": 28160
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7697615512271815e-05,
+      "loss": 1.0313,
+      "step": 28170
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.769324831863045e-05,
+      "loss": 0.8339,
+      "step": 28180
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.768888112498908e-05,
+      "loss": 0.9292,
+      "step": 28190
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.768451393134772e-05,
+      "loss": 0.9588,
+      "step": 28200
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.768014673770635e-05,
+      "loss": 0.8291,
+      "step": 28210
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.767577954406499e-05,
+      "loss": 0.8975,
+      "step": 28220
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.767141235042362e-05,
+      "loss": 0.95,
+      "step": 28230
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7667045156782255e-05,
+      "loss": 0.8907,
+      "step": 28240
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.766267796314089e-05,
+      "loss": 0.9476,
+      "step": 28250
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.765831076949952e-05,
+      "loss": 0.8179,
+      "step": 28260
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7653943575858154e-05,
+      "loss": 0.8824,
+      "step": 28270
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.764957638221679e-05,
+      "loss": 0.9275,
+      "step": 28280
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.764520918857543e-05,
+      "loss": 0.8864,
+      "step": 28290
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7640841994934054e-05,
+      "loss": 0.8628,
+      "step": 28300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7636474801292694e-05,
+      "loss": 0.9246,
+      "step": 28310
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.763210760765132e-05,
+      "loss": 0.8165,
+      "step": 28320
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.762774041400996e-05,
+      "loss": 0.8437,
+      "step": 28330
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7623373220368593e-05,
+      "loss": 0.9607,
+      "step": 28340
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.761900602672723e-05,
+      "loss": 0.8957,
+      "step": 28350
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.761463883308587e-05,
+      "loss": 0.9313,
+      "step": 28360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.761027163944449e-05,
+      "loss": 0.8727,
+      "step": 28370
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.760590444580313e-05,
+      "loss": 0.8245,
+      "step": 28380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.760153725216176e-05,
+      "loss": 0.9247,
+      "step": 28390
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.75971700585204e-05,
+      "loss": 0.8337,
+      "step": 28400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7592802864879026e-05,
+      "loss": 0.909,
+      "step": 28410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7588435671237666e-05,
+      "loss": 0.9037,
+      "step": 28420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.75840684775963e-05,
+      "loss": 1.0252,
+      "step": 28430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.757970128395493e-05,
+      "loss": 0.9458,
+      "step": 28440
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7575334090313565e-05,
+      "loss": 0.9845,
+      "step": 28450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.75709668966722e-05,
+      "loss": 0.8313,
+      "step": 28460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.756659970303084e-05,
+      "loss": 0.7355,
+      "step": 28470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7562232509389465e-05,
+      "loss": 0.7191,
+      "step": 28480
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7557865315748105e-05,
+      "loss": 0.8746,
+      "step": 28490
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.755349812210673e-05,
+      "loss": 0.8124,
+      "step": 28500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.754913092846537e-05,
+      "loss": 0.9495,
+      "step": 28510
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7544763734824005e-05,
+      "loss": 0.8358,
+      "step": 28520
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.754039654118264e-05,
+      "loss": 0.8624,
+      "step": 28530
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.753602934754127e-05,
+      "loss": 0.9682,
+      "step": 28540
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7531662153899904e-05,
+      "loss": 0.9362,
+      "step": 28550
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7527294960258544e-05,
+      "loss": 0.8355,
+      "step": 28560
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.752292776661717e-05,
+      "loss": 0.7724,
+      "step": 28570
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.751856057297581e-05,
+      "loss": 0.8557,
+      "step": 28580
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.751419337933444e-05,
+      "loss": 0.8543,
+      "step": 28590
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.750982618569308e-05,
+      "loss": 0.909,
+      "step": 28600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.750545899205171e-05,
+      "loss": 0.96,
+      "step": 28610
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7501091798410343e-05,
+      "loss": 0.9205,
+      "step": 28620
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.749672460476898e-05,
+      "loss": 0.8027,
+      "step": 28630
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.749235741112761e-05,
+      "loss": 0.8754,
+      "step": 28640
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.748799021748624e-05,
+      "loss": 0.8994,
+      "step": 28650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7483623023844876e-05,
+      "loss": 0.9041,
+      "step": 28660
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7479255830203516e-05,
+      "loss": 0.8703,
+      "step": 28670
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.747488863656214e-05,
+      "loss": 0.9414,
+      "step": 28680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.747052144292078e-05,
+      "loss": 0.8962,
+      "step": 28690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7466154249279416e-05,
+      "loss": 0.9463,
+      "step": 28700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.746178705563805e-05,
+      "loss": 0.9432,
+      "step": 28710
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.745741986199668e-05,
+      "loss": 0.8816,
+      "step": 28720
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7453052668355316e-05,
+      "loss": 0.9267,
+      "step": 28730
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.744868547471395e-05,
+      "loss": 0.9836,
+      "step": 28740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.744431828107258e-05,
+      "loss": 0.8484,
+      "step": 28750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.743995108743122e-05,
+      "loss": 0.9633,
+      "step": 28760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7435583893789855e-05,
+      "loss": 0.8619,
+      "step": 28770
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.743121670014849e-05,
+      "loss": 0.9429,
+      "step": 28780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.742684950650712e-05,
+      "loss": 0.895,
+      "step": 28790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7422482312865755e-05,
+      "loss": 0.9302,
+      "step": 28800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.741811511922439e-05,
+      "loss": 0.8122,
+      "step": 28810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.741374792558302e-05,
+      "loss": 0.8599,
+      "step": 28820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7409380731941654e-05,
+      "loss": 0.845,
+      "step": 28830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.740501353830029e-05,
+      "loss": 0.9022,
+      "step": 28840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.740064634465892e-05,
+      "loss": 0.8399,
+      "step": 28850
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.739627915101756e-05,
+      "loss": 0.9031,
+      "step": 28860
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7391911957376194e-05,
+      "loss": 0.773,
+      "step": 28870
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.738754476373483e-05,
+      "loss": 0.9786,
+      "step": 28880
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.738317757009346e-05,
+      "loss": 0.9155,
+      "step": 28890
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7378810376452094e-05,
+      "loss": 0.8825,
+      "step": 28900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.737444318281073e-05,
+      "loss": 0.9342,
+      "step": 28910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.737007598916936e-05,
+      "loss": 0.9824,
+      "step": 28920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.736570879552799e-05,
+      "loss": 0.8604,
+      "step": 28930
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7361341601886626e-05,
+      "loss": 0.9313,
+      "step": 28940
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7356974408245266e-05,
+      "loss": 0.9212,
+      "step": 28950
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.73526072146039e-05,
+      "loss": 0.9935,
+      "step": 28960
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.734824002096253e-05,
+      "loss": 0.9001,
+      "step": 28970
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7343872827321166e-05,
+      "loss": 0.8225,
+      "step": 28980
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.73395056336798e-05,
+      "loss": 0.8585,
+      "step": 28990
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.733513844003843e-05,
+      "loss": 0.7686,
+      "step": 29000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7330771246397066e-05,
+      "loss": 0.897,
+      "step": 29010
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7326404052755705e-05,
+      "loss": 0.9644,
+      "step": 29020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.732203685911433e-05,
+      "loss": 1.0603,
+      "step": 29030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.731766966547297e-05,
+      "loss": 0.9168,
+      "step": 29040
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.73133024718316e-05,
+      "loss": 0.9792,
+      "step": 29050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.730893527819024e-05,
+      "loss": 0.9122,
+      "step": 29060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.730456808454887e-05,
+      "loss": 0.9485,
+      "step": 29070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7300200890907505e-05,
+      "loss": 0.7853,
+      "step": 29080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.729583369726614e-05,
+      "loss": 0.8752,
+      "step": 29090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.729146650362477e-05,
+      "loss": 0.9133,
+      "step": 29100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.728709930998341e-05,
+      "loss": 0.8705,
+      "step": 29110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.728273211634204e-05,
+      "loss": 0.8165,
+      "step": 29120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.727836492270068e-05,
+      "loss": 0.862,
+      "step": 29130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7273997729059304e-05,
+      "loss": 0.8964,
+      "step": 29140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7269630535417944e-05,
+      "loss": 1.0126,
+      "step": 29150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.726526334177658e-05,
+      "loss": 0.8337,
+      "step": 29160
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.726089614813521e-05,
+      "loss": 0.9428,
+      "step": 29170
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7256528954493844e-05,
+      "loss": 0.8714,
+      "step": 29180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.725216176085248e-05,
+      "loss": 0.8689,
+      "step": 29190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.724779456721112e-05,
+      "loss": 0.8388,
+      "step": 29200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.724342737356974e-05,
+      "loss": 0.8633,
+      "step": 29210
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.723906017992838e-05,
+      "loss": 0.9142,
+      "step": 29220
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.723469298628701e-05,
+      "loss": 0.8784,
+      "step": 29230
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.723032579264565e-05,
+      "loss": 0.8121,
+      "step": 29240
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7225958599004276e-05,
+      "loss": 0.8571,
+      "step": 29250
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7221591405362916e-05,
+      "loss": 0.9242,
+      "step": 29260
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.721722421172155e-05,
+      "loss": 0.9927,
+      "step": 29270
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.721285701808018e-05,
+      "loss": 1.0068,
+      "step": 29280
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.720848982443882e-05,
+      "loss": 0.9295,
+      "step": 29290
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.720412263079745e-05,
+      "loss": 0.91,
+      "step": 29300
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.719975543715609e-05,
+      "loss": 0.8985,
+      "step": 29310
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7195388243514715e-05,
+      "loss": 0.8521,
+      "step": 29320
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7191021049873355e-05,
+      "loss": 0.9149,
+      "step": 29330
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.718665385623198e-05,
+      "loss": 0.8992,
+      "step": 29340
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.718228666259062e-05,
+      "loss": 0.8932,
+      "step": 29350
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7177919468949255e-05,
+      "loss": 0.9405,
+      "step": 29360
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.717355227530789e-05,
+      "loss": 0.8556,
+      "step": 29370
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.716918508166652e-05,
+      "loss": 0.8299,
+      "step": 29380
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7164817888025154e-05,
+      "loss": 0.8066,
+      "step": 29390
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7160450694383794e-05,
+      "loss": 0.816,
+      "step": 29400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.715608350074242e-05,
+      "loss": 0.8753,
+      "step": 29410
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.715171630710106e-05,
+      "loss": 0.8798,
+      "step": 29420
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7147349113459694e-05,
+      "loss": 0.9945,
+      "step": 29430
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.714298191981833e-05,
+      "loss": 0.9417,
+      "step": 29440
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.713861472617696e-05,
+      "loss": 0.9615,
+      "step": 29450
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7134247532535594e-05,
+      "loss": 0.8597,
+      "step": 29460
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.712988033889423e-05,
+      "loss": 0.951,
+      "step": 29470
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.712551314525286e-05,
+      "loss": 0.8804,
+      "step": 29480
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.71211459516115e-05,
+      "loss": 0.9168,
+      "step": 29490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7116778757970126e-05,
+      "loss": 0.8557,
+      "step": 29500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7112411564328766e-05,
+      "loss": 0.8645,
+      "step": 29510
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.71080443706874e-05,
+      "loss": 1.0844,
+      "step": 29520
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.710367717704603e-05,
+      "loss": 0.8833,
+      "step": 29530
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7099309983404666e-05,
+      "loss": 0.7537,
+      "step": 29540
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.70949427897633e-05,
+      "loss": 0.9232,
+      "step": 29550
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.709057559612193e-05,
+      "loss": 0.8953,
+      "step": 29560
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7086208402480566e-05,
+      "loss": 0.8938,
+      "step": 29570
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.70818412088392e-05,
+      "loss": 0.9393,
+      "step": 29580
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.707747401519784e-05,
+      "loss": 0.8734,
+      "step": 29590
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.707310682155647e-05,
+      "loss": 0.7744,
+      "step": 29600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7068739627915105e-05,
+      "loss": 0.8491,
+      "step": 29610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.706437243427374e-05,
+      "loss": 0.9464,
+      "step": 29620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.706000524063237e-05,
+      "loss": 1.1008,
+      "step": 29630
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7055638046991005e-05,
+      "loss": 0.8729,
+      "step": 29640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.705127085334964e-05,
+      "loss": 1.0173,
