diff --git "a/sft/checkpoint-14015/trainer_state.json" "b/sft/checkpoint-14015/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/sft/checkpoint-14015/trainer_state.json"
@@ -0,0 +1,9840 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 14015,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0028540849090260435,
+      "grad_norm": 7.177610874176025,
+      "learning_rate": 5.707762557077626e-07,
+      "loss": 5.7328,
+      "step": 10
+    },
+    {
+      "epoch": 0.005708169818052087,
+      "grad_norm": 7.953944206237793,
+      "learning_rate": 1.1415525114155251e-06,
+      "loss": 5.4553,
+      "step": 20
+    },
+    {
+      "epoch": 0.00856225472707813,
+      "grad_norm": 7.924602031707764,
+      "learning_rate": 1.7123287671232877e-06,
+      "loss": 5.6662,
+      "step": 30
+    },
+    {
+      "epoch": 0.011416339636104174,
+      "grad_norm": 8.706378936767578,
+      "learning_rate": 2.2831050228310503e-06,
+      "loss": 5.503,
+      "step": 40
+    },
+    {
+      "epoch": 0.014270424545130217,
+      "grad_norm": 8.828363418579102,
+      "learning_rate": 2.853881278538813e-06,
+      "loss": 5.2946,
+      "step": 50
+    },
+    {
+      "epoch": 0.01712450945415626,
+      "grad_norm": 11.124424934387207,
+      "learning_rate": 3.4246575342465754e-06,
+      "loss": 4.9995,
+      "step": 60
+    },
+    {
+      "epoch": 0.019978594363182306,
+      "grad_norm": 16.35590934753418,
+      "learning_rate": 3.995433789954338e-06,
+      "loss": 4.4733,
+      "step": 70
+    },
+    {
+      "epoch": 0.022832679272208348,
+      "grad_norm": 9.050875663757324,
+      "learning_rate": 4.566210045662101e-06,
+      "loss": 3.389,
+      "step": 80
+    },
+    {
+      "epoch": 0.025686764181234393,
+      "grad_norm": 10.061305046081543,
+      "learning_rate": 5.136986301369863e-06,
+      "loss": 2.475,
+      "step": 90
+    },
+    {
+      "epoch": 0.028540849090260435,
+      "grad_norm": 4.466187000274658,
+      "learning_rate": 5.707762557077626e-06,
+      "loss": 1.4959,
+      "step": 100
+    },
+    {
+      "epoch": 0.03139493399928648,
+      "grad_norm": 3.469721794128418,
+      "learning_rate": 6.278538812785388e-06,
+      "loss": 1.1382,
+      "step": 110
+    },
+    {
+      "epoch": 0.03424901890831252,
+      "grad_norm": 5.091841220855713,
+      "learning_rate": 6.849315068493151e-06,
+      "loss": 0.8312,
+      "step": 120
+    },
+    {
+      "epoch": 0.03710310381733856,
+      "grad_norm": 2.3277525901794434,
+      "learning_rate": 7.4200913242009134e-06,
+      "loss": 0.7045,
+      "step": 130
+    },
+    {
+      "epoch": 0.03995718872636461,
+      "grad_norm": 2.6082115173339844,
+      "learning_rate": 7.990867579908676e-06,
+      "loss": 0.7053,
+      "step": 140
+    },
+    {
+      "epoch": 0.042811273635390654,
+      "grad_norm": 2.5840566158294678,
+      "learning_rate": 8.561643835616438e-06,
+      "loss": 0.5612,
+      "step": 150
+    },
+    {
+      "epoch": 0.045665358544416695,
+      "grad_norm": 1.6896095275878906,
+      "learning_rate": 9.132420091324201e-06,
+      "loss": 0.536,
+      "step": 160
+    },
+    {
+      "epoch": 0.04851944345344274,
+      "grad_norm": 4.644432067871094,
+      "learning_rate": 9.703196347031963e-06,
+      "loss": 0.4792,
+      "step": 170
+    },
+    {
+      "epoch": 0.051373528362468786,
+      "grad_norm": 2.0224084854125977,
+      "learning_rate": 1.0273972602739726e-05,
+      "loss": 0.4958,
+      "step": 180
+    },
+    {
+      "epoch": 0.05422761327149483,
+      "grad_norm": 2.7462992668151855,
+      "learning_rate": 1.0844748858447488e-05,
+      "loss": 0.5412,
+      "step": 190
+    },
+    {
+      "epoch": 0.05708169818052087,
+      "grad_norm": 3.044191598892212,
+      "learning_rate": 1.1415525114155251e-05,
+      "loss": 0.5006,
+      "step": 200
+    },
+    {
+      "epoch": 0.05993578308954691,
+      "grad_norm": 2.856959819793701,
+      "learning_rate": 1.1986301369863013e-05,
+      "loss": 0.4759,
+      "step": 210
+    },
+    {
+      "epoch": 0.06278986799857296,
+      "grad_norm": 1.736061453819275,
+      "learning_rate": 1.2557077625570777e-05,
+      "loss": 0.4691,
+      "step": 220
+    },
+    {
+      "epoch": 0.065643952907599,
+      "grad_norm": 1.445332646369934,
+      "learning_rate": 1.312785388127854e-05,
+      "loss": 0.4628,
+      "step": 230
+    },
+    {
+      "epoch": 0.06849803781662504,
+      "grad_norm": 2.229682683944702,
+      "learning_rate": 1.3698630136986302e-05,
+      "loss": 0.4981,
+      "step": 240
+    },
+    {
+      "epoch": 0.07135212272565108,
+      "grad_norm": 4.1906232833862305,
+      "learning_rate": 1.4269406392694065e-05,
+      "loss": 0.4352,
+      "step": 250
+    },
+    {
+      "epoch": 0.07420620763467713,
+      "grad_norm": 3.204361915588379,
+      "learning_rate": 1.4840182648401827e-05,
+      "loss": 0.4872,
+      "step": 260
+    },
+    {
+      "epoch": 0.07706029254370317,
+      "grad_norm": 2.531733989715576,
+      "learning_rate": 1.541095890410959e-05,
+      "loss": 0.4652,
+      "step": 270
+    },
+    {
+      "epoch": 0.07991437745272922,
+      "grad_norm": 2.0460851192474365,
+      "learning_rate": 1.5981735159817352e-05,
+      "loss": 0.5005,
+      "step": 280
+    },
+    {
+      "epoch": 0.08276846236175527,
+      "grad_norm": 3.3922574520111084,
+      "learning_rate": 1.6552511415525115e-05,
+      "loss": 0.4896,
+      "step": 290
+    },
+    {
+      "epoch": 0.08562254727078131,
+      "grad_norm": 2.1291298866271973,
+      "learning_rate": 1.7123287671232875e-05,
+      "loss": 0.4396,
+      "step": 300
+    },
+    {
+      "epoch": 0.08847663217980735,
+      "grad_norm": 3.9003682136535645,
+      "learning_rate": 1.769406392694064e-05,
+      "loss": 0.467,
+      "step": 310
+    },
+    {
+      "epoch": 0.09133071708883339,
+      "grad_norm": 3.01649808883667,
+      "learning_rate": 1.8264840182648402e-05,
+      "loss": 0.4752,
+      "step": 320
+    },
+    {
+      "epoch": 0.09418480199785943,
+      "grad_norm": 1.4664050340652466,
+      "learning_rate": 1.8835616438356166e-05,
+      "loss": 0.4519,
+      "step": 330
+    },
+    {
+      "epoch": 0.09703888690688547,
+      "grad_norm": 1.8165019750595093,
+      "learning_rate": 1.9406392694063926e-05,
+      "loss": 0.4778,
+      "step": 340
+    },
+    {
+      "epoch": 0.09989297181591153,
+      "grad_norm": 1.9969810247421265,
+      "learning_rate": 1.997716894977169e-05,
+      "loss": 0.446,
+      "step": 350
+    },
+    {
+      "epoch": 0.10274705672493757,
+      "grad_norm": 1.4795584678649902,
+      "learning_rate": 2.0547945205479453e-05,
+      "loss": 0.4847,
+      "step": 360
+    },
+    {
+      "epoch": 0.10560114163396361,
+      "grad_norm": 1.3624811172485352,
+      "learning_rate": 2.1118721461187216e-05,
+      "loss": 0.4902,
+      "step": 370
+    },
+    {
+      "epoch": 0.10845522654298966,
+      "grad_norm": 2.039226770401001,
+      "learning_rate": 2.1689497716894976e-05,
+      "loss": 0.4595,
+      "step": 380
+    },
+    {
+      "epoch": 0.1113093114520157,
+      "grad_norm": 1.597818374633789,
+      "learning_rate": 2.226027397260274e-05,
+      "loss": 0.4481,
+      "step": 390
+    },
+    {
+      "epoch": 0.11416339636104174,
+      "grad_norm": 1.184199571609497,
+      "learning_rate": 2.2831050228310503e-05,
+      "loss": 0.4659,
+      "step": 400
+    },
+    {
+      "epoch": 0.11701748127006778,
+      "grad_norm": 2.6335010528564453,
+      "learning_rate": 2.3401826484018266e-05,
+      "loss": 0.4932,
+      "step": 410
+    },
+    {
+      "epoch": 0.11987156617909382,
+      "grad_norm": 1.4489330053329468,
+      "learning_rate": 2.3972602739726026e-05,
+      "loss": 0.4646,
+      "step": 420
+    },
+    {
+      "epoch": 0.12272565108811988,
+      "grad_norm": 2.0063445568084717,
+      "learning_rate": 2.454337899543379e-05,
+      "loss": 0.4544,
+      "step": 430
+    },
+    {
+      "epoch": 0.12557973599714592,
+      "grad_norm": 2.1495394706726074,
+      "learning_rate": 2.5114155251141553e-05,
+      "loss": 0.4479,
+      "step": 440
+    },
+    {
+      "epoch": 0.12843382090617195,
+      "grad_norm": 1.3138567209243774,
+      "learning_rate": 2.568493150684932e-05,
+      "loss": 0.4727,
+      "step": 450
+    },
+    {
+      "epoch": 0.131287905815198,
+      "grad_norm": 1.474777340888977,
+      "learning_rate": 2.625570776255708e-05,
+      "loss": 0.4613,
+      "step": 460
+    },
+    {
+      "epoch": 0.13414199072422406,
+      "grad_norm": 2.4200029373168945,
+      "learning_rate": 2.682648401826484e-05,
+      "loss": 0.4794,
+      "step": 470
+    },
+    {
+      "epoch": 0.1369960756332501,
+      "grad_norm": 1.476733922958374,
+      "learning_rate": 2.7397260273972603e-05,
+      "loss": 0.4764,
+      "step": 480
+    },
+    {
+      "epoch": 0.13985016054227614,
+      "grad_norm": 4.004658222198486,
+      "learning_rate": 2.796803652968037e-05,
+      "loss": 0.4796,
+      "step": 490
+    },
+    {
+      "epoch": 0.14270424545130217,
+      "grad_norm": 1.8049726486206055,
+      "learning_rate": 2.853881278538813e-05,
+      "loss": 0.4883,
+      "step": 500
+    },
+    {
+      "epoch": 0.14555833036032823,
+      "grad_norm": 1.7194474935531616,
+      "learning_rate": 2.910958904109589e-05,
+      "loss": 0.4749,
+      "step": 510
+    },
+    {
+      "epoch": 0.14841241526935425,
+      "grad_norm": 1.2992069721221924,
+      "learning_rate": 2.9680365296803654e-05,
+      "loss": 0.448,
+      "step": 520
+    },
+    {
+      "epoch": 0.1512665001783803,
+      "grad_norm": 1.0582475662231445,
+      "learning_rate": 3.025114155251142e-05,
+      "loss": 0.4781,
+      "step": 530
+    },
+    {
+      "epoch": 0.15412058508740634,
+      "grad_norm": 2.1239373683929443,
+      "learning_rate": 3.082191780821918e-05,
+      "loss": 0.5293,
+      "step": 540
+    },
+    {
+      "epoch": 0.1569746699964324,
+      "grad_norm": 1.3766371011734009,
+      "learning_rate": 3.1392694063926944e-05,
+      "loss": 0.4626,
+      "step": 550
+    },
+    {
+      "epoch": 0.15982875490545845,
+      "grad_norm": 1.2140746116638184,
+      "learning_rate": 3.1963470319634704e-05,
+      "loss": 0.4471,
+      "step": 560
+    },
+    {
+      "epoch": 0.16268283981448448,
+      "grad_norm": 1.7624636888504028,
+      "learning_rate": 3.253424657534247e-05,
+      "loss": 0.4691,
+      "step": 570
+    },
+    {
+      "epoch": 0.16553692472351053,
+      "grad_norm": 3.22637939453125,
+      "learning_rate": 3.310502283105023e-05,
+      "loss": 0.5514,
+      "step": 580
+    },
+    {
+      "epoch": 0.16839100963253656,
+      "grad_norm": 3.3661112785339355,
+      "learning_rate": 3.367579908675799e-05,
+      "loss": 0.5277,
+      "step": 590
+    },
+    {
+      "epoch": 0.17124509454156261,
+      "grad_norm": 2.009028673171997,
+      "learning_rate": 3.424657534246575e-05,
+      "loss": 0.4961,
+      "step": 600
+    },
+    {
+      "epoch": 0.17409917945058864,
+      "grad_norm": 1.145951271057129,
+      "learning_rate": 3.481735159817352e-05,
+      "loss": 0.484,
+      "step": 610
+    },
+    {
+      "epoch": 0.1769532643596147,
+      "grad_norm": 1.4422398805618286,
+      "learning_rate": 3.538812785388128e-05,
+      "loss": 0.4913,
+      "step": 620
+    },
+    {
+      "epoch": 0.17980734926864075,
+      "grad_norm": 1.5047334432601929,
+      "learning_rate": 3.5958904109589045e-05,
+      "loss": 0.468,
+      "step": 630
+    },
+    {
+      "epoch": 0.18266143417766678,
+      "grad_norm": 1.358590841293335,
+      "learning_rate": 3.6529680365296805e-05,
+      "loss": 0.4549,
+      "step": 640
+    },
+    {
+      "epoch": 0.18551551908669284,
+      "grad_norm": 1.2067798376083374,
+      "learning_rate": 3.710045662100457e-05,
+      "loss": 0.4726,
+      "step": 650
+    },
+    {
+      "epoch": 0.18836960399571887,
+      "grad_norm": 1.3069053888320923,
+      "learning_rate": 3.767123287671233e-05,
+      "loss": 0.4998,
+      "step": 660
+    },
+    {
+      "epoch": 0.19122368890474492,
+      "grad_norm": 1.4003655910491943,
+      "learning_rate": 3.824200913242009e-05,
+      "loss": 0.4396,
+      "step": 670
+    },
+    {
+      "epoch": 0.19407777381377095,
+      "grad_norm": 1.6738018989562988,
+      "learning_rate": 3.881278538812785e-05,
+      "loss": 0.4782,
+      "step": 680
+    },
+    {
+      "epoch": 0.196931858722797,
+      "grad_norm": 1.3846429586410522,
+      "learning_rate": 3.938356164383562e-05,
+      "loss": 0.4873,
+      "step": 690
+    },
+    {
+      "epoch": 0.19978594363182306,
+      "grad_norm": 1.5841200351715088,
+      "learning_rate": 3.995433789954338e-05,
+      "loss": 0.4568,
+      "step": 700
+    },
+    {
+      "epoch": 0.2026400285408491,
+      "grad_norm": 1.7131880521774292,
+      "learning_rate": 4.0525114155251145e-05,
+      "loss": 0.5105,
+      "step": 710
+    },
+    {
+      "epoch": 0.20549411344987514,
+      "grad_norm": 0.7919635772705078,
+      "learning_rate": 4.1095890410958905e-05,
+      "loss": 0.4417,
+      "step": 720
+    },
+    {
+      "epoch": 0.20834819835890117,
+      "grad_norm": 1.2708427906036377,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.454,
+      "step": 730
+    },
+    {
+      "epoch": 0.21120228326792723,
+      "grad_norm": 1.4845744371414185,
+      "learning_rate": 4.223744292237443e-05,
+      "loss": 0.4498,
+      "step": 740
+    },
+    {
+      "epoch": 0.21405636817695325,
+      "grad_norm": 1.4763469696044922,
+      "learning_rate": 4.280821917808219e-05,
+      "loss": 0.4724,
+      "step": 750
+    },
+    {
+      "epoch": 0.2169104530859793,
+      "grad_norm": 1.6870049238204956,
+      "learning_rate": 4.337899543378995e-05,
+      "loss": 0.4885,
+      "step": 760
+    },
+    {
+      "epoch": 0.21976453799500534,
+      "grad_norm": 0.9296655058860779,
+      "learning_rate": 4.394977168949772e-05,
+      "loss": 0.4669,
+      "step": 770
+    },
+    {
+      "epoch": 0.2226186229040314,
+      "grad_norm": 0.876143753528595,
+      "learning_rate": 4.452054794520548e-05,
+      "loss": 0.4569,
+      "step": 780
+    },
+    {
+      "epoch": 0.22547270781305745,
+      "grad_norm": 0.651347815990448,
+      "learning_rate": 4.5091324200913246e-05,
+      "loss": 0.4307,
+      "step": 790
+    },
+    {
+      "epoch": 0.22832679272208348,
+      "grad_norm": 0.7317978739738464,
+      "learning_rate": 4.5662100456621006e-05,
+      "loss": 0.4713,
+      "step": 800
+    },
+    {
+      "epoch": 0.23118087763110953,
+      "grad_norm": 1.1030404567718506,
+      "learning_rate": 4.623287671232877e-05,
+      "loss": 0.4387,
+      "step": 810
+    },
+    {
+      "epoch": 0.23403496254013556,
+      "grad_norm": 1.3097269535064697,
+      "learning_rate": 4.680365296803653e-05,
+      "loss": 0.4327,
+      "step": 820
+    },
+    {
+      "epoch": 0.23688904744916162,
+      "grad_norm": 1.6860710382461548,
+      "learning_rate": 4.737442922374429e-05,
+      "loss": 0.4658,
+      "step": 830
+    },
+    {
+      "epoch": 0.23974313235818764,
+      "grad_norm": 0.9696588516235352,
+      "learning_rate": 4.794520547945205e-05,
+      "loss": 0.485,
+      "step": 840
+    },
+    {
+      "epoch": 0.2425972172672137,
+      "grad_norm": 1.1960479021072388,
+      "learning_rate": 4.851598173515982e-05,
+      "loss": 0.454,
+      "step": 850
+    },
+    {
+      "epoch": 0.24545130217623976,
+      "grad_norm": 1.4926533699035645,
+      "learning_rate": 4.908675799086758e-05,
+      "loss": 0.5058,
+      "step": 860
+    },
+    {
+      "epoch": 0.24830538708526578,
+      "grad_norm": 0.5784097909927368,
+      "learning_rate": 4.9657534246575346e-05,
+      "loss": 0.4072,
+      "step": 870
+    },
+    {
+      "epoch": 0.25115947199429184,
+      "grad_norm": 1.0179357528686523,
+      "learning_rate": 5.0228310502283106e-05,
+      "loss": 0.4462,
+      "step": 880
+    },
+    {
+      "epoch": 0.2540135569033179,