+      "step": 29650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.704690365970827e-05,
+      "loss": 0.8467,
+      "step": 29660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7042536466066904e-05,
+      "loss": 0.831,
+      "step": 29670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7038169272425544e-05,
+      "loss": 0.8869,
+      "step": 29680
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.703380207878418e-05,
+      "loss": 0.9159,
+      "step": 29690
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.702943488514281e-05,
+      "loss": 0.7965,
+      "step": 29700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7025067691501444e-05,
+      "loss": 0.8959,
+      "step": 29710
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.702070049786008e-05,
+      "loss": 0.8991,
+      "step": 29720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.701633330421871e-05,
+      "loss": 0.7936,
+      "step": 29730
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7011966110577344e-05,
+      "loss": 0.7891,
+      "step": 29740
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.700759891693598e-05,
+      "loss": 0.7465,
+      "step": 29750
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.700323172329461e-05,
+      "loss": 0.9482,
+      "step": 29760
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.699886452965325e-05,
+      "loss": 0.8593,
+      "step": 29770
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.6994497336011876e-05,
+      "loss": 0.8967,
+      "step": 29780
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.6990130142370516e-05,
+      "loss": 0.8984,
+      "step": 29790
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.698576294872915e-05,
+      "loss": 0.7997,
+      "step": 29800
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.698139575508778e-05,
+      "loss": 1.0243,
+      "step": 29810
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.6977028561446416e-05,
+      "loss": 0.8945,
+      "step": 29820
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.697266136780505e-05,
+      "loss": 0.824,
+      "step": 29830
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.696829417416369e-05,
+      "loss": 0.8862,
+      "step": 29840
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6963926980522316e-05,
+      "loss": 0.9838,
+      "step": 29850
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6959559786880956e-05,
+      "loss": 0.8577,
+      "step": 29860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.695519259323958e-05,
+      "loss": 0.9283,
+      "step": 29870
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.695082539959822e-05,
+      "loss": 0.844,
+      "step": 29880
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6946458205956855e-05,
+      "loss": 0.8483,
+      "step": 29890
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.694209101231549e-05,
+      "loss": 0.9538,
+      "step": 29900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.693772381867412e-05,
+      "loss": 0.9602,
+      "step": 29910
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6933356625032755e-05,
+      "loss": 0.9144,
+      "step": 29920
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6928989431391395e-05,
+      "loss": 0.897,
+      "step": 29930
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.692462223775002e-05,
+      "loss": 0.9545,
+      "step": 29940
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.692025504410866e-05,
+      "loss": 0.929,
+      "step": 29950
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.691588785046729e-05,
+      "loss": 0.9091,
+      "step": 29960
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.691152065682593e-05,
+      "loss": 0.8406,
+      "step": 29970
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6907153463184554e-05,
+      "loss": 0.7906,
+      "step": 29980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6902786269543194e-05,
+      "loss": 0.9229,
+      "step": 29990
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.689841907590183e-05,
+      "loss": 0.7617,
+      "step": 30000
+    },
+    {
+      "epoch": 2.62,
+      "eval_accuracy": 0.5697160263338902,
+      "eval_loss": 0.8902987241744995,
+      "eval_runtime": 84.089,
+      "eval_samples_per_second": 121.027,
+      "eval_steps_per_second": 15.139,
+      "step": 30000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.689405188226046e-05,
+      "loss": 0.8725,
+      "step": 30010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.68896846886191e-05,
+      "loss": 0.9209,
+      "step": 30020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.688531749497773e-05,
+      "loss": 0.8024,
+      "step": 30030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.688095030133637e-05,
+      "loss": 0.8148,
+      "step": 30040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.687658310769499e-05,
+      "loss": 0.9597,
+      "step": 30050
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.687221591405363e-05,
+      "loss": 0.9644,
+      "step": 30060
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.686784872041226e-05,
+      "loss": 0.8896,
+      "step": 30070
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.68634815267709e-05,
+      "loss": 0.8064,
+      "step": 30080
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.685911433312953e-05,
+      "loss": 0.9304,
+      "step": 30090
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6854747139488166e-05,
+      "loss": 0.7495,
+      "step": 30100
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.68503799458468e-05,
+      "loss": 0.876,
+      "step": 30110
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.684601275220543e-05,
+      "loss": 0.87,
+      "step": 30120
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.684164555856407e-05,
+      "loss": 0.923,
+      "step": 30130
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.68372783649227e-05,
+      "loss": 0.7467,
+      "step": 30140
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.683291117128134e-05,
+      "loss": 0.9045,
+      "step": 30150
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6828543977639965e-05,
+      "loss": 0.8363,
+      "step": 30160
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6824176783998605e-05,
+      "loss": 0.8909,
+      "step": 30170
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.681980959035724e-05,
+      "loss": 0.9214,
+      "step": 30180
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.681544239671587e-05,
+      "loss": 0.8907,
+      "step": 30190
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6811075203074505e-05,
+      "loss": 0.9207,
+      "step": 30200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.680670800943314e-05,
+      "loss": 0.895,
+      "step": 30210
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.680234081579178e-05,
+      "loss": 0.826,
+      "step": 30220
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6797973622150404e-05,
+      "loss": 0.95,
+      "step": 30230
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6793606428509044e-05,
+      "loss": 0.845,
+      "step": 30240
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.678923923486768e-05,
+      "loss": 0.8591,
+      "step": 30250
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.678487204122631e-05,
+      "loss": 0.8775,
+      "step": 30260
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6780504847584944e-05,
+      "loss": 1.0453,
+      "step": 30270
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.677613765394358e-05,
+      "loss": 0.7653,
+      "step": 30280
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.677177046030221e-05,
+      "loss": 0.9071,
+      "step": 30290
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6767403266660844e-05,
+      "loss": 0.8831,
+      "step": 30300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.676303607301948e-05,
+      "loss": 0.9214,
+      "step": 30310
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.675866887937811e-05,
+      "loss": 0.7844,
+      "step": 30320
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.675430168573675e-05,
+      "loss": 0.8768,
+      "step": 30330
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.674993449209538e-05,
+      "loss": 0.9314,
+      "step": 30340
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6745567298454016e-05,
+      "loss": 0.8272,
+      "step": 30350
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.674120010481265e-05,
+      "loss": 0.971,
+      "step": 30360
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.673683291117128e-05,
+      "loss": 0.8493,
+      "step": 30370
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6732465717529916e-05,
+      "loss": 0.8179,
+      "step": 30380
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.672809852388855e-05,
+      "loss": 1.0002,
+      "step": 30390
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.672373133024718e-05,
+      "loss": 0.9528,
+      "step": 30400
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6719364136605816e-05,
+      "loss": 0.8391,
+      "step": 30410
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6714996942964456e-05,
+      "loss": 0.9359,
+      "step": 30420
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.671062974932309e-05,
+      "loss": 0.7887,
+      "step": 30430
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.670626255568172e-05,
+      "loss": 0.8466,
+      "step": 30440
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6701895362040355e-05,
+      "loss": 0.8466,
+      "step": 30450
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.669752816839899e-05,
+      "loss": 0.7982,
+      "step": 30460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.669316097475762e-05,
+      "loss": 1.0767,
+      "step": 30470
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6688793781116255e-05,
+      "loss": 0.9009,
+      "step": 30480
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.668442658747489e-05,
+      "loss": 0.9746,
+      "step": 30490
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.668005939383353e-05,
+      "loss": 0.8384,
+      "step": 30500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6675692200192154e-05,
+      "loss": 0.9489,
+      "step": 30510
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6671325006550794e-05,
+      "loss": 0.7793,
+      "step": 30520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.666695781290943e-05,
+      "loss": 0.966,
+      "step": 30530
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.666259061926806e-05,
+      "loss": 0.8517,
+      "step": 30540
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6658223425626694e-05,
+      "loss": 0.8766,
+      "step": 30550
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.665385623198533e-05,
+      "loss": 0.8056,
+      "step": 30560
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.664948903834396e-05,
+      "loss": 0.9925,
+      "step": 30570
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6645121844702594e-05,
+      "loss": 0.9064,
+      "step": 30580
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6640754651061234e-05,
+      "loss": 0.9196,
+      "step": 30590
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.663638745741986e-05,
+      "loss": 0.9677,
+      "step": 30600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.66320202637785e-05,
+      "loss": 0.8676,
+      "step": 30610
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.662765307013713e-05,
+      "loss": 0.9462,
+      "step": 30620
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6623285876495766e-05,
+      "loss": 1.055,
+      "step": 30630
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.66189186828544e-05,
+      "loss": 0.8715,
+      "step": 30640
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.661455148921303e-05,
+      "loss": 0.9595,
+      "step": 30650
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.661018429557167e-05,
+      "loss": 0.8803,
+      "step": 30660
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.66058171019303e-05,
+      "loss": 0.8949,
+      "step": 30670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.660144990828894e-05,
+      "loss": 0.9004,
+      "step": 30680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6597082714647566e-05,
+      "loss": 0.9795,
+      "step": 30690
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6592715521006206e-05,
+      "loss": 0.86,
+      "step": 30700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.658834832736483e-05,
+      "loss": 0.8744,
+      "step": 30710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.658398113372347e-05,
+      "loss": 0.8781,
+      "step": 30720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6579613940082105e-05,
+      "loss": 0.987,
+      "step": 30730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.657524674644074e-05,
+      "loss": 0.9166,
+      "step": 30740
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.657087955279938e-05,
+      "loss": 0.8625,
+      "step": 30750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6566512359158005e-05,
+      "loss": 0.8649,
+      "step": 30760
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6562145165516645e-05,
+      "loss": 0.827,
+      "step": 30770
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.655777797187527e-05,
+      "loss": 0.829,
+      "step": 30780
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.655341077823391e-05,
+      "loss": 0.8834,
+      "step": 30790
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.654904358459254e-05,
+      "loss": 0.9434,
+      "step": 30800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.654467639095118e-05,
+      "loss": 0.7786,
+      "step": 30810
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.654030919730981e-05,
+      "loss": 0.8089,
+      "step": 30820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6535942003668444e-05,
+      "loss": 0.9265,
+      "step": 30830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.653157481002708e-05,
+      "loss": 0.8696,
+      "step": 30840
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.652720761638571e-05,
+      "loss": 0.9755,
+      "step": 30850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.652284042274435e-05,
+      "loss": 0.8557,
+      "step": 30860
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.651847322910298e-05,
+      "loss": 0.8577,
+      "step": 30870
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.651410603546162e-05,
+      "loss": 0.9624,
+      "step": 30880
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.650973884182024e-05,
+      "loss": 0.7929,
+      "step": 30890
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.650537164817888e-05,
+      "loss": 0.8877,
+      "step": 30900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6501004454537516e-05,
+      "loss": 0.9239,
+      "step": 30910
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.649663726089615e-05,
+      "loss": 0.804,
+      "step": 30920
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.649227006725478e-05,
+      "loss": 0.8929,
+      "step": 30930
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6487902873613416e-05,
+      "loss": 0.9311,
+      "step": 30940
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6483535679972056e-05,
+      "loss": 0.9447,
+      "step": 30950
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.647916848633068e-05,
+      "loss": 0.8262,
+      "step": 30960
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.647480129268932e-05,
+      "loss": 0.8579,
+      "step": 30970
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.647043409904795e-05,
+      "loss": 0.9707,
+      "step": 30980
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.646606690540659e-05,
+      "loss": 0.9639,
+      "step": 30990
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.646169971176522e-05,
+      "loss": 0.9303,
+      "step": 31000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6457332518123855e-05,
+      "loss": 0.9284,
+      "step": 31010
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.645296532448249e-05,
+      "loss": 0.812,
+      "step": 31020
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.644859813084112e-05,
+      "loss": 0.8999,
+      "step": 31030
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6444230937199755e-05,
+      "loss": 0.7755,
+      "step": 31040
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.643986374355839e-05,
+      "loss": 0.9287,
+      "step": 31050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.643549654991703e-05,
+      "loss": 0.8968,
+      "step": 31060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.643112935627566e-05,
+      "loss": 0.8468,
+      "step": 31070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6426762162634294e-05,
+      "loss": 0.8325,
+      "step": 31080
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.642239496899293e-05,
+      "loss": 0.8397,
+      "step": 31090
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.641802777535156e-05,
+      "loss": 1.0066,
+      "step": 31100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6413660581710194e-05,
+      "loss": 0.9081,
+      "step": 31110
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.640929338806883e-05,
+      "loss": 0.8058,
+      "step": 31120