+      "grad_norm": 0.9068560600280762,
+      "learning_rate": 5.0799086757990866e-05,
+      "loss": 0.4674,
+      "step": 890
+    },
+    {
+      "epoch": 0.2568676418123439,
+      "grad_norm": 1.4015111923217773,
+      "learning_rate": 5.136986301369864e-05,
+      "loss": 0.4528,
+      "step": 900
+    },
+    {
+      "epoch": 0.25972172672136995,
+      "grad_norm": 0.6935715675354004,
+      "learning_rate": 5.19406392694064e-05,
+      "loss": 0.4841,
+      "step": 910
+    },
+    {
+      "epoch": 0.262575811630396,
+      "grad_norm": 1.1978791952133179,
+      "learning_rate": 5.251141552511416e-05,
+      "loss": 0.4707,
+      "step": 920
+    },
+    {
+      "epoch": 0.26542989653942206,
+      "grad_norm": 0.9195595979690552,
+      "learning_rate": 5.308219178082192e-05,
+      "loss": 0.4538,
+      "step": 930
+    },
+    {
+      "epoch": 0.2682839814484481,
+      "grad_norm": 1.5978795289993286,
+      "learning_rate": 5.365296803652968e-05,
+      "loss": 0.4536,
+      "step": 940
+    },
+    {
+      "epoch": 0.2711380663574741,
+      "grad_norm": 1.3926838636398315,
+      "learning_rate": 5.422374429223745e-05,
+      "loss": 0.49,
+      "step": 950
+    },
+    {
+      "epoch": 0.2739921512665002,
+      "grad_norm": 0.575034499168396,
+      "learning_rate": 5.479452054794521e-05,
+      "loss": 0.5136,
+      "step": 960
+    },
+    {
+      "epoch": 0.27684623617552623,
+      "grad_norm": 0.6068254709243774,
+      "learning_rate": 5.536529680365297e-05,
+      "loss": 0.4552,
+      "step": 970
+    },
+    {
+      "epoch": 0.2797003210845523,
+      "grad_norm": 0.7864544987678528,
+      "learning_rate": 5.593607305936074e-05,
+      "loss": 0.492,
+      "step": 980
+    },
+    {
+      "epoch": 0.2825544059935783,
+      "grad_norm": 1.3433411121368408,
+      "learning_rate": 5.65068493150685e-05,
+      "loss": 0.4795,
+      "step": 990
+    },
+    {
+      "epoch": 0.28540849090260434,
+      "grad_norm": 0.7735553979873657,
+      "learning_rate": 5.707762557077626e-05,
+      "loss": 0.4538,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2882625758116304,
+      "grad_norm": 0.6869261264801025,
+      "learning_rate": 5.764840182648402e-05,
+      "loss": 0.4821,
+      "step": 1010
+    },
+    {
+      "epoch": 0.29111666072065645,
+      "grad_norm": 1.081992506980896,
+      "learning_rate": 5.821917808219178e-05,
+      "loss": 0.4669,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2939707456296825,
+      "grad_norm": 0.8736602663993835,
+      "learning_rate": 5.878995433789955e-05,
+      "loss": 0.4253,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2968248305387085,
+      "grad_norm": 1.1186368465423584,
+      "learning_rate": 5.936073059360731e-05,
+      "loss": 0.469,
+      "step": 1040
+    },
+    {
+      "epoch": 0.29967891544773456,
+      "grad_norm": 0.6147318482398987,
+      "learning_rate": 5.993150684931507e-05,
+      "loss": 0.4573,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3025330003567606,
+      "grad_norm": 1.0282665491104126,
+      "learning_rate": 6.050228310502284e-05,
+      "loss": 0.4448,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3053870852657867,
+      "grad_norm": 1.3858377933502197,
+      "learning_rate": 6.10730593607306e-05,
+      "loss": 0.4453,
+      "step": 1070
+    },
+    {
+      "epoch": 0.3082411701748127,
+      "grad_norm": 0.9879347085952759,
+      "learning_rate": 6.164383561643835e-05,
+      "loss": 0.4441,
+      "step": 1080
+    },
+    {
+      "epoch": 0.31109525508383873,
+      "grad_norm": 0.7240331172943115,
+      "learning_rate": 6.221461187214613e-05,
+      "loss": 0.4408,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3139493399928648,
+      "grad_norm": 0.7207432389259338,
+      "learning_rate": 6.278538812785389e-05,
+      "loss": 0.4753,
+      "step": 1100
+    },
+    {
+      "epoch": 0.31680342490189084,
+      "grad_norm": 1.230485439300537,
+      "learning_rate": 6.335616438356165e-05,
+      "loss": 0.4815,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3196575098109169,
+      "grad_norm": 0.7309743762016296,
+      "learning_rate": 6.392694063926941e-05,
+      "loss": 0.424,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3225115947199429,
+      "grad_norm": 0.919762134552002,
+      "learning_rate": 6.449771689497717e-05,
+      "loss": 0.4576,
+      "step": 1130
+    },
+    {
+      "epoch": 0.32536567962896895,
+      "grad_norm": 0.6879755258560181,
+      "learning_rate": 6.506849315068494e-05,
+      "loss": 0.4349,
+      "step": 1140
+    },
+    {
+      "epoch": 0.328219764537995,
+      "grad_norm": 0.8196412920951843,
+      "learning_rate": 6.56392694063927e-05,
+      "loss": 0.4955,
+      "step": 1150
+    },
+    {
+      "epoch": 0.33107384944702106,
+      "grad_norm": 1.016493558883667,
+      "learning_rate": 6.621004566210046e-05,
+      "loss": 0.4656,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3339279343560471,
+      "grad_norm": 0.560720682144165,
+      "learning_rate": 6.678082191780822e-05,
+      "loss": 0.4251,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3367820192650731,
+      "grad_norm": 0.5806992053985596,
+      "learning_rate": 6.735159817351598e-05,
+      "loss": 0.4293,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3396361041740992,
+      "grad_norm": 0.9094499349594116,
+      "learning_rate": 6.792237442922374e-05,
+      "loss": 0.4143,
+      "step": 1190
+    },
+    {
+      "epoch": 0.34249018908312523,
+      "grad_norm": 0.5154547691345215,
+      "learning_rate": 6.84931506849315e-05,
+      "loss": 0.4682,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3453442739921513,
+      "grad_norm": 0.7153740525245667,
+      "learning_rate": 6.906392694063926e-05,
+      "loss": 0.4717,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3481983589011773,
+      "grad_norm": 0.6453626155853271,
+      "learning_rate": 6.963470319634704e-05,
+      "loss": 0.4609,
+      "step": 1220
+    },
+    {
+      "epoch": 0.35105244381020334,
+      "grad_norm": 0.6944445371627808,
+      "learning_rate": 7.02054794520548e-05,
+      "loss": 0.4446,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3539065287192294,
+      "grad_norm": 0.5920615196228027,
+      "learning_rate": 7.077625570776256e-05,
+      "loss": 0.4863,
+      "step": 1240
+    },
+    {
+      "epoch": 0.35676061362825545,
+      "grad_norm": 0.8108448386192322,
+      "learning_rate": 7.134703196347033e-05,
+      "loss": 0.4713,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3596146985372815,
+      "grad_norm": 0.6366176605224609,
+      "learning_rate": 7.191780821917809e-05,
+      "loss": 0.4338,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3624687834463075,
+      "grad_norm": 0.9244728088378906,
+      "learning_rate": 7.248858447488585e-05,
+      "loss": 0.4319,
+      "step": 1270
+    },
+    {
+      "epoch": 0.36532286835533356,
+      "grad_norm": 0.7847846150398254,
+      "learning_rate": 7.305936073059361e-05,
+      "loss": 0.4692,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3681769532643596,
+      "grad_norm": 0.9149333238601685,
+      "learning_rate": 7.363013698630137e-05,
+      "loss": 0.4677,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3710310381733857,
+      "grad_norm": 0.7170394659042358,
+      "learning_rate": 7.420091324200914e-05,
+      "loss": 0.4261,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3738851230824117,
+      "grad_norm": 1.0459473133087158,
+      "learning_rate": 7.47716894977169e-05,
+      "loss": 0.4709,
+      "step": 1310
+    },
+    {
+      "epoch": 0.37673920799143773,
+      "grad_norm": 1.1265268325805664,
+      "learning_rate": 7.534246575342466e-05,
+      "loss": 0.4799,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3795932929004638,
+      "grad_norm": 0.7591957449913025,
+      "learning_rate": 7.591324200913242e-05,
+      "loss": 0.4438,
+      "step": 1330
+    },
+    {
+      "epoch": 0.38244737780948984,
+      "grad_norm": 0.5419506430625916,
+      "learning_rate": 7.648401826484018e-05,
+      "loss": 0.4578,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3853014627185159,
+      "grad_norm": 0.5713040828704834,
+      "learning_rate": 7.705479452054794e-05,
+      "loss": 0.4633,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3881555476275419,
+      "grad_norm": 0.9401557445526123,
+      "learning_rate": 7.76255707762557e-05,
+      "loss": 0.4517,
+      "step": 1360
+    },
+    {
+      "epoch": 0.39100963253656795,
+      "grad_norm": 0.9244952201843262,
+      "learning_rate": 7.819634703196348e-05,
+      "loss": 0.4429,
+      "step": 1370
+    },
+    {
+      "epoch": 0.393863717445594,
+      "grad_norm": 0.6746932864189148,
+      "learning_rate": 7.876712328767124e-05,
+      "loss": 0.4595,
+      "step": 1380
+    },
+    {
+      "epoch": 0.39671780235462006,
+      "grad_norm": 0.7216023206710815,
+      "learning_rate": 7.9337899543379e-05,
+      "loss": 0.4364,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3995718872636461,
+      "grad_norm": 0.8057281374931335,
+      "learning_rate": 7.990867579908676e-05,
+      "loss": 0.4532,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4024259721726721,
+      "grad_norm": 0.5679146647453308,
+      "learning_rate": 8.047945205479453e-05,
+      "loss": 0.42,
+      "step": 1410
+    },
+    {
+      "epoch": 0.4052800570816982,
+      "grad_norm": 0.45307865738868713,
+      "learning_rate": 8.105022831050229e-05,
+      "loss": 0.4454,
+      "step": 1420
+    },
+    {
+      "epoch": 0.40813414199072423,
+      "grad_norm": 0.559577226638794,
+      "learning_rate": 8.162100456621005e-05,
+      "loss": 0.4723,
+      "step": 1430
+    },
+    {
+      "epoch": 0.4109882268997503,
+      "grad_norm": 1.2319309711456299,
+      "learning_rate": 8.219178082191781e-05,
+      "loss": 0.4659,
+      "step": 1440
+    },
+    {
+      "epoch": 0.4138423118087763,
+      "grad_norm": 0.6527321338653564,
+      "learning_rate": 8.276255707762558e-05,
+      "loss": 0.4235,
+      "step": 1450
+    },
+    {
+      "epoch": 0.41669639671780234,
+      "grad_norm": 0.6203667521476746,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 0.4565,
+      "step": 1460
+    },
+    {
+      "epoch": 0.4195504816268284,
+      "grad_norm": 0.9035118818283081,
+      "learning_rate": 8.39041095890411e-05,
+      "loss": 0.4528,
+      "step": 1470
+    },
+    {
+      "epoch": 0.42240456653585445,
+      "grad_norm": 1.0064053535461426,
+      "learning_rate": 8.447488584474886e-05,
+      "loss": 0.4358,
+      "step": 1480
+    },
+    {
+      "epoch": 0.4252586514448805,
+      "grad_norm": 0.7333278656005859,
+      "learning_rate": 8.504566210045662e-05,
+      "loss": 0.4523,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4281127363539065,
+      "grad_norm": 0.7744036912918091,
+      "learning_rate": 8.561643835616438e-05,
+      "loss": 0.4208,
+      "step": 1500
+    },
+    {
+      "epoch": 0.43096682126293256,
+      "grad_norm": 1.0139415264129639,
+      "learning_rate": 8.618721461187214e-05,
+      "loss": 0.4566,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4338209061719586,
+      "grad_norm": 0.83629310131073,
+      "learning_rate": 8.67579908675799e-05,
+      "loss": 0.4434,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4366749910809847,
+      "grad_norm": 0.7974062561988831,
+      "learning_rate": 8.732876712328768e-05,
+      "loss": 0.4562,
+      "step": 1530
+    },
+    {
+      "epoch": 0.4395290759900107,
+      "grad_norm": 0.6136748194694519,
+      "learning_rate": 8.789954337899544e-05,
+      "loss": 0.4556,
+      "step": 1540
+    },
+    {
+      "epoch": 0.44238316089903673,
+      "grad_norm": 0.6411725282669067,
+      "learning_rate": 8.84703196347032e-05,
+      "loss": 0.4601,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4452372458080628,
+      "grad_norm": 0.9157238006591797,
+      "learning_rate": 8.904109589041096e-05,
+      "loss": 0.4473,
+      "step": 1560
+    },
+    {
+      "epoch": 0.44809133071708884,
+      "grad_norm": 0.5935428142547607,
+      "learning_rate": 8.961187214611873e-05,
+      "loss": 0.4374,
+      "step": 1570
+    },
+    {
+      "epoch": 0.4509454156261149,
+      "grad_norm": 0.6518482565879822,
+      "learning_rate": 9.018264840182649e-05,
+      "loss": 0.4898,
+      "step": 1580
+    },
+    {
+      "epoch": 0.4537995005351409,
+      "grad_norm": 0.990638792514801,
+      "learning_rate": 9.075342465753425e-05,
+      "loss": 0.4623,
+      "step": 1590
+    },
+    {
+      "epoch": 0.45665358544416695,
+      "grad_norm": 0.7768042683601379,
+      "learning_rate": 9.132420091324201e-05,
+      "loss": 0.4203,
+      "step": 1600
+    },
+    {
+      "epoch": 0.459507670353193,
+      "grad_norm": 0.49145108461380005,
+      "learning_rate": 9.189497716894979e-05,
+      "loss": 0.4454,
+      "step": 1610
+    },
+    {
+      "epoch": 0.46236175526221907,
+      "grad_norm": 0.7254750728607178,
+      "learning_rate": 9.246575342465755e-05,
+      "loss": 0.4204,
+      "step": 1620
+    },
+    {
+      "epoch": 0.4652158401712451,
+      "grad_norm": 0.5650269985198975,
+      "learning_rate": 9.30365296803653e-05,
+      "loss": 0.4431,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4680699250802711,
+      "grad_norm": 0.9648821353912354,
+      "learning_rate": 9.360730593607307e-05,
+      "loss": 0.4389,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4709240099892972,
+      "grad_norm": 0.6625390648841858,
+      "learning_rate": 9.417808219178083e-05,
+      "loss": 0.4491,
+      "step": 1650
+    },
+    {
+      "epoch": 0.47377809489832323,
+      "grad_norm": 0.8872827291488647,
+      "learning_rate": 9.474885844748859e-05,
+      "loss": 0.443,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4766321798073493,
+      "grad_norm": 0.6523913741111755,
+      "learning_rate": 9.531963470319635e-05,
+      "loss": 0.421,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4794862647163753,
+      "grad_norm": 0.48721668124198914,
+      "learning_rate": 9.58904109589041e-05,
+      "loss": 0.4762,
+      "step": 1680
+    },
+    {
+      "epoch": 0.48234034962540134,
+      "grad_norm": 0.808788537979126,
+      "learning_rate": 9.646118721461188e-05,
+      "loss": 0.4383,
+      "step": 1690
+    },
+    {
+      "epoch": 0.4851944345344274,
+      "grad_norm": 1.0816537141799927,
+      "learning_rate": 9.703196347031964e-05,
+      "loss": 0.4273,
+      "step": 1700
+    },
+    {
+      "epoch": 0.48804851944345345,
+      "grad_norm": 0.9300740361213684,
+      "learning_rate": 9.76027397260274e-05,
+      "loss": 0.4383,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4909026043524795,
+      "grad_norm": 0.6394598484039307,
+      "learning_rate": 9.817351598173516e-05,
+      "loss": 0.4864,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4937566892615055,
+      "grad_norm": 0.6434561014175415,
+      "learning_rate": 9.874429223744292e-05,
+      "loss": 0.4815,
+      "step": 1730
+    },
+    {
+      "epoch": 0.49661077417053157,
+      "grad_norm": 0.7413051128387451,
+      "learning_rate": 9.931506849315069e-05,
+      "loss": 0.465,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4994648590795576,
+      "grad_norm": 0.6108921766281128,
+      "learning_rate": 9.988584474885845e-05,
+      "loss": 0.4645,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5023189439885837,
+      "grad_norm": 0.6062872409820557,
+      "learning_rate": 9.999993644614703e-05,
+      "loss": 0.4244,
+      "step": 1760
+    },
+    {
+      "epoch": 0.5051730288976097,
+      "grad_norm": 0.7179297208786011,
+      "learning_rate": 9.999967825889622e-05,
+      "loss": 0.437,
+      "step": 1770
+    },
+    {
+      "epoch": 0.5080271138066358,
+      "grad_norm": 0.7407391667366028,
+      "learning_rate": 9.999922146715655e-05,
+      "loss": 0.4506,
+      "step": 1780
+    },
+    {
+      "epoch": 0.5108811987156618,