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.640492619442746e-05,
+      "loss": 0.8724,
+      "step": 31130
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6400559000786094e-05,
+      "loss": 0.938,
+      "step": 31140
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6396191807144734e-05,
+      "loss": 0.811,
+      "step": 31150
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.639182461350337e-05,
+      "loss": 0.8148,
+      "step": 31160
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6387457419862e-05,
+      "loss": 0.9617,
+      "step": 31170
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.638309022622063e-05,
+      "loss": 0.8239,
+      "step": 31180
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6378723032579266e-05,
+      "loss": 0.8597,
+      "step": 31190
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.63743558389379e-05,
+      "loss": 0.7862,
+      "step": 31200
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.636998864529653e-05,
+      "loss": 1.0252,
+      "step": 31210
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6365621451655166e-05,
+      "loss": 0.8375,
+      "step": 31220
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.63612542580138e-05,
+      "loss": 0.7994,
+      "step": 31230
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.635688706437243e-05,
+      "loss": 0.8623,
+      "step": 31240
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.635251987073107e-05,
+      "loss": 0.8096,
+      "step": 31250
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6348152677089706e-05,
+      "loss": 0.8325,
+      "step": 31260
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.634378548344834e-05,
+      "loss": 0.9191,
+      "step": 31270
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.633941828980697e-05,
+      "loss": 0.9055,
+      "step": 31280
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6335051096165605e-05,
+      "loss": 0.9339,
+      "step": 31290
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.633068390252424e-05,
+      "loss": 0.8545,
+      "step": 31300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.632631670888287e-05,
+      "loss": 0.9295,
+      "step": 31310
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.632194951524151e-05,
+      "loss": 0.857,
+      "step": 31320
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.631758232160014e-05,
+      "loss": 0.9053,
+      "step": 31330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.631321512795878e-05,
+      "loss": 0.869,
+      "step": 31340
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.630884793431741e-05,
+      "loss": 0.8556,
+      "step": 31350
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6304480740676044e-05,
+      "loss": 0.9302,
+      "step": 31360
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.630011354703468e-05,
+      "loss": 0.9028,
+      "step": 31370
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.629574635339331e-05,
+      "loss": 0.8506,
+      "step": 31380
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6291379159751944e-05,
+      "loss": 1.0644,
+      "step": 31390
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.628701196611058e-05,
+      "loss": 0.8551,
+      "step": 31400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.628264477246922e-05,
+      "loss": 0.9015,
+      "step": 31410
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6278277578827844e-05,
+      "loss": 0.9056,
+      "step": 31420
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6273910385186484e-05,
+      "loss": 0.837,
+      "step": 31430
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.626954319154511e-05,
+      "loss": 0.9594,
+      "step": 31440
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.626517599790375e-05,
+      "loss": 0.9665,
+      "step": 31450
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.626080880426238e-05,
+      "loss": 0.9836,
+      "step": 31460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6256441610621016e-05,
+      "loss": 0.9962,
+      "step": 31470
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6252074416979656e-05,
+      "loss": 0.9267,
+      "step": 31480
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.624770722333828e-05,
+      "loss": 0.9897,
+      "step": 31490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.624334002969692e-05,
+      "loss": 0.9022,
+      "step": 31500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.623897283605555e-05,
+      "loss": 0.8704,
+      "step": 31510
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.623460564241419e-05,
+      "loss": 0.8266,
+      "step": 31520
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6230238448772816e-05,
+      "loss": 0.9212,
+      "step": 31530
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6225871255131456e-05,
+      "loss": 0.8737,
+      "step": 31540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.622150406149009e-05,
+      "loss": 0.8937,
+      "step": 31550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.621713686784872e-05,
+      "loss": 0.807,
+      "step": 31560
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6212769674207355e-05,
+      "loss": 0.8526,
+      "step": 31570
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.620840248056599e-05,
+      "loss": 0.8936,
+      "step": 31580
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.620403528692463e-05,
+      "loss": 0.9244,
+      "step": 31590
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6199668093283255e-05,
+      "loss": 1.0083,
+      "step": 31600
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6195300899641895e-05,
+      "loss": 0.9569,
+      "step": 31610
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.619093370600052e-05,
+      "loss": 0.9787,
+      "step": 31620
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.618656651235916e-05,
+      "loss": 1.012,
+      "step": 31630
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.618219931871779e-05,
+      "loss": 0.889,
+      "step": 31640
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.617783212507643e-05,
+      "loss": 0.8541,
+      "step": 31650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.617346493143506e-05,
+      "loss": 0.8598,
+      "step": 31660
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6169097737793694e-05,
+      "loss": 0.9854,
+      "step": 31670
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6164730544152334e-05,
+      "loss": 0.8487,
+      "step": 31680
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.616036335051096e-05,
+      "loss": 0.9482,
+      "step": 31690
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.61559961568696e-05,
+      "loss": 0.8416,
+      "step": 31700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.615162896322823e-05,
+      "loss": 0.9164,
+      "step": 31710
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.614726176958687e-05,
+      "loss": 0.9078,
+      "step": 31720
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.61428945759455e-05,
+      "loss": 0.9142,
+      "step": 31730
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.613852738230413e-05,
+      "loss": 0.9093,
+      "step": 31740
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6134160188662766e-05,
+      "loss": 0.8964,
+      "step": 31750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.61297929950214e-05,
+      "loss": 0.8281,
+      "step": 31760
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.612542580138003e-05,
+      "loss": 0.9545,
+      "step": 31770
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6121058607738666e-05,
+      "loss": 1.0251,
+      "step": 31780
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6116691414097306e-05,
+      "loss": 0.879,
+      "step": 31790
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.611232422045593e-05,
+      "loss": 0.9087,
+      "step": 31800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.610795702681457e-05,
+      "loss": 0.9483,
+      "step": 31810
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6103589833173206e-05,
+      "loss": 0.8455,
+      "step": 31820
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.609922263953184e-05,
+      "loss": 0.7958,
+      "step": 31830
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.609485544589047e-05,
+      "loss": 0.9556,
+      "step": 31840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6090488252249105e-05,
+      "loss": 0.9793,
+      "step": 31850
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.608612105860774e-05,
+      "loss": 1.0086,
+      "step": 31860
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.608175386496637e-05,
+      "loss": 0.8565,
+      "step": 31870
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.607738667132501e-05,
+      "loss": 0.8732,
+      "step": 31880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6073019477683645e-05,
+      "loss": 0.9144,
+      "step": 31890
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.606865228404228e-05,
+      "loss": 0.7981,
+      "step": 31900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.606428509040091e-05,
+      "loss": 0.9067,
+      "step": 31910
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6059917896759544e-05,
+      "loss": 0.7604,
+      "step": 31920
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.605555070311818e-05,
+      "loss": 0.9703,
+      "step": 31930
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.605118350947681e-05,
+      "loss": 0.9444,
+      "step": 31940
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6046816315835444e-05,
+      "loss": 0.9026,
+      "step": 31950
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.604244912219408e-05,
+      "loss": 0.8295,
+      "step": 31960
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.603808192855271e-05,
+      "loss": 1.0002,
+      "step": 31970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.603371473491135e-05,
+      "loss": 0.9038,
+      "step": 31980
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6029347541269984e-05,
+      "loss": 0.9065,
+      "step": 31990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.602498034762862e-05,
+      "loss": 0.8089,
+      "step": 32000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.602061315398725e-05,
+      "loss": 0.8224,
+      "step": 32010
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.601624596034588e-05,
+      "loss": 0.8463,
+      "step": 32020
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6011878766704516e-05,
+      "loss": 0.9144,
+      "step": 32030
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.600751157306315e-05,
+      "loss": 0.8298,
+      "step": 32040
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.600314437942178e-05,
+      "loss": 0.9097,
+      "step": 32050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.5998777185780416e-05,
+      "loss": 0.7905,
+      "step": 32060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.5994409992139056e-05,
+      "loss": 1.0139,
+      "step": 32070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.599004279849769e-05,
+      "loss": 0.933,
+      "step": 32080
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.598567560485632e-05,
+      "loss": 0.947,
+      "step": 32090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.5981308411214956e-05,
+      "loss": 0.9221,
+      "step": 32100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.597694121757359e-05,
+      "loss": 0.948,
+      "step": 32110
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.597257402393222e-05,
+      "loss": 0.8697,
+      "step": 32120
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5968206830290855e-05,
+      "loss": 0.9354,
+      "step": 32130
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5963839636649495e-05,
+      "loss": 0.9762,
+      "step": 32140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.595947244300812e-05,
+      "loss": 0.8958,
+      "step": 32150
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.595510524936676e-05,
+      "loss": 0.8151,
+      "step": 32160
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5950738055725395e-05,
+      "loss": 0.9358,
+      "step": 32170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.594637086208403e-05,
+      "loss": 1.0005,
+      "step": 32180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.594200366844266e-05,
+      "loss": 1.0432,
+      "step": 32190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5937636474801294e-05,
+      "loss": 0.8377,
+      "step": 32200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.593326928115993e-05,
+      "loss": 0.891,
+      "step": 32210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.592890208751856e-05,
+      "loss": 0.9114,
+      "step": 32220
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.59245348938772e-05,
+      "loss": 0.967,
+      "step": 32230
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.592016770023583e-05,
+      "loss": 0.949,
+      "step": 32240
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.591580050659447e-05,
+      "loss": 0.9408,
+      "step": 32250
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5911433312953094e-05,
+      "loss": 0.9918,
+      "step": 32260
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5907066119311734e-05,
+      "loss": 0.9556,
+      "step": 32270
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.590269892567037e-05,
+      "loss": 0.7735,
+      "step": 32280
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5898331732029e-05,
+      "loss": 0.9057,
+      "step": 32290
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.589396453838763e-05,
+      "loss": 0.9906,
+      "step": 32300
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5889597344746266e-05,
+      "loss": 0.8508,
+      "step": 32310
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5885230151104906e-05,
+      "loss": 0.8245,
+      "step": 32320
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.588086295746353e-05,
+      "loss": 0.8635,
+      "step": 32330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.587649576382217e-05,
+      "loss": 0.9695,
+      "step": 32340
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.58721285701808e-05,
+      "loss": 0.9106,
+      "step": 32350
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.586776137653944e-05,
+      "loss": 0.8739,
+      "step": 32360
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.586339418289807e-05,
+      "loss": 0.9221,
+      "step": 32370
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5859026989256706e-05,
+      "loss": 1.082,
+      "step": 32380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.585465979561534e-05,
+      "loss": 0.7639,
+      "step": 32390
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.585029260197397e-05,
+      "loss": 0.9124,
+      "step": 32400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.584592540833261e-05,
+      "loss": 0.9329,
+      "step": 32410
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.584155821469124e-05,
+      "loss": 0.8865,
+      "step": 32420
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.583719102104988e-05,
+      "loss": 0.8719,
+      "step": 32430
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5832823827408505e-05,
+      "loss": 0.9509,
+      "step": 32440
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5828456633767145e-05,
+      "loss": 0.8419,
+      "step": 32450
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.582408944012577e-05,
+      "loss": 0.8648,
+      "step": 32460
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.581972224648441e-05,
+      "loss": 0.8786,
+      "step": 32470
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5815355052843044e-05,
+      "loss": 0.7561,
+      "step": 32480
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.581098785920168e-05,
+      "loss": 0.7937,
+      "step": 32490
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.580662066556032e-05,
+      "loss": 0.8721,
+      "step": 32500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5802253471918944e-05,
+      "loss": 0.9386,
+      "step": 32510
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5797886278277584e-05,
+      "loss": 0.8892,
+      "step": 32520
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.579351908463621e-05,
+      "loss": 0.9537,
+      "step": 32530
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.578915189099485e-05,
+      "loss": 0.9444,
+      "step": 32540
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5784784697353484e-05,
+      "loss": 0.9662,
+      "step": 32550
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.578041750371212e-05,
+      "loss": 0.8605,
+      "step": 32560
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.577605031007075e-05,
+      "loss": 0.8425,
+      "step": 32570
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.577168311642938e-05,
+      "loss": 0.8483,
+      "step": 32580
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5767315922788016e-05,
+      "loss": 0.7909,
+      "step": 32590
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.576294872914665e-05,
+      "loss": 0.8687,
+      "step": 32600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.575858153550529e-05,
+      "loss": 0.8529,
+      "step": 32610
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5754214341863916e-05,