+      "grad_norm": 0.38226214051246643,
+      "learning_rate": 9.999856607274242e-05,
+      "loss": 0.3888,
+      "step": 1790
+    },
+    {
+      "epoch": 0.5137352836246878,
+      "grad_norm": 0.8683974146842957,
+      "learning_rate": 9.999771207825713e-05,
+      "loss": 0.445,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5165893685337138,
+      "grad_norm": 0.6052406430244446,
+      "learning_rate": 9.999665948709287e-05,
+      "loss": 0.393,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5194434534427399,
+      "grad_norm": 0.9383887052536011,
+      "learning_rate": 9.999540830343064e-05,
+      "loss": 0.4655,
+      "step": 1820
+    },
+    {
+      "epoch": 0.522297538351766,
+      "grad_norm": 0.6511524319648743,
+      "learning_rate": 9.999395853224028e-05,
+      "loss": 0.4633,
+      "step": 1830
+    },
+    {
+      "epoch": 0.525151623260792,
+      "grad_norm": 0.3971940875053406,
+      "learning_rate": 9.999231017928045e-05,
+      "loss": 0.408,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5280057081698181,
+      "grad_norm": 0.6968293786048889,
+      "learning_rate": 9.99904632510986e-05,
+      "loss": 0.4559,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5308597930788441,
+      "grad_norm": 0.47270509600639343,
+      "learning_rate": 9.998841775503096e-05,
+      "loss": 0.4257,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5337138779878702,
+      "grad_norm": 0.7707135081291199,
+      "learning_rate": 9.998617369920244e-05,
+      "loss": 0.4326,
+      "step": 1870
+    },
+    {
+      "epoch": 0.5365679628968962,
+      "grad_norm": 0.720313549041748,
+      "learning_rate": 9.998373109252672e-05,
+      "loss": 0.4257,
+      "step": 1880
+    },
+    {
+      "epoch": 0.5394220478059222,
+      "grad_norm": 0.9472813606262207,
+      "learning_rate": 9.998108994470612e-05,
+      "loss": 0.4477,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5422761327149482,
+      "grad_norm": 0.5608553290367126,
+      "learning_rate": 9.997825026623155e-05,
+      "loss": 0.4542,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5451302176239743,
+      "grad_norm": 1.084322452545166,
+      "learning_rate": 9.997521206838255e-05,
+      "loss": 0.4208,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5479843025330003,
+      "grad_norm": 0.6105670928955078,
+      "learning_rate": 9.997197536322722e-05,
+      "loss": 0.4282,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5508383874420264,
+      "grad_norm": 0.5767210721969604,
+      "learning_rate": 9.99685401636221e-05,
+      "loss": 0.455,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5536924723510525,
+      "grad_norm": 0.5981484055519104,
+      "learning_rate": 9.996490648321219e-05,
+      "loss": 0.4424,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5565465572600785,
+      "grad_norm": 0.7510049939155579,
+      "learning_rate": 9.996107433643092e-05,
+      "loss": 0.4533,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5594006421691046,
+      "grad_norm": 0.9091824889183044,
+      "learning_rate": 9.99570437385e-05,
+      "loss": 0.4232,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5622547270781306,
+      "grad_norm": 0.5205058455467224,
+      "learning_rate": 9.995281470542943e-05,
+      "loss": 0.4501,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5651088119871566,
+      "grad_norm": 0.5514882206916809,
+      "learning_rate": 9.994838725401743e-05,
+      "loss": 0.4258,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5679628968961826,
+      "grad_norm": 0.6759376525878906,
+      "learning_rate": 9.994376140185032e-05,
+      "loss": 0.4764,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5708169818052087,
+      "grad_norm": 0.5412331223487854,
+      "learning_rate": 9.993893716730255e-05,
+      "loss": 0.4606,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5736710667142347,
+      "grad_norm": 0.7248188257217407,
+      "learning_rate": 9.993391456953651e-05,
+      "loss": 0.4417,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5765251516232608,
+      "grad_norm": 0.6278032660484314,
+      "learning_rate": 9.992869362850256e-05,
+      "loss": 0.4354,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5793792365322868,
+      "grad_norm": 0.7813991904258728,
+      "learning_rate": 9.992327436493886e-05,
+      "loss": 0.4386,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5822333214413129,
+      "grad_norm": 0.6477668881416321,
+      "learning_rate": 9.991765680037137e-05,
+      "loss": 0.3989,
+      "step": 2040
+    },
+    {
+      "epoch": 0.585087406350339,
+      "grad_norm": 0.6790643930435181,
+      "learning_rate": 9.991184095711368e-05,
+      "loss": 0.4251,
+      "step": 2050
+    },
+    {
+      "epoch": 0.587941491259365,
+      "grad_norm": 0.6338703632354736,
+      "learning_rate": 9.990582685826701e-05,
+      "loss": 0.4251,
+      "step": 2060
+    },
+    {
+      "epoch": 0.590795576168391,
+      "grad_norm": 0.5298501253128052,
+      "learning_rate": 9.989961452772005e-05,
+      "loss": 0.4457,
+      "step": 2070
+    },
+    {
+      "epoch": 0.593649661077417,
+      "grad_norm": 0.547738790512085,
+      "learning_rate": 9.98932039901489e-05,
+      "loss": 0.447,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5965037459864431,
+      "grad_norm": 0.6610301733016968,
+      "learning_rate": 9.988659527101693e-05,
+      "loss": 0.43,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5993578308954691,
+      "grad_norm": 0.5846606492996216,
+      "learning_rate": 9.987978839657475e-05,
+      "loss": 0.4594,
+      "step": 2100
+    },
+    {
+      "epoch": 0.6022119158044952,
+      "grad_norm": 0.5017244815826416,
+      "learning_rate": 9.987278339386006e-05,
+      "loss": 0.4136,
+      "step": 2110
+    },
+    {
+      "epoch": 0.6050660007135212,
+      "grad_norm": 0.6239372491836548,
+      "learning_rate": 9.986558029069753e-05,
+      "loss": 0.4765,
+      "step": 2120
+    },
+    {
+      "epoch": 0.6079200856225473,
+      "grad_norm": 0.4270966053009033,
+      "learning_rate": 9.98581791156987e-05,
+      "loss": 0.421,
+      "step": 2130
+    },
+    {
+      "epoch": 0.6107741705315733,
+      "grad_norm": 0.9911853075027466,
+      "learning_rate": 9.985057989826193e-05,
+      "loss": 0.4398,
+      "step": 2140
+    },
+    {
+      "epoch": 0.6136282554405994,
+      "grad_norm": 0.5237308740615845,
+      "learning_rate": 9.984278266857215e-05,
+      "loss": 0.4036,
+      "step": 2150
+    },
+    {
+      "epoch": 0.6164823403496253,
+      "grad_norm": 0.7680717706680298,
+      "learning_rate": 9.983478745760088e-05,
+      "loss": 0.4579,
+      "step": 2160
+    },
+    {
+      "epoch": 0.6193364252586514,
+      "grad_norm": 0.6343058347702026,
+      "learning_rate": 9.982659429710601e-05,
+      "loss": 0.4119,
+      "step": 2170
+    },
+    {
+      "epoch": 0.6221905101676775,
+      "grad_norm": 0.4680488109588623,
+      "learning_rate": 9.981820321963171e-05,
+      "loss": 0.4047,
+      "step": 2180
+    },
+    {
+      "epoch": 0.6250445950767035,
+      "grad_norm": 0.4934472143650055,
+      "learning_rate": 9.980961425850832e-05,
+      "loss": 0.4377,
+      "step": 2190
+    },
+    {
+      "epoch": 0.6278986799857296,
+      "grad_norm": 0.6324202418327332,
+      "learning_rate": 9.980082744785219e-05,
+      "loss": 0.4417,
+      "step": 2200
+    },
+    {
+      "epoch": 0.6307527648947556,
+      "grad_norm": 0.5263383984565735,
+      "learning_rate": 9.979184282256553e-05,
+      "loss": 0.4166,
+      "step": 2210
+    },
+    {
+      "epoch": 0.6336068498037817,
+      "grad_norm": 1.1357102394104004,
+      "learning_rate": 9.978266041833631e-05,
+      "loss": 0.4536,
+      "step": 2220
+    },
+    {
+      "epoch": 0.6364609347128077,
+      "grad_norm": 1.1446198225021362,
+      "learning_rate": 9.977328027163811e-05,
+      "loss": 0.4449,
+      "step": 2230
+    },
+    {
+      "epoch": 0.6393150196218338,