+      "loss": 0.9204,
+      "step": 32620
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5749847148222556e-05,
+      "loss": 0.9534,
+      "step": 32630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.574547995458119e-05,
+      "loss": 0.8468,
+      "step": 32640
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.574111276093982e-05,
+      "loss": 0.8563,
+      "step": 32650
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5736745567298456e-05,
+      "loss": 0.9261,
+      "step": 32660
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.573237837365709e-05,
+      "loss": 0.9533,
+      "step": 32670
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.572801118001572e-05,
+      "loss": 0.8493,
+      "step": 32680
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5723643986374355e-05,
+      "loss": 0.8436,
+      "step": 32690
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5719276792732995e-05,
+      "loss": 0.9649,
+      "step": 32700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.571490959909162e-05,
+      "loss": 0.7837,
+      "step": 32710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.571054240545026e-05,
+      "loss": 1.0539,
+      "step": 32720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5706175211808895e-05,
+      "loss": 0.8843,
+      "step": 32730
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.570180801816753e-05,
+      "loss": 0.8199,
+      "step": 32740
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.569744082452616e-05,
+      "loss": 0.7991,
+      "step": 32750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5693073630884794e-05,
+      "loss": 0.9486,
+      "step": 32760
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.568870643724343e-05,
+      "loss": 0.8535,
+      "step": 32770
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.568433924360206e-05,
+      "loss": 0.9214,
+      "step": 32780
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5679972049960694e-05,
+      "loss": 0.9739,
+      "step": 32790
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5675604856319334e-05,
+      "loss": 0.9556,
+      "step": 32800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.567123766267797e-05,
+      "loss": 0.9765,
+      "step": 32810
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.56668704690366e-05,
+      "loss": 0.9093,
+      "step": 32820
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5662503275395234e-05,
+      "loss": 0.972,
+      "step": 32830
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.565813608175387e-05,
+      "loss": 0.8315,
+      "step": 32840
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.56537688881125e-05,
+      "loss": 0.8443,
+      "step": 32850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.564940169447113e-05,
+      "loss": 0.9462,
+      "step": 32860
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5645034500829766e-05,
+      "loss": 0.8609,
+      "step": 32870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.56406673071884e-05,
+      "loss": 0.8787,
+      "step": 32880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.563630011354704e-05,
+      "loss": 0.9766,
+      "step": 32890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.563193291990567e-05,
+      "loss": 1.0129,
+      "step": 32900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5627565726264306e-05,
+      "loss": 1.0724,
+      "step": 32910
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.562319853262294e-05,
+      "loss": 0.8128,
+      "step": 32920
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.561883133898157e-05,
+      "loss": 0.8227,
+      "step": 32930
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5614464145340206e-05,
+      "loss": 0.785,
+      "step": 32940
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.561009695169884e-05,
+      "loss": 0.9274,
+      "step": 32950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.560572975805748e-05,
+      "loss": 0.8945,
+      "step": 32960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5601362564416105e-05,
+      "loss": 0.8248,
+      "step": 32970
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5596995370774745e-05,
+      "loss": 0.9042,
+      "step": 32980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.559262817713337e-05,
+      "loss": 0.91,
+      "step": 32990
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.558826098349201e-05,
+      "loss": 0.8769,
+      "step": 33000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5583893789850645e-05,
+      "loss": 0.9262,
+      "step": 33010
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.557952659620928e-05,
+      "loss": 0.9182,
+      "step": 33020
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.557515940256791e-05,
+      "loss": 0.8805,
+      "step": 33030
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5570792208926544e-05,
+      "loss": 0.8361,
+      "step": 33040
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5566425015285184e-05,
+      "loss": 0.7815,
+      "step": 33050
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.556205782164381e-05,
+      "loss": 0.8394,
+      "step": 33060
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.555769062800245e-05,
+      "loss": 0.848,
+      "step": 33070
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.555332343436108e-05,
+      "loss": 0.9131,
+      "step": 33080
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.554895624071972e-05,
+      "loss": 0.8752,
+      "step": 33090
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.554458904707835e-05,
+      "loss": 0.9267,
+      "step": 33100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5540221853436984e-05,
+      "loss": 0.9302,
+      "step": 33110
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.553585465979562e-05,
+      "loss": 0.9191,
+      "step": 33120
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.553148746615425e-05,
+      "loss": 0.8011,
+      "step": 33130
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.552712027251289e-05,
+      "loss": 0.8981,
+      "step": 33140
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5522753078871516e-05,
+      "loss": 0.8593,
+      "step": 33150
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5518385885230156e-05,
+      "loss": 0.8256,
+      "step": 33160
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.551401869158878e-05,
+      "loss": 0.8501,
+      "step": 33170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.550965149794742e-05,
+      "loss": 0.8409,
+      "step": 33180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.550528430430605e-05,
+      "loss": 0.8818,
+      "step": 33190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.550091711066469e-05,
+      "loss": 0.8558,
+      "step": 33200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.549654991702332e-05,
+      "loss": 0.8975,
+      "step": 33210
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5492182723381956e-05,
+      "loss": 0.8991,
+      "step": 33220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5487815529740596e-05,
+      "loss": 0.9684,
+      "step": 33230
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.548344833609922e-05,
+      "loss": 0.8138,
+      "step": 33240
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.547908114245786e-05,
+      "loss": 0.884,
+      "step": 33250
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.547471394881649e-05,
+      "loss": 0.766,
+      "step": 33260
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.547034675517513e-05,
+      "loss": 0.9252,
+      "step": 33270
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5465979561533755e-05,
+      "loss": 0.856,
+      "step": 33280
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5461612367892395e-05,
+      "loss": 0.9005,
+      "step": 33290
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.545724517425103e-05,
+      "loss": 0.835,
+      "step": 33300
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.545287798060966e-05,
+      "loss": 0.8151,
+      "step": 33310
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5448510786968294e-05,
+      "loss": 0.8401,
+      "step": 33320
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.544414359332693e-05,
+      "loss": 1.054,
+      "step": 33330
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.543977639968557e-05,
+      "loss": 0.7953,
+      "step": 33340
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5435409206044194e-05,
+      "loss": 0.8432,
+      "step": 33350
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5431042012402834e-05,
+      "loss": 0.8792,
+      "step": 33360
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.542667481876147e-05,
+      "loss": 0.8561,
+      "step": 33370
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.54223076251201e-05,
+      "loss": 0.8802,
+      "step": 33380
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5417940431478734e-05,
+      "loss": 0.7915,
+      "step": 33390
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.541357323783737e-05,
+      "loss": 0.77,
+      "step": 33400
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5409206044196e-05,
+      "loss": 1.0202,
+      "step": 33410
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.540483885055463e-05,
+      "loss": 0.8132,
+      "step": 33420
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.540047165691327e-05,
+      "loss": 0.8365,
+      "step": 33430
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.53961044632719e-05,
+      "loss": 0.8153,
+      "step": 33440
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.539173726963054e-05,
+      "loss": 0.8786,
+      "step": 33450
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.538737007598917e-05,
+      "loss": 0.9782,
+      "step": 33460
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5383002882347806e-05,
+      "loss": 0.9631,
+      "step": 33470
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.537863568870644e-05,
+      "loss": 0.8552,
+      "step": 33480
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.537426849506507e-05,
+      "loss": 0.8429,
+      "step": 33490
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5369901301423706e-05,
+      "loss": 0.8336,
+      "step": 33500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.536553410778234e-05,
+      "loss": 0.8604,
+      "step": 33510
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.536116691414097e-05,
+      "loss": 0.9319,
+      "step": 33520
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5356799720499605e-05,
+      "loss": 0.9256,
+      "step": 33530
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5352432526858245e-05,
+      "loss": 0.8423,
+      "step": 33540
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.534806533321688e-05,
+      "loss": 0.9648,
+      "step": 33550
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.534369813957551e-05,
+      "loss": 1.0002,
+      "step": 33560
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5339330945934145e-05,
+      "loss": 0.8719,
+      "step": 33570
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.533496375229278e-05,
+      "loss": 0.8496,
+      "step": 33580
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.533059655865141e-05,
+      "loss": 0.8903,
+      "step": 33590
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5326229365010044e-05,
+      "loss": 0.8984,
+      "step": 33600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.532186217136868e-05,
+      "loss": 0.9814,
+      "step": 33610
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.531749497772732e-05,
+      "loss": 0.8566,
+      "step": 33620
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.531312778408595e-05,
+      "loss": 0.9102,
+      "step": 33630
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5308760590444584e-05,
+      "loss": 0.9665,
+      "step": 33640
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.530439339680322e-05,
+      "loss": 0.8606,
+      "step": 33650
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.530002620316185e-05,
+      "loss": 0.9535,
+      "step": 33660
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5295659009520484e-05,
+      "loss": 0.9405,
+      "step": 33670
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.529129181587912e-05,
+      "loss": 0.893,
+      "step": 33680
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.528692462223775e-05,
+      "loss": 1.0814,
+      "step": 33690
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.528255742859638e-05,
+      "loss": 0.7987,
+      "step": 33700
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.527819023495502e-05,
+      "loss": 0.8133,
+      "step": 33710
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.527382304131365e-05,
+      "loss": 0.7475,
+      "step": 33720
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.526945584767229e-05,
+      "loss": 0.9253,
+      "step": 33730
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.526508865403092e-05,
+      "loss": 0.8634,
+      "step": 33740
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5260721460389556e-05,
+      "loss": 0.9544,
+      "step": 33750
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.525635426674819e-05,
+      "loss": 0.8355,
+      "step": 33760
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.525198707310682e-05,
+      "loss": 0.7776,
+      "step": 33770
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.524761987946546e-05,
+      "loss": 0.8795,
+      "step": 33780
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.524325268582409e-05,
+      "loss": 0.981,
+      "step": 33790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.523888549218273e-05,
+      "loss": 0.8933,
+      "step": 33800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5234518298541355e-05,
+      "loss": 0.8578,
+      "step": 33810
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5230151104899995e-05,
+      "loss": 0.8661,
+      "step": 33820
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.522578391125863e-05,
+      "loss": 0.8563,
+      "step": 33830
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.522141671761726e-05,
+      "loss": 0.9361,
+      "step": 33840
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5217049523975895e-05,
+      "loss": 0.9242,
+      "step": 33850
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.521268233033453e-05,
+      "loss": 0.9063,
+      "step": 33860
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.520831513669317e-05,
+      "loss": 1.0139,
+      "step": 33870
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5203947943051795e-05,
+      "loss": 0.7928,
+      "step": 33880
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5199580749410434e-05,
+      "loss": 0.8645,
+      "step": 33890
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.519521355576906e-05,
+      "loss": 0.8043,
+      "step": 33900
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.51908463621277e-05,
+      "loss": 0.8156,
+      "step": 33910
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.518647916848633e-05,
+      "loss": 0.9783,
+      "step": 33920
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.518211197484497e-05,
+      "loss": 0.8923,
+      "step": 33930
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.51777447812036e-05,
+      "loss": 0.8076,
+      "step": 33940
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5173377587562234e-05,
+      "loss": 0.8826,
+      "step": 33950
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5169010393920874e-05,
+      "loss": 0.9647,
+      "step": 33960
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.51646432002795e-05,
+      "loss": 0.9384,
+      "step": 33970
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.516027600663814e-05,
+      "loss": 0.8704,
+      "step": 33980
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5155908812996767e-05,
+      "loss": 0.8816,
+      "step": 33990
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5151541619355406e-05,
+      "loss": 0.9729,
+      "step": 34000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.514717442571403e-05,
+      "loss": 0.8491,
+      "step": 34010
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.514280723207267e-05,
+      "loss": 0.8071,
+      "step": 34020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5138440038431306e-05,
+      "loss": 0.8743,
+      "step": 34030
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.513407284478994e-05,
+      "loss": 0.7653,
+      "step": 34040
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.512970565114857e-05,
+      "loss": 0.9276,
+      "step": 34050
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5125338457507206e-05,
+      "loss": 0.8856,
+      "step": 34060
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5120971263865846e-05,
+      "loss": 0.8502,
+      "step": 34070
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.511660407022447e-05,
+      "loss": 0.775,
+      "step": 34080
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.511223687658311e-05,
+      "loss": 0.9047,
+      "step": 34090
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.510786968294174e-05,
+      "loss": 0.88,
+      "step": 34100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.510350248930038e-05,
+      "loss": 0.8273,
+      "step": 34110
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.509913529565901e-05,
+      "loss": 0.7624,
+      "step": 34120
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5094768102017645e-05,
+      "loss": 0.8721,
+      "step": 34130
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.509040090837628e-05,
+      "loss": 0.8909,
+      "step": 34140
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.508603371473491e-05,
+      "loss": 0.8245,
+      "step": 34150
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.508166652109355e-05,
+      "loss": 0.8409,
+      "step": 34160
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.507729932745218e-05,
+      "loss": 0.7988,
+      "step": 34170
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.507293213381082e-05,
+      "loss": 0.8641,
+      "step": 34180
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5068564940169444e-05,
+      "loss": 0.8359,
+      "step": 34190
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5064197746528084e-05,
+      "loss": 0.8981,
+      "step": 34200
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.505983055288672e-05,
+      "loss": 0.8315,
+      "step": 34210
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.505546335924535e-05,
+      "loss": 0.8974,
+      "step": 34220
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5051096165603984e-05,
+      "loss": 0.9257,
+      "step": 34230
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.504672897196262e-05,
+      "loss": 0.948,
+      "step": 34240
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.504236177832125e-05,
+      "loss": 0.8309,
+      "step": 34250
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.503799458467988e-05,
+      "loss": 0.9047,
+      "step": 34260
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.503362739103852e-05,
+      "loss": 0.8787,
+      "step": 34270
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5029260197397157e-05,
+      "loss": 0.9212,
+      "step": 34280
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502489300375579e-05,
+      "loss": 0.9506,
+      "step": 34290
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502052581011442e-05,
+      "loss": 0.8778,
+      "step": 34300
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5016158616473056e-05,
+      "loss": 1.0212,
+      "step": 34310
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.501179142283169e-05,
+      "loss": 0.9111,
+      "step": 34320
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.500742422919032e-05,
+      "loss": 0.9281,
+      "step": 34330
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5003057035548956e-05,
+      "loss": 0.8172,
+      "step": 34340
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.499868984190759e-05,
+      "loss": 1.0171,
+      "step": 34350
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.499432264826623e-05,
+      "loss": 0.938,
+      "step": 34360
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.498995545462486e-05,
+      "loss": 0.8832,
+      "step": 34370
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.4985588260983495e-05,
+      "loss": 0.9279,
+      "step": 34380
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.498122106734213e-05,
+      "loss": 0.7592,
+      "step": 34390
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.497685387370076e-05,
+      "loss": 0.9636,
+      "step": 34400
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4972486680059395e-05,
+      "loss": 0.7962,
+      "step": 34410
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.496811948641803e-05,
+      "loss": 0.9219,
+      "step": 34420
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.496375229277666e-05,
+      "loss": 0.8833,
+      "step": 34430
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.49593850991353e-05,
+      "loss": 0.7864,
+      "step": 34440
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.495501790549393e-05,
+      "loss": 0.9062,
+      "step": 34450
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.495065071185257e-05,
+      "loss": 0.7872,
+      "step": 34460
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.49462835182112e-05,
+      "loss": 0.7374,
+      "step": 34470
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4941916324569834e-05,
+      "loss": 0.8927,
+      "step": 34480
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.493754913092847e-05,
+      "loss": 1.083,
+      "step": 34490
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.49331819372871e-05,
+      "loss": 0.9656,
+      "step": 34500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4928814743645734e-05,
+      "loss": 0.9358,
+      "step": 34510
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.492444755000437e-05,
+      "loss": 0.9888,
+      "step": 34520
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.492008035636301e-05,
+      "loss": 0.972,
+      "step": 34530
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.491571316272163e-05,
+      "loss": 0.8618,
+      "step": 34540
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.491134596908027e-05,
+      "loss": 0.929,
+      "step": 34550
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4906978775438907e-05,
+      "loss": 0.8937,
+      "step": 34560
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.490261158179754e-05,
+      "loss": 0.8863,
+      "step": 34570
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.489824438815617e-05,
+      "loss": 0.9848,
+      "step": 34580
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4893877194514806e-05,
+      "loss": 0.9367,
+      "step": 34590
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.488951000087344e-05,
+      "loss": 1.0042,
+      "step": 34600
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.488514280723207e-05,
+      "loss": 0.9252,
+      "step": 34610
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.488077561359071e-05,
+      "loss": 0.9834,
+      "step": 34620
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.487640841994934e-05,
+      "loss": 0.7942,
+      "step": 34630
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.487204122630798e-05,
+      "loss": 0.8642,
+      "step": 34640
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4867674032666605e-05,
+      "loss": 0.8466,
+      "step": 34650
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4863306839025245e-05,
+      "loss": 0.8247,
+      "step": 34660
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.485893964538388e-05,
+      "loss": 0.94,
+      "step": 34670
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.485457245174251e-05,
+      "loss": 0.8429,
+      "step": 34680
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.485020525810115e-05,
+      "loss": 0.8034,
+      "step": 34690
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.484583806445978e-05,
+      "loss": 0.8194,
+      "step": 34700
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.484147087081842e-05,
+      "loss": 0.9079,
+      "step": 34710
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4837103677177045e-05,
+      "loss": 0.8624,
+      "step": 34720
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4832736483535685e-05,
+      "loss": 0.9443,
+      "step": 34730
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.482836928989431e-05,
+      "loss": 0.8354,
+      "step": 34740
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.482400209625295e-05,
+      "loss": 0.8489,
+      "step": 34750
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4819634902611584e-05,
+      "loss": 0.8837,
+      "step": 34760
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.481526770897022e-05,
+      "loss": 0.8671,
+      "step": 34770
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.481090051532885e-05,
+      "loss": 0.8916,
+      "step": 34780
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4806533321687484e-05,
+      "loss": 0.994,
+      "step": 34790
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4802166128046124e-05,
+      "loss": 0.9229,
+      "step": 34800
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.479779893440475e-05,
+      "loss": 0.9352,
+      "step": 34810
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.479343174076339e-05,
+      "loss": 0.8805,
+      "step": 34820
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4789064547122017e-05,
+      "loss": 0.8819,
+      "step": 34830
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4784697353480657e-05,
+      "loss": 0.8586,
+      "step": 34840
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.478033015983929e-05,
+      "loss": 0.9049,
+      "step": 34850
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.477596296619792e-05,
+      "loss": 1.0219,
+      "step": 34860
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4771595772556556e-05,
+      "loss": 0.9363,
+      "step": 34870
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.476722857891519e-05,
+      "loss": 0.8157,
+      "step": 34880
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.476286138527383e-05,
+      "loss": 0.783,
+      "step": 34890
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4758494191632456e-05,
+      "loss": 0.8574,
+      "step": 34900
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4754126997991096e-05,
+      "loss": 0.9588,
+      "step": 34910
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.474975980434972e-05,
+      "loss": 0.8942,
+      "step": 34920
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.474539261070836e-05,
+      "loss": 0.9236,
+      "step": 34930
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4741025417066995e-05,
+      "loss": 1.0235,
+      "step": 34940
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.473665822342563e-05,
+      "loss": 0.6991,
+      "step": 34950
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.473229102978426e-05,
+      "loss": 0.8242,
+      "step": 34960
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4727923836142895e-05,
+      "loss": 0.8821,
+      "step": 34970
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.472355664250153e-05,
+      "loss": 0.7784,
+      "step": 34980
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.471918944886016e-05,
+      "loss": 0.9724,
+      "step": 34990
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.47148222552188e-05,
+      "loss": 0.8615,
+      "step": 35000
+    },
+    {
+      "epoch": 3.06,
+      "eval_accuracy": 0.5885820968851332,
+      "eval_loss": 0.882483720779419,
+      "eval_runtime": 84.0427,
+      "eval_samples_per_second": 121.093,
+      "eval_steps_per_second": 15.147,
+      "step": 35000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.471045506157743e-05,
+      "loss": 1.0345,
+      "step": 35010
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.470608786793607e-05,
+      "loss": 0.7732,
+      "step": 35020
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.47017206742947e-05,
+      "loss": 0.7744,
+      "step": 35030
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4697353480653334e-05,
+      "loss": 1.058,
+      "step": 35040
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.469298628701197e-05,
+      "loss": 0.8635,
+      "step": 35050
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.46886190933706e-05,
+      "loss": 0.7907,
+      "step": 35060
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4684251899729234e-05,
+      "loss": 0.8665,
+      "step": 35070
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.467988470608787e-05,
+      "loss": 0.9788,
+      "step": 35080
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.467551751244651e-05,
+      "loss": 0.9071,
+      "step": 35090
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.467115031880514e-05,
+      "loss": 0.8321,
+      "step": 35100
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.466678312516377e-05,
+      "loss": 0.9209,
+      "step": 35110
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4662415931522407e-05,
+      "loss": 0.8042,
+      "step": 35120
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.465804873788104e-05,
+      "loss": 0.9322,
+      "step": 35130
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.465368154423967e-05,
+      "loss": 0.8444,
+      "step": 35140
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4649314350598306e-05,
+      "loss": 0.8347,
+      "step": 35150
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.464494715695694e-05,
+      "loss": 0.8465,
+      "step": 35160
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.464057996331557e-05,
+      "loss": 0.8566,
+      "step": 35170
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4636212769674206e-05,
+      "loss": 0.877,
+      "step": 35180
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4631845576032846e-05,
+      "loss": 0.9483,
+      "step": 35190
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.462747838239148e-05,
+      "loss": 0.8888,
+      "step": 35200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.462311118875011e-05,
+      "loss": 1.0432,
+      "step": 35210
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4618743995108745e-05,
+      "loss": 0.873,
+      "step": 35220
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.461437680146738e-05,
+      "loss": 1.0026,
+      "step": 35230
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.461000960782601e-05,
+      "loss": 1.0515,
+      "step": 35240
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4605642414184645e-05,
+      "loss": 0.9022,
+      "step": 35250
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4601275220543285e-05,
+      "loss": 0.8203,
+      "step": 35260
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.459690802690191e-05,
+      "loss": 0.8713,
+      "step": 35270
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.459254083326055e-05,
+      "loss": 0.8287,
+      "step": 35280
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4588173639619185e-05,
+      "loss": 0.9076,
+      "step": 35290
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.458380644597782e-05,
+      "loss": 0.8194,
+      "step": 35300
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.457943925233645e-05,
+      "loss": 0.9893,
+      "step": 35310
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4575072058695084e-05,
+      "loss": 0.8751,
+      "step": 35320
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.457070486505372e-05,
+      "loss": 0.9858,
+      "step": 35330
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.456633767141235e-05,
+      "loss": 0.8483,
+      "step": 35340
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.456197047777099e-05,
+      "loss": 0.7207,
+      "step": 35350
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.455760328412962e-05,
+      "loss": 0.9631,
+      "step": 35360
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.455323609048826e-05,
+      "loss": 0.798,
+      "step": 35370
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.454886889684688e-05,
+      "loss": 0.7873,
+      "step": 35380
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.454450170320552e-05,
+      "loss": 0.801,
+      "step": 35390
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4540134509564157e-05,
+      "loss": 1.0151,
+      "step": 35400
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.453576731592279e-05,
+      "loss": 1.0433,
+      "step": 35410
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.453140012228142e-05,
+      "loss": 0.813,
+      "step": 35420
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4527032928640056e-05,
+      "loss": 0.8712,
+      "step": 35430
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4522665734998696e-05,
+      "loss": 0.882,
+      "step": 35440
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.451829854135732e-05,
+      "loss": 0.94,
+      "step": 35450
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.451393134771596e-05,
+      "loss": 0.8956,
+      "step": 35460
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.450956415407459e-05,
+      "loss": 0.841,
+      "step": 35470
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.450519696043323e-05,
+      "loss": 1.0206,
+      "step": 35480
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.450082976679186e-05,
+      "loss": 0.9371,
+      "step": 35490
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4496462573150495e-05,
+      "loss": 0.9155,
+      "step": 35500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.449209537950913e-05,
+      "loss": 0.9004,
+      "step": 35510
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.448772818586776e-05,
+      "loss": 0.8114,
+      "step": 35520
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.44833609922264e-05,
+      "loss": 0.7819,
+      "step": 35530
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.447899379858503e-05,
+      "loss": 0.9102,
+      "step": 35540
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.447462660494367e-05,
+      "loss": 0.8473,
+      "step": 35550
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4470259411302295e-05,
+      "loss": 0.8636,
+      "step": 35560
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4465892217660935e-05,
+      "loss": 0.8907,
+      "step": 35570