+      "grad_norm": 0.4317645728588104,
+      "learning_rate": 9.976370241972995e-05,
+      "loss": 0.4161,
+      "step": 2240
+    },
+    {
+      "epoch": 0.6421691045308598,
+      "grad_norm": 0.6288971304893494,
+      "learning_rate": 9.975392690065616e-05,
+      "loss": 0.4469,
+      "step": 2250
+    },
+    {
+      "epoch": 0.6450231894398858,
+      "grad_norm": 0.5151458978652954,
+      "learning_rate": 9.974395375324622e-05,
+      "loss": 0.4478,
+      "step": 2260
+    },
+    {
+      "epoch": 0.6478772743489118,
+      "grad_norm": 0.6855816841125488,
+      "learning_rate": 9.973378301711465e-05,
+      "loss": 0.437,
+      "step": 2270
+    },
+    {
+      "epoch": 0.6507313592579379,
+      "grad_norm": 0.5615636110305786,
+      "learning_rate": 9.972341473266074e-05,
+      "loss": 0.4475,
+      "step": 2280
+    },
+    {
+      "epoch": 0.653585444166964,
+      "grad_norm": 0.758758544921875,
+      "learning_rate": 9.971284894106856e-05,
+      "loss": 0.4066,
+      "step": 2290
+    },
+    {
+      "epoch": 0.65643952907599,
+      "grad_norm": 0.5864000916481018,
+      "learning_rate": 9.970208568430662e-05,
+      "loss": 0.4327,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6592936139850161,
+      "grad_norm": 0.5502752065658569,
+      "learning_rate": 9.969112500512784e-05,
+      "loss": 0.4266,
+      "step": 2310
+    },
+    {
+      "epoch": 0.6621476988940421,
+      "grad_norm": 0.9890068769454956,
+      "learning_rate": 9.96799669470693e-05,
+      "loss": 0.4279,
+      "step": 2320
+    },
+    {
+      "epoch": 0.6650017838030682,
+      "grad_norm": 0.6124778389930725,
+      "learning_rate": 9.966861155445207e-05,
+      "loss": 0.4138,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6678558687120942,
+      "grad_norm": 0.6781124472618103,
+      "learning_rate": 9.965705887238109e-05,
+      "loss": 0.4264,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6707099536211202,
+      "grad_norm": 0.5691377520561218,
+      "learning_rate": 9.964530894674495e-05,
+      "loss": 0.4407,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6735640385301462,
+      "grad_norm": 0.5900623202323914,
+      "learning_rate": 9.963336182421572e-05,
+      "loss": 0.4134,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6764181234391723,
+      "grad_norm": 0.9385659694671631,
+      "learning_rate": 9.962121755224874e-05,
+      "loss": 0.4488,
+      "step": 2370
+    },
+    {
+      "epoch": 0.6792722083481983,
+      "grad_norm": 0.5161449909210205,
+      "learning_rate": 9.960887617908245e-05,
+      "loss": 0.4443,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6821262932572244,
+      "grad_norm": 0.636301577091217,
+      "learning_rate": 9.959633775373823e-05,
+      "loss": 0.4416,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6849803781662505,
+      "grad_norm": 0.666099488735199,
+      "learning_rate": 9.958360232602013e-05,
+      "loss": 0.4629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6878344630752765,
+      "grad_norm": 0.8457536697387695,
+      "learning_rate": 9.957066994651474e-05,
+      "loss": 0.4516,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6906885479843026,
+      "grad_norm": 0.753018319606781,
+      "learning_rate": 9.955754066659096e-05,
+      "loss": 0.4199,
+      "step": 2420
+    },
+    {
+      "epoch": 0.6935426328933286,
+      "grad_norm": 0.6624453663825989,
+      "learning_rate": 9.954421453839983e-05,
+      "loss": 0.4301,
+      "step": 2430
+    },
+    {
+      "epoch": 0.6963967178023546,
+      "grad_norm": 0.5142033696174622,
+      "learning_rate": 9.953069161487422e-05,
+      "loss": 0.445,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6992508027113806,
+      "grad_norm": 0.6645984053611755,
+      "learning_rate": 9.95169719497288e-05,
+      "loss": 0.4784,
+      "step": 2450
+    },
+    {
+      "epoch": 0.7021048876204067,
+      "grad_norm": 0.767155647277832,
+      "learning_rate": 9.950305559745963e-05,
+      "loss": 0.4029,
+      "step": 2460
+    },
+    {
+      "epoch": 0.7049589725294327,
+      "grad_norm": 0.7413857579231262,
+      "learning_rate": 9.948894261334408e-05,
+      "loss": 0.4558,
+      "step": 2470
+    },
+    {
+      "epoch": 0.7078130574384588,
+      "grad_norm": 0.6691272854804993,
+      "learning_rate": 9.947463305344051e-05,
+      "loss": 0.4196,
+      "step": 2480
+    },
+    {
+      "epoch": 0.7106671423474848,
+      "grad_norm": 0.7997813820838928,
+      "learning_rate": 9.946012697458819e-05,
+      "loss": 0.4381,
+      "step": 2490
+    },
+    {
+      "epoch": 0.7135212272565109,
+      "grad_norm": 1.3109943866729736,
+      "learning_rate": 9.944542443440693e-05,
+      "loss": 0.4342,
+      "step": 2500
+    },
+    {
+      "epoch": 0.716375312165537,
+      "grad_norm": 0.6375836730003357,
+      "learning_rate": 9.94305254912969e-05,
+      "loss": 0.4314,
+      "step": 2510
+    },
+    {
+      "epoch": 0.719229397074563,
+      "grad_norm": 0.5794687271118164,
+      "learning_rate": 9.941543020443843e-05,
+      "loss": 0.444,
+      "step": 2520
+    },
+    {
+      "epoch": 0.722083481983589,
+      "grad_norm": 0.5072224736213684,
+      "learning_rate": 9.940013863379173e-05,
+      "loss": 0.4442,
+      "step": 2530
+    },
+    {
+      "epoch": 0.724937566892615,
+      "grad_norm": 0.6260116696357727,
+      "learning_rate": 9.93846508400967e-05,
+      "loss": 0.4761,
+      "step": 2540
+    },
+    {
+      "epoch": 0.7277916518016411,
+      "grad_norm": 0.5651645660400391,
+      "learning_rate": 9.936896688487262e-05,
+      "loss": 0.48,
+      "step": 2550
+    },
+    {
+      "epoch": 0.7306457367106671,
+      "grad_norm": 1.0688337087631226,
+      "learning_rate": 9.9353086830418e-05,
+      "loss": 0.4256,
+      "step": 2560
+    },
+    {
+      "epoch": 0.7334998216196932,
+      "grad_norm": 0.7255706191062927,
+      "learning_rate": 9.933701073981023e-05,
+      "loss": 0.4464,
+      "step": 2570
+    },
+    {
+      "epoch": 0.7363539065287192,
+      "grad_norm": 0.5594896078109741,
+      "learning_rate": 9.932073867690539e-05,
+      "loss": 0.4456,
+      "step": 2580
+    },
+    {
+      "epoch": 0.7392079914377453,
+      "grad_norm": 0.5755929946899414,
+      "learning_rate": 9.930427070633798e-05,
+      "loss": 0.4013,
+      "step": 2590
+    },
+    {
+      "epoch": 0.7420620763467713,
+      "grad_norm": 0.8663392066955566,
+      "learning_rate": 9.928760689352072e-05,
+      "loss": 0.4306,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7449161612557974,
+      "grad_norm": 0.6546819806098938,
+      "learning_rate": 9.927074730464417e-05,
+      "loss": 0.4352,
+      "step": 2610
+    },
+    {
+      "epoch": 0.7477702461648233,
+      "grad_norm": 0.6611933708190918,
+      "learning_rate": 9.925369200667655e-05,
+      "loss": 0.4681,
+      "step": 2620
+    },
+    {
+      "epoch": 0.7506243310738494,
+      "grad_norm": 0.9629554152488708,
+      "learning_rate": 9.923644106736348e-05,
+      "loss": 0.4626,
+      "step": 2630
+    },
+    {
+      "epoch": 0.7534784159828755,
+      "grad_norm": 0.6556516289710999,
+      "learning_rate": 9.92189945552277e-05,
+      "loss": 0.4009,
+      "step": 2640
+    },
+    {
+      "epoch": 0.7563325008919015,
+      "grad_norm": 0.5907385945320129,
+      "learning_rate": 9.920135253956875e-05,
+      "loss": 0.4303,
+      "step": 2650
+    },
+    {
+      "epoch": 0.7591865858009276,
+      "grad_norm": 0.5351846218109131,
+      "learning_rate": 9.918351509046276e-05,
+      "loss": 0.4192,
+      "step": 2660
+    },
+    {
+      "epoch": 0.7620406707099536,
+      "grad_norm": 0.8660167455673218,
+      "learning_rate": 9.91654822787621e-05,
+      "loss": 0.4525,
+      "step": 2670
+    },
+    {
+      "epoch": 0.7648947556189797,
+      "grad_norm": 0.5204272270202637,
+      "learning_rate": 9.914725417609523e-05,
+      "loss": 0.3889,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7677488405280057,
+      "grad_norm": 0.8953636884689331,
+      "learning_rate": 9.912883085486626e-05,
+      "loss": 0.4623,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7706029254370318,
+      "grad_norm": 0.6904731392860413,
+      "learning_rate": 9.911021238825473e-05,
+      "loss": 0.4251,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7734570103460578,
+      "grad_norm": 0.4080071449279785,
+      "learning_rate": 9.909139885021535e-05,
+      "loss": 0.4477,
+      "step": 2710
+    },
+    {
+      "epoch": 0.7763110952550838,
+      "grad_norm": 0.6172974705696106,
+      "learning_rate": 9.907239031547765e-05,
+      "loss": 0.4164,
+      "step": 2720
+    },
+    {
+      "epoch": 0.7791651801641098,
+      "grad_norm": 0.4886259436607361,
+      "learning_rate": 9.905318685954574e-05,
+      "loss": 0.421,
+      "step": 2730
+    },
+    {
+      "epoch": 0.7820192650731359,
+      "grad_norm": 1.2826932668685913,
+      "learning_rate": 9.903378855869797e-05,
+      "loss": 0.4093,
+      "step": 2740
+    },
+    {
+      "epoch": 0.784873349982162,
+      "grad_norm": 0.7179902195930481,
+      "learning_rate": 9.901419548998658e-05,
+      "loss": 0.4022,
+      "step": 2750
+    },
+    {
+      "epoch": 0.787727434891188,
+      "grad_norm": 0.559853732585907,
+      "learning_rate": 9.899440773123756e-05,
+      "loss": 0.4364,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7905815198002141,
+      "grad_norm": 1.151147723197937,
+      "learning_rate": 9.897442536105013e-05,
+      "loss": 0.4179,
+      "step": 2770
+    },
+    {
+      "epoch": 0.7934356047092401,
+      "grad_norm": 0.7972052097320557,
+      "learning_rate": 9.895424845879657e-05,
+      "loss": 0.4534,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7962896896182662,
+      "grad_norm": 1.4866856336593628,
+      "learning_rate": 9.893387710462189e-05,
+      "loss": 0.4501,
+      "step": 2790
+    },
+    {
+      "epoch": 0.7991437745272922,
+      "grad_norm": 0.6486208438873291,
+      "learning_rate": 9.891331137944344e-05,
+      "loss": 0.4327,
+      "step": 2800
+    },
+    {
+      "epoch": 0.8019978594363182,
+      "grad_norm": 0.4823531210422516,
+      "learning_rate": 9.889255136495063e-05,
+      "loss": 0.4188,
+      "step": 2810
+    },
+    {
+      "epoch": 0.8048519443453442,
+      "grad_norm": 0.4850338399410248,
+      "learning_rate": 9.887159714360469e-05,
+      "loss": 0.4096,
+      "step": 2820
+    },
+    {
+      "epoch": 0.8077060292543703,
+      "grad_norm": 0.6615185737609863,
+      "learning_rate": 9.885044879863816e-05,
+      "loss": 0.4626,
+      "step": 2830
+    },
+    {
+      "epoch": 0.8105601141633964,
+      "grad_norm": 0.639529824256897,
+      "learning_rate": 9.88291064140547e-05,
+      "loss": 0.4125,
+      "step": 2840
+    },
+    {
+      "epoch": 0.8134141990724224,
+      "grad_norm": 0.6544278860092163,
+      "learning_rate": 9.880757007462876e-05,
+      "loss": 0.4319,
+      "step": 2850
+    },
+    {
+      "epoch": 0.8162682839814485,
+      "grad_norm": 0.8010094165802002,
+      "learning_rate": 9.878583986590513e-05,
+      "loss": 0.4206,
+      "step": 2860
+    },
+    {
+      "epoch": 0.8191223688904745,
+      "grad_norm": 0.9454021453857422,
+      "learning_rate": 9.876391587419871e-05,
+      "loss": 0.4327,
+      "step": 2870
+    },
+    {
+      "epoch": 0.8219764537995006,
+      "grad_norm": 0.5547480583190918,
+      "learning_rate": 9.874179818659415e-05,
+      "loss": 0.448,
+      "step": 2880
+    },
+    {
+      "epoch": 0.8248305387085266,
+      "grad_norm": 0.526059091091156,
+      "learning_rate": 9.871948689094542e-05,
+      "loss": 0.413,
+      "step": 2890
+    },
+    {
+      "epoch": 0.8276846236175526,
+      "grad_norm": 0.5893587470054626,
+      "learning_rate": 9.869698207587558e-05,
+      "loss": 0.4044,
+      "step": 2900
+    },
+    {
+      "epoch": 0.8305387085265786,
+      "grad_norm": 0.5328799486160278,
+      "learning_rate": 9.867428383077637e-05,
+      "loss": 0.4255,
+      "step": 2910
+    },
+    {
+      "epoch": 0.8333927934356047,
+      "grad_norm": 0.4376719295978546,
+      "learning_rate": 9.865139224580779e-05,
+      "loss": 0.3751,
+      "step": 2920
+    },
+    {
+      "epoch": 0.8362468783446307,
+      "grad_norm": 0.9210516214370728,
+      "learning_rate": 9.86283074118979e-05,
+      "loss": 0.4657,
+      "step": 2930
+    },
+    {
+      "epoch": 0.8391009632536568,
+      "grad_norm": 0.5722871422767639,
+      "learning_rate": 9.86050294207423e-05,
+      "loss": 0.435,
+      "step": 2940
+    },
+    {
+      "epoch": 0.8419550481626829,
+      "grad_norm": 0.6614554524421692,
+      "learning_rate": 9.858155836480387e-05,
+      "loss": 0.4466,
+      "step": 2950
+    },
+    {
+      "epoch": 0.8448091330717089,
+      "grad_norm": 0.8073657751083374,
+      "learning_rate": 9.855789433731234e-05,
+      "loss": 0.405,
+      "step": 2960
+    },
+    {
+      "epoch": 0.847663217980735,
+      "grad_norm": 0.5362779498100281,
+      "learning_rate": 9.853403743226393e-05,
+      "loss": 0.4344,
+      "step": 2970
+    },
+    {
+      "epoch": 0.850517302889761,
+      "grad_norm": 0.6456807851791382,
+      "learning_rate": 9.850998774442106e-05,
+      "loss": 0.4305,
+      "step": 2980
+    },
+    {
+      "epoch": 0.853371387798787,
+      "grad_norm": 0.5409958958625793,
+      "learning_rate": 9.848574536931185e-05,
+      "loss": 0.4312,
+      "step": 2990
+    },
+    {
+      "epoch": 0.856225472707813,
+      "grad_norm": 0.5728853940963745,
+      "learning_rate": 9.846131040322983e-05,
+      "loss": 0.4364,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8590795576168391,
+      "grad_norm": 0.5256248116493225,
+      "learning_rate": 9.843668294323348e-05,
+      "loss": 0.4201,
+      "step": 3010
+    },
+    {
+      "epoch": 0.8619336425258651,
+      "grad_norm": 0.8561432957649231,
+      "learning_rate": 9.841186308714589e-05,
+      "loss": 0.4336,
+      "step": 3020
+    },
+    {
+      "epoch": 0.8647877274348912,
+      "grad_norm": 0.7229063510894775,
+      "learning_rate": 9.838685093355444e-05,
+      "loss": 0.407,
+      "step": 3030
+    },
+    {
+      "epoch": 0.8676418123439172,
+      "grad_norm": 0.8342879414558411,
+      "learning_rate": 9.836164658181026e-05,
+      "loss": 0.4489,
+      "step": 3040
+    },
+    {
+      "epoch": 0.8704958972529433,
+      "grad_norm": 0.5657804608345032,
+      "learning_rate": 9.833625013202794e-05,
+      "loss": 0.4169,
+      "step": 3050
+    },
+    {
+      "epoch": 0.8733499821619694,
+      "grad_norm": 0.5953682065010071,
+      "learning_rate": 9.831066168508514e-05,
+      "loss": 0.408,
+      "step": 3060
+    },
+    {
+      "epoch": 0.8762040670709954,
+      "grad_norm": 0.456233948469162,
+      "learning_rate": 9.828488134262208e-05,
+      "loss": 0.4163,
+      "step": 3070
+    },
+    {
+      "epoch": 0.8790581519800214,
+      "grad_norm": 0.7670127749443054,
+      "learning_rate": 9.825890920704126e-05,
+      "loss": 0.4373,
+      "step": 3080
+    },
+    {
+      "epoch": 0.8819122368890474,
+      "grad_norm": 0.9011564254760742,
+      "learning_rate": 9.823274538150702e-05,
+      "loss": 0.3886,
+      "step": 3090
+    },
+    {
+      "epoch": 0.8847663217980735,
+      "grad_norm": 0.5500415563583374,
+      "learning_rate": 9.820638996994509e-05,
+      "loss": 0.4382,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8876204067070995,
+      "grad_norm": 0.6603699922561646,
+      "learning_rate": 9.81798430770422e-05,
+      "loss": 0.4768,
+      "step": 3110
+    },
+    {
+      "epoch": 0.8904744916161256,
+      "grad_norm": 0.43730512261390686,
+      "learning_rate": 9.815310480824564e-05,
+      "loss": 0.3954,
+      "step": 3120
+    },
+    {
+      "epoch": 0.8933285765251516,
+      "grad_norm": 0.5685926675796509,
+      "learning_rate": 9.812617526976295e-05,
+      "loss": 0.4326,
+      "step": 3130
+    },
+    {