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.446152502401956e-05,
+      "loss": 0.8656,
+      "step": 35580
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.44571578303782e-05,
+      "loss": 0.8956,
+      "step": 35590
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4452790636736834e-05,
+      "loss": 0.9364,
+      "step": 35600
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.444842344309547e-05,
+      "loss": 0.9816,
+      "step": 35610
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.444405624945411e-05,
+      "loss": 0.8587,
+      "step": 35620
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4439689055812734e-05,
+      "loss": 0.9352,
+      "step": 35630
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4435321862171374e-05,
+      "loss": 0.7623,
+      "step": 35640
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.443095466853e-05,
+      "loss": 0.9557,
+      "step": 35650
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.442658747488864e-05,
+      "loss": 1.0131,
+      "step": 35660
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4422220281247267e-05,
+      "loss": 0.9015,
+      "step": 35670
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4417853087605907e-05,
+      "loss": 0.8346,
+      "step": 35680
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.441348589396454e-05,
+      "loss": 0.9611,
+      "step": 35690
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.440911870032317e-05,
+      "loss": 0.9278,
+      "step": 35700
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4404751506681806e-05,
+      "loss": 0.9588,
+      "step": 35710
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.440038431304044e-05,
+      "loss": 0.9484,
+      "step": 35720
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.439601711939908e-05,
+      "loss": 1.0451,
+      "step": 35730
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4391649925757706e-05,
+      "loss": 0.801,
+      "step": 35740
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4387282732116346e-05,
+      "loss": 0.9293,
+      "step": 35750
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.438291553847498e-05,
+      "loss": 0.9039,
+      "step": 35760
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.437854834483361e-05,
+      "loss": 0.8244,
+      "step": 35770
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4374181151192245e-05,
+      "loss": 0.8739,
+      "step": 35780
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.436981395755088e-05,
+      "loss": 0.8162,
+      "step": 35790
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.436544676390951e-05,
+      "loss": 0.9777,
+      "step": 35800
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4361079570268145e-05,
+      "loss": 0.839,
+      "step": 35810
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4356712376626785e-05,
+      "loss": 0.9514,
+      "step": 35820
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.435234518298541e-05,
+      "loss": 0.9178,
+      "step": 35830
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.434797798934405e-05,
+      "loss": 0.7832,
+      "step": 35840
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4343610795702685e-05,
+      "loss": 0.8735,
+      "step": 35850
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.433924360206132e-05,
+      "loss": 0.8604,
+      "step": 35860
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.433487640841995e-05,
+      "loss": 1.109,
+      "step": 35870
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4330509214778584e-05,
+      "loss": 0.8711,
+      "step": 35880
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.432614202113722e-05,
+      "loss": 0.8646,
+      "step": 35890
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.432177482749585e-05,
+      "loss": 0.8587,
+      "step": 35900
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4317407633854484e-05,
+      "loss": 0.8208,
+      "step": 35910
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4313040440213124e-05,
+      "loss": 0.9191,
+      "step": 35920
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.430867324657176e-05,
+      "loss": 0.8553,
+      "step": 35930
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.430430605293039e-05,
+      "loss": 0.8445,
+      "step": 35940
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.429993885928902e-05,
+      "loss": 0.8122,
+      "step": 35950
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4295571665647657e-05,
+      "loss": 0.8396,
+      "step": 35960
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.429120447200629e-05,
+      "loss": 0.8943,
+      "step": 35970
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.428683727836492e-05,
+      "loss": 0.8213,
+      "step": 35980
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4282470084723556e-05,
+      "loss": 0.8706,
+      "step": 35990
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.427810289108219e-05,
+      "loss": 0.8227,
+      "step": 36000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.427373569744083e-05,
+      "loss": 0.9031,
+      "step": 36010
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.426936850379946e-05,
+      "loss": 0.8812,
+      "step": 36020
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4265001310158096e-05,
+      "loss": 1.0123,
+      "step": 36030
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.426063411651673e-05,
+      "loss": 0.8268,
+      "step": 36040
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.425626692287536e-05,
+      "loss": 0.9628,
+      "step": 36050
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4251899729233995e-05,
+      "loss": 0.8805,
+      "step": 36060
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.424753253559263e-05,
+      "loss": 0.7904,
+      "step": 36070
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.424316534195126e-05,
+      "loss": 1.0054,
+      "step": 36080
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4238798148309895e-05,
+      "loss": 0.889,
+      "step": 36090
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4234430954668535e-05,
+      "loss": 0.9051,
+      "step": 36100
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.423006376102716e-05,
+      "loss": 0.9296,
+      "step": 36110
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.42256965673858e-05,
+      "loss": 0.9426,
+      "step": 36120
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4221329373744435e-05,
+      "loss": 0.9493,
+      "step": 36130
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.421696218010307e-05,
+      "loss": 0.916,
+      "step": 36140
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.42125949864617e-05,
+      "loss": 0.8907,
+      "step": 36150
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4208227792820334e-05,
+      "loss": 0.8568,
+      "step": 36160
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4203860599178974e-05,
+      "loss": 0.9123,
+      "step": 36170
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.41994934055376e-05,
+      "loss": 0.9181,
+      "step": 36180
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.419512621189624e-05,
+      "loss": 0.8457,
+      "step": 36190
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.419075901825487e-05,
+      "loss": 0.934,
+      "step": 36200
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.418639182461351e-05,
+      "loss": 0.876,
+      "step": 36210
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.418202463097214e-05,
+      "loss": 0.8449,
+      "step": 36220
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.417765743733077e-05,
+      "loss": 0.9457,
+      "step": 36230
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4173290243689407e-05,
+      "loss": 0.8618,
+      "step": 36240
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.416892305004804e-05,
+      "loss": 0.9077,
+      "step": 36250
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.416455585640668e-05,
+      "loss": 0.942,
+      "step": 36260
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4160188662765306e-05,
+      "loss": 0.995,
+      "step": 36270
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4155821469123946e-05,
+      "loss": 0.7341,
+      "step": 36280
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.415145427548257e-05,
+      "loss": 0.8964,
+      "step": 36290
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.414708708184121e-05,
+      "loss": 0.9073,
+      "step": 36300
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.414271988819984e-05,
+      "loss": 0.8333,
+      "step": 36310
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.413835269455848e-05,
+      "loss": 0.8367,
+      "step": 36320
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.413398550091711e-05,
+      "loss": 0.9353,
+      "step": 36330
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4129618307275745e-05,
+      "loss": 0.8865,
+      "step": 36340
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4125251113634385e-05,
+      "loss": 0.7973,
+      "step": 36350
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.412088391999301e-05,
+      "loss": 0.9568,
+      "step": 36360
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.411651672635165e-05,
+      "loss": 0.9927,
+      "step": 36370
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.411214953271028e-05,
+      "loss": 0.9476,
+      "step": 36380
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.410778233906892e-05,
+      "loss": 0.8401,
+      "step": 36390
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4103415145427545e-05,
+      "loss": 0.7536,
+      "step": 36400
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4099047951786185e-05,
+      "loss": 0.8481,
+      "step": 36410
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.409468075814482e-05,
+      "loss": 1.0072,
+      "step": 36420
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.409031356450345e-05,
+      "loss": 0.978,
+      "step": 36430
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4085946370862084e-05,
+      "loss": 0.7873,
+      "step": 36440
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.408157917722072e-05,
+      "loss": 0.8129,
+      "step": 36450
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.407721198357936e-05,
+      "loss": 0.8795,
+      "step": 36460
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4072844789937984e-05,
+      "loss": 0.8698,
+      "step": 36470
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4068477596296624e-05,
+      "loss": 0.9103,
+      "step": 36480
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.406411040265525e-05,
+      "loss": 1.0156,
+      "step": 36490
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.405974320901389e-05,
+      "loss": 0.9182,
+      "step": 36500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4055376015372523e-05,
+      "loss": 1.0087,
+      "step": 36510
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4051008821731157e-05,
+      "loss": 0.8094,
+      "step": 36520
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.404664162808979e-05,
+      "loss": 0.9239,
+      "step": 36530
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.404227443444842e-05,
+      "loss": 0.862,
+      "step": 36540
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.403790724080706e-05,
+      "loss": 0.8401,
+      "step": 36550
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.403354004716569e-05,
+      "loss": 0.8969,
+      "step": 36560
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.402917285352433e-05,
+      "loss": 0.9551,
+      "step": 36570
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.402480565988296e-05,
+      "loss": 0.8493,
+      "step": 36580
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4020438466241596e-05,
+      "loss": 0.8277,
+      "step": 36590
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.401607127260023e-05,
+      "loss": 0.8707,
+      "step": 36600
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.401170407895886e-05,
+      "loss": 1.0136,
+      "step": 36610
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4007336885317495e-05,
+      "loss": 0.9579,
+      "step": 36620
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.400296969167613e-05,
+      "loss": 0.867,
+      "step": 36630
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.399860249803476e-05,
+      "loss": 0.8497,
+      "step": 36640
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3994235304393395e-05,
+      "loss": 1.0245,
+      "step": 36650
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3989868110752035e-05,
+      "loss": 0.9093,
+      "step": 36660
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.398550091711067e-05,
+      "loss": 0.8243,
+      "step": 36670
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.39811337234693e-05,
+      "loss": 0.8868,
+      "step": 36680
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3976766529827935e-05,
+      "loss": 0.8731,
+      "step": 36690
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.397239933618657e-05,
+      "loss": 0.9336,
+      "step": 36700
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.39680321425452e-05,
+      "loss": 0.8461,
+      "step": 36710
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3963664948903834e-05,
+      "loss": 0.9806,
+      "step": 36720
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.395929775526247e-05,
+      "loss": 0.9746,
+      "step": 36730
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.395493056162111e-05,
+      "loss": 0.9219,
+      "step": 36740
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.395056336797974e-05,
+      "loss": 0.8256,
+      "step": 36750
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3946196174338374e-05,
+      "loss": 0.8811,
+      "step": 36760
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.394182898069701e-05,
+      "loss": 0.9958,
+      "step": 36770
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.393746178705564e-05,
+      "loss": 1.0068,
+      "step": 36780
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3933094593414273e-05,
+      "loss": 0.784,
+      "step": 36790
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.392872739977291e-05,
+      "loss": 0.9068,
+      "step": 36800
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.392436020613154e-05,
+      "loss": 0.8763,
+      "step": 36810
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.391999301249017e-05,
+      "loss": 0.9686,
+      "step": 36820
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.391562581884881e-05,
+      "loss": 0.9457,
+      "step": 36830
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.391125862520744e-05,
+      "loss": 0.8564,
+      "step": 36840
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.390689143156608e-05,
+      "loss": 0.99,
+      "step": 36850
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.390252423792471e-05,
+      "loss": 0.9033,
+      "step": 36860
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3898157044283346e-05,
+      "loss": 0.9307,
+      "step": 36870
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.389378985064198e-05,
+      "loss": 1.0438,
+      "step": 36880
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.388942265700061e-05,
+      "loss": 0.7963,
+      "step": 36890
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3885055463359245e-05,
+      "loss": 0.8291,
+      "step": 36900
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.388068826971788e-05,
+      "loss": 0.8747,
+      "step": 36910
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.387632107607652e-05,
+      "loss": 0.9417,
+      "step": 36920
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3871953882435145e-05,
+      "loss": 0.8939,
+      "step": 36930
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3867586688793785e-05,
+      "loss": 0.8849,
+      "step": 36940
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.386321949515242e-05,
+      "loss": 0.8524,
+      "step": 36950
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.385885230151105e-05,
+      "loss": 0.6721,
+      "step": 36960
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3854485107869685e-05,
+      "loss": 0.8793,
+      "step": 36970
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.385011791422832e-05,
+      "loss": 0.8334,
+      "step": 36980
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.384575072058696e-05,
+      "loss": 0.9656,
+      "step": 36990
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3841383526945584e-05,
+      "loss": 0.8265,
+      "step": 37000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3837016333304224e-05,
+      "loss": 0.8985,
+      "step": 37010
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.383264913966285e-05,
+      "loss": 0.9337,
+      "step": 37020
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.382828194602149e-05,
+      "loss": 0.9311,
+      "step": 37030
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.382391475238012e-05,
+      "loss": 0.8259,
+      "step": 37040
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.381954755873876e-05,
+      "loss": 0.9188,
+      "step": 37050
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.381518036509739e-05,
+      "loss": 0.8173,
+      "step": 37060
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3810813171456023e-05,
+      "loss": 0.978,
+      "step": 37070
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3806445977814663e-05,
+      "loss": 0.8295,
+      "step": 37080
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.380207878417329e-05,
+      "loss": 0.7591,
+      "step": 37090
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.379771159053193e-05,
+      "loss": 0.9092,
+      "step": 37100
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3793344396890556e-05,
+      "loss": 0.7545,
+      "step": 37110
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3788977203249196e-05,
+      "loss": 0.9288,
+      "step": 37120
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.378461000960782e-05,
+      "loss": 0.9138,
+      "step": 37130
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.378024281596646e-05,
+      "loss": 0.8662,
+      "step": 37140
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3775875622325096e-05,
+      "loss": 0.7862,
+      "step": 37150
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.377150842868373e-05,
+      "loss": 0.8301,
+      "step": 37160
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.376714123504236e-05,
+      "loss": 0.9223,
+      "step": 37170
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3762774041400995e-05,
+      "loss": 0.9533,
+      "step": 37180
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3758406847759635e-05,
+      "loss": 0.8741,
+      "step": 37190
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.375403965411826e-05,
+      "loss": 0.957,
+      "step": 37200
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.37496724604769e-05,
+      "loss": 0.8571,
+      "step": 37210
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.374530526683553e-05,
+      "loss": 0.8671,
+      "step": 37220
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.374093807319417e-05,
+      "loss": 0.8366,
+      "step": 37230
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.37365708795528e-05,
+      "loss": 0.8003,
+      "step": 37240
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3732203685911435e-05,
+      "loss": 0.9867,
+      "step": 37250
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.372783649227007e-05,
+      "loss": 1.0672,
+      "step": 37260
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.37234692986287e-05,
+      "loss": 0.9091,
+      "step": 37270
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.371910210498734e-05,
+      "loss": 0.8926,
+      "step": 37280
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.371473491134597e-05,
+      "loss": 0.8997,
+      "step": 37290
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.371036771770461e-05,
+      "loss": 0.8784,
+      "step": 37300
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3706000524063234e-05,
+      "loss": 0.9578,
+      "step": 37310
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3701633330421874e-05,
+      "loss": 1.0252,
+      "step": 37320
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.369726613678051e-05,
+      "loss": 0.9283,
+      "step": 37330
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.369289894313914e-05,
+      "loss": 0.8785,
+      "step": 37340
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3688531749497773e-05,
+      "loss": 0.9622,
+      "step": 37350
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.368416455585641e-05,
+      "loss": 0.8911,
+      "step": 37360
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.367979736221504e-05,
+      "loss": 0.9444,
+      "step": 37370
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.367543016857367e-05,
+      "loss": 0.9206,
+      "step": 37380
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.367106297493231e-05,
+      "loss": 0.8737,
+      "step": 37390
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3666695781290946e-05,
+      "loss": 0.9084,
+      "step": 37400
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.366232858764958e-05,
+      "loss": 0.8716,
+      "step": 37410
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.365796139400821e-05,
+      "loss": 0.8926,
+      "step": 37420
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3653594200366846e-05,
+      "loss": 0.991,
+      "step": 37430
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.364922700672548e-05,
+      "loss": 0.981,
+      "step": 37440
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.364485981308411e-05,
+      "loss": 0.8926,
+      "step": 37450
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3640492619442745e-05,
+      "loss": 0.8259,
+      "step": 37460
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.363612542580138e-05,
+      "loss": 0.8469,
+      "step": 37470
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.363175823216002e-05,
+      "loss": 0.9116,
+      "step": 37480
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.362739103851865e-05,
+      "loss": 0.843,
+      "step": 37490
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3623023844877285e-05,
+      "loss": 0.9118,
+      "step": 37500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.361865665123592e-05,
+      "loss": 0.9026,
+      "step": 37510
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.361428945759455e-05,
+      "loss": 0.8457,
+      "step": 37520
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3609922263953185e-05,
+      "loss": 0.8222,
+      "step": 37530
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.360555507031182e-05,
+      "loss": 0.8583,
+      "step": 37540
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.360118787667045e-05,
+      "loss": 0.7543,
+      "step": 37550
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3596820683029084e-05,
+      "loss": 0.8891,
+      "step": 37560
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.359245348938772e-05,
+      "loss": 0.8562,
+      "step": 37570
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.358808629574636e-05,
+      "loss": 0.875,
+      "step": 37580
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.358371910210499e-05,
+      "loss": 1.0304,
+      "step": 37590
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3579351908463624e-05,
+      "loss": 0.824,
+      "step": 37600
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.357498471482226e-05,
+      "loss": 0.9239,
+      "step": 37610
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.357061752118089e-05,
+      "loss": 0.8727,
+      "step": 37620
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3566250327539523e-05,
+      "loss": 0.9683,
+      "step": 37630
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.356188313389816e-05,
+      "loss": 0.8903,
+      "step": 37640
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.35575159402568e-05,
+      "loss": 0.7583,
+      "step": 37650
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.355314874661542e-05,
+      "loss": 0.8851,
+      "step": 37660
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.354878155297406e-05,
+      "loss": 1.0015,
+      "step": 37670
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3544414359332696e-05,
+      "loss": 0.8163,
+      "step": 37680
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.354004716569133e-05,
+      "loss": 0.8691,
+      "step": 37690
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.353567997204996e-05,
+      "loss": 0.9769,
+      "step": 37700
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3531312778408596e-05,
+      "loss": 0.8945,
+      "step": 37710
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.352694558476723e-05,
+      "loss": 0.8299,
+      "step": 37720
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.352257839112586e-05,
+      "loss": 0.8719,
+      "step": 37730
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.35182111974845e-05,
+      "loss": 0.8188,
+      "step": 37740
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.351384400384313e-05,
+      "loss": 0.7876,
+      "step": 37750
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.350947681020177e-05,
+      "loss": 0.9175,
+      "step": 37760
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3505109616560395e-05,
+      "loss": 0.7864,
+      "step": 37770
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3500742422919035e-05,
+      "loss": 0.9697,
+      "step": 37780
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.349637522927767e-05,
+      "loss": 0.9064,
+      "step": 37790
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.34920080356363e-05,
+      "loss": 0.9331,
+      "step": 37800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.348764084199494e-05,
+      "loss": 0.9566,
+      "step": 37810
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.348327364835357e-05,
+      "loss": 0.8627,
+      "step": 37820
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.347890645471221e-05,
+      "loss": 0.8814,
+      "step": 37830
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3474539261070834e-05,
+      "loss": 0.8672,
+      "step": 37840
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3470172067429474e-05,
+      "loss": 0.9348,
+      "step": 37850
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.34658048737881e-05,
+      "loss": 0.914,
+      "step": 37860
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.346143768014674e-05,
+      "loss": 0.8979,
+      "step": 37870
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3457070486505374e-05,
+      "loss": 0.9306,
+      "step": 37880
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.345270329286401e-05,
+      "loss": 0.9296,
+      "step": 37890
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.344833609922264e-05,
+      "loss": 0.8789,
+      "step": 37900
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3443968905581273e-05,
+      "loss": 0.7948,
+      "step": 37910
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3439601711939913e-05,
+      "loss": 0.7723,
+      "step": 37920
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.343523451829854e-05,
+      "loss": 0.9166,
+      "step": 37930
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.343086732465718e-05,
+      "loss": 0.7961,
+      "step": 37940
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3426500131015806e-05,
+      "loss": 0.8378,
+      "step": 37950
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3422132937374446e-05,
+      "loss": 0.9409,
+      "step": 37960
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.341776574373307e-05,
+      "loss": 1.0128,
+      "step": 37970
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.341339855009171e-05,
+      "loss": 0.9386,
+      "step": 37980
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3409031356450346e-05,
+      "loss": 0.8847,
+      "step": 37990
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.340466416280898e-05,
+      "loss": 0.8472,
+      "step": 38000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.340029696916762e-05,
+      "loss": 0.8086,
+      "step": 38010
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3395929775526245e-05,
+      "loss": 0.8825,
+      "step": 38020
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3391562581884885e-05,
+      "loss": 0.8778,
+      "step": 38030
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.338719538824351e-05,
+      "loss": 1.017,
+      "step": 38040
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.338282819460215e-05,
+      "loss": 0.768,
+      "step": 38050
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3378461000960785e-05,
+      "loss": 1.0126,
+      "step": 38060
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.337409380731942e-05,
+      "loss": 0.7972,
+      "step": 38070
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.336972661367805e-05,
+      "loss": 0.8973,
+      "step": 38080
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3365359420036685e-05,
+      "loss": 0.9751,
+      "step": 38090
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.336099222639532e-05,
+      "loss": 0.8613,
+      "step": 38100
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.335662503275395e-05,
+      "loss": 0.8612,
+      "step": 38110
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.335225783911259e-05,
+      "loss": 0.8904,
+      "step": 38120
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.334789064547122e-05,
+      "loss": 0.8445,
+      "step": 38130
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.334352345182986e-05,
+      "loss": 0.9083,
+      "step": 38140
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.333915625818849e-05,
+      "loss": 0.9217,
+      "step": 38150
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3334789064547124e-05,
+      "loss": 0.9023,
+      "step": 38160
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.333042187090576e-05,
+      "loss": 0.9067,
+      "step": 38170
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.332605467726439e-05,
+      "loss": 0.8068,
+      "step": 38180
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3321687483623023e-05,
+      "loss": 0.805,
+      "step": 38190
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.331732028998166e-05,
+      "loss": 0.8529,
+      "step": 38200
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.33129530963403e-05,
+      "loss": 0.957,
+      "step": 38210
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.330858590269893e-05,
+      "loss": 0.7811,
+      "step": 38220
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.330421870905756e-05,
+      "loss": 0.9307,
+      "step": 38230
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3299851515416196e-05,
+      "loss": 0.8822,
+      "step": 38240
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.329548432177483e-05,
+      "loss": 0.8742,
+      "step": 38250
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.329111712813346e-05,
+      "loss": 0.9279,
+      "step": 38260
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3286749934492096e-05,
+      "loss": 0.8408,
+      "step": 38270
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.328238274085073e-05,
+      "loss": 0.9177,
+      "step": 38280
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.327801554720936e-05,
+      "loss": 0.8692,
+      "step": 38290
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3273648353567995e-05,
+      "loss": 0.8729,
+      "step": 38300
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3269281159926635e-05,
+      "loss": 1.0512,
+      "step": 38310
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.326491396628527e-05,
+      "loss": 0.7673,
+      "step": 38320
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.32605467726439e-05,
+      "loss": 0.8329,
+      "step": 38330
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3256179579002535e-05,
+      "loss": 0.9095,
+      "step": 38340
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.325181238536117e-05,
+      "loss": 0.8141,
+      "step": 38350
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.32474451917198e-05,
+      "loss": 0.8884,
+      "step": 38360
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3243077998078435e-05,
+      "loss": 0.9357,
+      "step": 38370
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.323871080443707e-05,
+      "loss": 0.8722,
+      "step": 38380
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.32343436107957e-05,
+      "loss": 1.0003,
+      "step": 38390
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.322997641715434e-05,
+      "loss": 0.7806,
+      "step": 38400
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3225609223512974e-05,
+      "loss": 0.7441,
+      "step": 38410
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.322124202987161e-05,
+      "loss": 0.8462,
+      "step": 38420
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.321687483623024e-05,
+      "loss": 0.9057,
+      "step": 38430
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3212507642588874e-05,
+      "loss": 0.7867,
+      "step": 38440
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.320814044894751e-05,
+      "loss": 0.9724,
+      "step": 38450
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.320377325530614e-05,
+      "loss": 0.9636,
+      "step": 38460
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.319940606166478e-05,
+      "loss": 0.9747,
+      "step": 38470
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.319503886802341e-05,
+      "loss": 0.9546,
+      "step": 38480
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.319067167438205e-05,
+      "loss": 0.7638,
+      "step": 38490
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.318630448074067e-05,
+      "loss": 0.8517,
+      "step": 38500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.318193728709931e-05,
+      "loss": 0.9352,
+      "step": 38510
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3177570093457946e-05,
+      "loss": 0.7845,
+      "step": 38520
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.317320289981658e-05,
+      "loss": 0.8812,
+      "step": 38530
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.316883570617521e-05,
+      "loss": 0.8305,
+      "step": 38540
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3164468512533846e-05,
+      "loss": 0.8574,
+      "step": 38550
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3160101318892486e-05,
+      "loss": 0.9807,
+      "step": 38560
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.315573412525111e-05,
+      "loss": 0.9211,
+      "step": 38570
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.315136693160975e-05,
+      "loss": 0.8707,
+      "step": 38580
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.314699973796838e-05,
+      "loss": 0.9616,
+      "step": 38590
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.314263254432702e-05,
+      "loss": 0.9155,
+      "step": 38600
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.313826535068565e-05,
+      "loss": 0.8248,
+      "step": 38610
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3133898157044285e-05,
+      "loss": 0.9311,
+      "step": 38620
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.312953096340292e-05,
+      "loss": 0.9414,
+      "step": 38630
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.312516376976155e-05,
+      "loss": 0.8742,
+      "step": 38640
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.312079657612019e-05,
+      "loss": 0.8834,
+      "step": 38650
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.311642938247882e-05,
+      "loss": 0.9586,
+      "step": 38660
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.311206218883746e-05,
+      "loss": 0.9079,
+      "step": 38670
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3107694995196084e-05,
+      "loss": 0.8603,
+      "step": 38680
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3103327801554724e-05,
+      "loss": 0.9403,
+      "step": 38690
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.309896060791335e-05,
+      "loss": 0.8825,
+      "step": 38700
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.309459341427199e-05,
+      "loss": 0.8225,
+      "step": 38710
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3090226220630624e-05,
+      "loss": 0.9175,
+      "step": 38720
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.308585902698926e-05,
+      "loss": 0.8398,
+      "step": 38730
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.30814918333479e-05,
+      "loss": 0.9789,
+      "step": 38740
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3077124639706523e-05,
+      "loss": 0.9611,
+      "step": 38750
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3072757446065163e-05,
+      "loss": 0.8102,
+      "step": 38760
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.306839025242379e-05,
+      "loss": 0.7973,
+      "step": 38770
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.306402305878243e-05,
+      "loss": 0.9467,
+      "step": 38780
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3059655865141056e-05,
+      "loss": 0.8624,
+      "step": 38790
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3055288671499696e-05,
+      "loss": 0.8698,
+      "step": 38800
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.305092147785833e-05,
+      "loss": 0.889,
+      "step": 38810
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.304655428421696e-05,
+      "loss": 0.8917,
+      "step": 38820
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3042187090575596e-05,
+      "loss": 0.9046,
+      "step": 38830
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.303781989693423e-05,
+      "loss": 0.8106,
+      "step": 38840
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.303345270329287e-05,
+      "loss": 1.0245,
+      "step": 38850
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3029085509651496e-05,
+      "loss": 0.8319,
+      "step": 38860
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3024718316010135e-05,
+      "loss": 0.994,
+      "step": 38870
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.302035112236877e-05,
+      "loss": 0.8914,
+      "step": 38880
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.30159839287274e-05,
+      "loss": 0.9563,
+      "step": 38890
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3011616735086035e-05,
+      "loss": 0.8352,
+      "step": 38900
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.300724954144467e-05,
+      "loss": 0.8798,
+      "step": 38910
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.30028823478033e-05,
+      "loss": 0.8841,
+      "step": 38920
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2998515154161935e-05,
+      "loss": 0.9059,
+      "step": 38930
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2994147960520575e-05,
+      "loss": 0.8982,
+      "step": 38940
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.29897807668792e-05,
+      "loss": 0.8473,
+      "step": 38950
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.298541357323784e-05,
+      "loss": 0.8248,
+      "step": 38960
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2981046379596474e-05,
+      "loss": 0.925,
+      "step": 38970
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.297667918595511e-05,
+      "loss": 0.7976,
+      "step": 38980
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.297231199231374e-05,
+      "loss": 0.8149,
+      "step": 38990
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2967944798672374e-05,
+      "loss": 0.9039,
+      "step": 39000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.296357760503101e-05,
+      "loss": 0.8951,
+      "step": 39010
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.295921041138964e-05,
+      "loss": 0.8596,
+      "step": 39020
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2954843217748274e-05,
+      "loss": 0.9184,
+      "step": 39030
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.295047602410691e-05,
+      "loss": 0.7691,
+      "step": 39040
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.294610883046555e-05,
+      "loss": 0.9152,
+      "step": 39050
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.294174163682418e-05,
+      "loss": 0.9746,
+      "step": 39060
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.293737444318281e-05,
+      "loss": 0.8897,
+      "step": 39070
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2933007249541446e-05,
+      "loss": 0.8328,
+      "step": 39080
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.292864005590008e-05,
+      "loss": 0.9095,
+      "step": 39090
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.292427286225871e-05,
+      "loss": 0.9215,
+      "step": 39100
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2919905668617346e-05,
+      "loss": 0.8609,
+      "step": 39110
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.291553847497598e-05,
+      "loss": 0.912,
+      "step": 39120
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.291117128133462e-05,
+      "loss": 0.937,
+      "step": 39130
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.290680408769325e-05,
+      "loss": 0.9144,
+      "step": 39140
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2902436894051885e-05,
+      "loss": 0.9416,
+      "step": 39150
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.289806970041052e-05,
+      "loss": 0.9167,
+      "step": 39160
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.289370250676915e-05,
+      "loss": 0.9428,
+      "step": 39170
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2889335313127785e-05,
+      "loss": 0.8021,
+      "step": 39180
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.288496811948642e-05,
+      "loss": 0.8767,
+      "step": 39190
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.288060092584505e-05,
+      "loss": 0.9318,
+      "step": 39200
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2876233732203685e-05,
+      "loss": 0.9166,
+      "step": 39210
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2871866538562325e-05,
+      "loss": 0.7278,
+      "step": 39220
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.286749934492095e-05,
+      "loss": 0.9183,
+      "step": 39230
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.286313215127959e-05,
+      "loss": 0.8492,
+      "step": 39240
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2858764957638224e-05,
+      "loss": 0.8909,
+      "step": 39250
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.285439776399686e-05,
+      "loss": 0.684,
+      "step": 39260
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.285003057035549e-05,
+      "loss": 0.9357,
+      "step": 39270
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2845663376714124e-05,
+      "loss": 0.8707,
+      "step": 39280
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2841296183072764e-05,
+      "loss": 0.7633,
+      "step": 39290
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.283692898943139e-05,
+      "loss": 0.8987,
+      "step": 39300
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.283256179579003e-05,
+      "loss": 0.9131,
+      "step": 39310
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.282819460214866e-05,
+      "loss": 0.8101,
+      "step": 39320
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.28238274085073e-05,
+      "loss": 0.7963,
+      "step": 39330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.281946021486593e-05,
+      "loss": 0.766,
+      "step": 39340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.281509302122456e-05,
+      "loss": 0.7346,
+      "step": 39350
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2810725827583196e-05,
+      "loss": 0.9294,
+      "step": 39360
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.280635863394183e-05,
+      "loss": 0.8245,
+      "step": 39370
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.280199144030047e-05,
+      "loss": 1.1549,
+      "step": 39380
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2797624246659096e-05,
+      "loss": 0.8841,
+      "step": 39390
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2793257053017736e-05,
+      "loss": 0.8597,
+      "step": 39400
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.278888985937636e-05,
+      "loss": 0.917,
+      "step": 39410
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2784522665735e-05,
+      "loss": 0.8988,
+      "step": 39420
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2780155472093636e-05,
+      "loss": 0.8952,
+      "step": 39430
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.277578827845227e-05,
+      "loss": 0.8713,
+      "step": 39440
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.27714210848109e-05,
+      "loss": 0.7879,
+      "step": 39450
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2767053891169535e-05,
+      "loss": 0.8459,
+      "step": 39460
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2762686697528175e-05,
+      "loss": 0.9478,
+      "step": 39470
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.27583195038868e-05,
+      "loss": 0.8545,
+      "step": 39480
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.275395231024544e-05,
+      "loss": 0.7723,
+      "step": 39490
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.274958511660407e-05,
+      "loss": 0.9551,
+      "step": 39500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.274521792296271e-05,
+      "loss": 0.867,
+      "step": 39510
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2740850729321334e-05,
+      "loss": 0.8125,
+      "step": 39520
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2736483535679974e-05,
+      "loss": 0.8484,
+      "step": 39530
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.273211634203861e-05,
+      "loss": 0.8548,
+      "step": 39540
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.272774914839724e-05,
+      "loss": 1.0267,
+      "step": 39550
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.272338195475588e-05,
+      "loss": 0.8919,
+      "step": 39560
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.271901476111451e-05,
+      "loss": 0.9334,
+      "step": 39570
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.271464756747315e-05,
+      "loss": 0.8413,
+      "step": 39580
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2710280373831774e-05,
+      "loss": 0.8609,
+      "step": 39590
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2705913180190414e-05,
+      "loss": 0.9368,
+      "step": 39600
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.270154598654904e-05,
+      "loss": 0.936,
+      "step": 39610
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.269717879290768e-05,
+      "loss": 0.9436,
+      "step": 39620
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.269281159926631e-05,
+      "loss": 0.8715,
+      "step": 39630
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2688444405624946e-05,
+      "loss": 0.8315,
+      "step": 39640
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.268407721198358e-05,
+      "loss": 0.796,
+      "step": 39650
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.267971001834221e-05,
+      "loss": 0.8514,
+      "step": 39660
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.267534282470085e-05,
+      "loss": 0.9298,
+      "step": 39670
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.267097563105948e-05,
+      "loss": 0.9868,
+      "step": 39680
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.266660843741812e-05,
+      "loss": 0.8787,
+      "step": 39690
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.266224124377675e-05,
+      "loss": 0.8319,
+      "step": 39700
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2657874050135386e-05,
+      "loss": 0.8805,
+      "step": 39710
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.265350685649402e-05,
+      "loss": 0.9283,
+      "step": 39720
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.264913966285265e-05,
+      "loss": 0.8665,
+      "step": 39730
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2644772469211285e-05,
+      "loss": 0.886,
+      "step": 39740
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.264040527556992e-05,
+      "loss": 0.9228,
+      "step": 39750
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.263603808192856e-05,
+      "loss": 1.0164,
+      "step": 39760
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2631670888287185e-05,
+      "loss": 0.8293,
+      "step": 39770
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2627303694645825e-05,
+      "loss": 0.8743,
+      "step": 39780
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.262293650100446e-05,
+      "loss": 0.917,
+      "step": 39790
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.261856930736309e-05,
+      "loss": 0.9465,
+      "step": 39800
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2614202113721724e-05,
+      "loss": 0.8184,
+      "step": 39810
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.260983492008036e-05,
+      "loss": 0.8406,
+      "step": 39820
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.260546772643899e-05,
+      "loss": 0.8139,
+      "step": 39830
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2601100532797624e-05,
+      "loss": 0.7886,
+      "step": 39840
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.259673333915626e-05,
+      "loss": 0.7649,
+      "step": 39850
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.259236614551489e-05,
+      "loss": 0.8103,
+      "step": 39860
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.258799895187353e-05,
+      "loss": 0.8564,
+      "step": 39870
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2583631758232164e-05,
+      "loss": 0.8123,
+      "step": 39880
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.25792645645908e-05,
+      "loss": 1.0089,
+      "step": 39890
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.257489737094943e-05,
+      "loss": 0.8636,
+      "step": 39900
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.257053017730806e-05,
+      "loss": 0.8431,
+      "step": 39910
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2566162983666696e-05,
+      "loss": 0.8657,
+      "step": 39920
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.256179579002533e-05,
+      "loss": 0.8181,
+      "step": 39930
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.255742859638396e-05,
+      "loss": 0.8357,
+      "step": 39940
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.25530614027426e-05,
+      "loss": 0.9168,
+      "step": 39950
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2548694209101236e-05,
+      "loss": 0.9096,
+      "step": 39960
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.254432701545987e-05,
+      "loss": 0.9179,
+      "step": 39970
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.25399598218185e-05,
+      "loss": 0.9418,
+      "step": 39980
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2535592628177136e-05,
+      "loss": 0.8835,
+      "step": 39990
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.253122543453577e-05,
+      "loss": 0.8974,
+      "step": 40000
+    },
+    {
+      "epoch": 3.49,
+      "eval_accuracy": 0.5760047165176378,
+      "eval_loss": 0.8895514011383057,
+      "eval_runtime": 84.0745,
+      "eval_samples_per_second": 121.047,
+      "eval_steps_per_second": 15.141,
+      "step": 40000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 114490,
+  "num_train_epochs": 10,
+  "save_steps": 5000,
+  "total_flos": 8.448609560044954e+16,
+  "trial_name": null,
+  "trial_params": null
+}