+      "epoch": 0.8961826614341777,
+      "grad_norm": 0.4876101016998291,
+      "learning_rate": 9.809905456856135e-05,
+      "loss": 0.4197,
+      "step": 3140
+    },
+    {
+      "epoch": 0.8990367463432037,
+      "grad_norm": 0.6296002864837646,
+      "learning_rate": 9.807174281236735e-05,
+      "loss": 0.4574,
+      "step": 3150
+    },
+    {
+      "epoch": 0.9018908312522298,
+      "grad_norm": 0.5232740640640259,
+      "learning_rate": 9.804424010966644e-05,
+      "loss": 0.4251,
+      "step": 3160
+    },
+    {
+      "epoch": 0.9047449161612559,
+      "grad_norm": 0.42207297682762146,
+      "learning_rate": 9.801654656970252e-05,
+      "loss": 0.3989,
+      "step": 3170
+    },
+    {
+      "epoch": 0.9075990010702818,
+      "grad_norm": 0.6646615266799927,
+      "learning_rate": 9.798866230247752e-05,
+      "loss": 0.4171,
+      "step": 3180
+    },
+    {
+      "epoch": 0.9104530859793079,
+      "grad_norm": 0.6919196248054504,
+      "learning_rate": 9.796058741875095e-05,
+      "loss": 0.4811,
+      "step": 3190
+    },
+    {
+      "epoch": 0.9133071708883339,
+      "grad_norm": 0.5413965582847595,
+      "learning_rate": 9.79323220300395e-05,
+      "loss": 0.4665,
+      "step": 3200
+    },
+    {
+      "epoch": 0.91616125579736,
+      "grad_norm": 0.8154647946357727,
+      "learning_rate": 9.790386624861657e-05,
+      "loss": 0.4236,
+      "step": 3210
+    },
+    {
+      "epoch": 0.919015340706386,
+      "grad_norm": 0.48265111446380615,
+      "learning_rate": 9.787522018751179e-05,
+      "loss": 0.419,
+      "step": 3220
+    },
+    {
+      "epoch": 0.9218694256154121,
+      "grad_norm": 0.7374817728996277,
+      "learning_rate": 9.78463839605106e-05,
+      "loss": 0.4372,
+      "step": 3230
+    },
+    {
+      "epoch": 0.9247235105244381,
+      "grad_norm": 0.7971628904342651,
+      "learning_rate": 9.781735768215385e-05,
+      "loss": 0.4593,
+      "step": 3240
+    },
+    {
+      "epoch": 0.9275775954334642,
+      "grad_norm": 0.49169448018074036,
+      "learning_rate": 9.778814146773727e-05,
+      "loss": 0.4407,
+      "step": 3250
+    },
+    {
+      "epoch": 0.9304316803424902,
+      "grad_norm": 0.8247508406639099,
+      "learning_rate": 9.7758735433311e-05,
+      "loss": 0.3936,
+      "step": 3260
+    },
+    {
+      "epoch": 0.9332857652515162,
+      "grad_norm": 0.4885646402835846,
+      "learning_rate": 9.772913969567923e-05,
+      "loss": 0.4159,
+      "step": 3270
+    },
+    {
+      "epoch": 0.9361398501605422,
+      "grad_norm": 0.5817092657089233,
+      "learning_rate": 9.76993543723996e-05,
+      "loss": 0.4822,
+      "step": 3280
+    },
+    {
+      "epoch": 0.9389939350695683,
+      "grad_norm": 0.5657879114151001,
+      "learning_rate": 9.766937958178293e-05,
+      "loss": 0.4002,
+      "step": 3290
+    },
+    {
+      "epoch": 0.9418480199785944,
+      "grad_norm": 0.6319302916526794,
+      "learning_rate": 9.763921544289245e-05,
+      "loss": 0.4502,
+      "step": 3300
+    },
+    {
+      "epoch": 0.9447021048876204,
+      "grad_norm": 0.7825123071670532,
+      "learning_rate": 9.760886207554367e-05,
+      "loss": 0.426,
+      "step": 3310
+    },
+    {
+      "epoch": 0.9475561897966465,
+      "grad_norm": 0.6912452578544617,
+      "learning_rate": 9.757831960030364e-05,
+      "loss": 0.4202,
+      "step": 3320
+    },
+    {
+      "epoch": 0.9504102747056725,
+      "grad_norm": 0.6349189877510071,
+      "learning_rate": 9.754758813849059e-05,
+      "loss": 0.4484,
+      "step": 3330
+    },
+    {
+      "epoch": 0.9532643596146986,
+      "grad_norm": 0.5008338689804077,
+      "learning_rate": 9.751666781217343e-05,
+      "loss": 0.4398,
+      "step": 3340
+    },
+    {
+      "epoch": 0.9561184445237246,
+      "grad_norm": 0.8321357369422913,
+      "learning_rate": 9.748555874417128e-05,
+      "loss": 0.42,
+      "step": 3350
+    },
+    {
+      "epoch": 0.9589725294327506,
+      "grad_norm": 0.454465389251709,
+      "learning_rate": 9.745426105805293e-05,
+      "loss": 0.3906,
+      "step": 3360
+    },
+    {
+      "epoch": 0.9618266143417766,
+      "grad_norm": 0.48528704047203064,
+      "learning_rate": 9.74227748781364e-05,
+      "loss": 0.4359,
+      "step": 3370
+    },
+    {
+      "epoch": 0.9646806992508027,
+      "grad_norm": 0.47498252987861633,
+      "learning_rate": 9.739110032948844e-05,
+      "loss": 0.3875,
+      "step": 3380
+    },
+    {
+      "epoch": 0.9675347841598287,
+      "grad_norm": 0.5873503088951111,
+      "learning_rate": 9.7359237537924e-05,
+      "loss": 0.4297,
+      "step": 3390
+    },
+    {
+      "epoch": 0.9703888690688548,
+      "grad_norm": 0.47873497009277344,
+      "learning_rate": 9.732718663000577e-05,
+      "loss": 0.4172,
+      "step": 3400
+    },
+    {
+      "epoch": 0.9732429539778809,
+      "grad_norm": 0.5299973487854004,
+      "learning_rate": 9.729494773304364e-05,
+      "loss": 0.4078,
+      "step": 3410
+    },
+    {
+      "epoch": 0.9760970388869069,
+      "grad_norm": 0.5915161967277527,
+      "learning_rate": 9.726252097509423e-05,
+      "loss": 0.4236,
+      "step": 3420
+    },
+    {
+      "epoch": 0.978951123795933,
+      "grad_norm": 0.87859046459198,
+      "learning_rate": 9.722990648496038e-05,
+      "loss": 0.4555,
+      "step": 3430
+    },
+    {
+      "epoch": 0.981805208704959,
+      "grad_norm": 0.6019312739372253,
+      "learning_rate": 9.719710439219058e-05,
+      "loss": 0.4058,
+      "step": 3440
+    },
+    {
+      "epoch": 0.984659293613985,
+      "grad_norm": 0.8020815253257751,
+      "learning_rate": 9.716411482707857e-05,
+      "loss": 0.4319,
+      "step": 3450
+    },
+    {
+      "epoch": 0.987513378523011,
+      "grad_norm": 0.7268204689025879,
+      "learning_rate": 9.713093792066266e-05,
+      "loss": 0.4416,
+      "step": 3460
+    },
+    {
+      "epoch": 0.9903674634320371,
+      "grad_norm": 0.6523804664611816,
+      "learning_rate": 9.709757380472536e-05,
+      "loss": 0.4043,
+      "step": 3470
+    },
+    {
+      "epoch": 0.9932215483410631,
+      "grad_norm": 0.6375195980072021,
+      "learning_rate": 9.706402261179279e-05,
+      "loss": 0.3974,
+      "step": 3480
+    },
+    {
+      "epoch": 0.9960756332500892,
+      "grad_norm": 0.9598755836486816,
+      "learning_rate": 9.703028447513417e-05,
+      "loss": 0.4353,
+      "step": 3490
+    },
+    {
+      "epoch": 0.9989297181591152,
+      "grad_norm": 0.4769824147224426,
+      "learning_rate": 9.699635952876127e-05,
+      "loss": 0.4413,
+      "step": 3500
+    },
+    {
+      "epoch": 1.0017838030681412,
+      "grad_norm": 0.6290175318717957,
+      "learning_rate": 9.696224790742787e-05,
+      "loss": 0.3933,
+      "step": 3510
+    },
+    {
+      "epoch": 1.0046378879771674,
+      "grad_norm": 0.6167151927947998,
+      "learning_rate": 9.692794974662927e-05,
+      "loss": 0.4242,
+      "step": 3520
+    },
+    {
+      "epoch": 1.0074919728861933,
+      "grad_norm": 0.44450104236602783,
+      "learning_rate": 9.689346518260174e-05,
+      "loss": 0.3619,
+      "step": 3530
+    },
+    {
+      "epoch": 1.0103460577952195,
+      "grad_norm": 0.9434749484062195,
+      "learning_rate": 9.685879435232193e-05,
+      "loss": 0.43,
+      "step": 3540
+    },
+    {
+      "epoch": 1.0132001427042454,
+      "grad_norm": 0.479621022939682,
+      "learning_rate": 9.682393739350637e-05,
+      "loss": 0.4174,
+      "step": 3550
+    },
+    {
+      "epoch": 1.0160542276132716,
+      "grad_norm": 0.5258568525314331,
+      "learning_rate": 9.678889444461093e-05,
+      "loss": 0.434,
+      "step": 3560
+    },
+    {
+      "epoch": 1.0189083125222975,
+      "grad_norm": 0.4461469054222107,
+      "learning_rate": 9.675366564483026e-05,
+      "loss": 0.3974,
+      "step": 3570
+    },
+    {
+      "epoch": 1.0217623974313237,
+      "grad_norm": 0.38621410727500916,
+      "learning_rate": 9.671825113409718e-05,
+      "loss": 0.4231,
+      "step": 3580
+    },
+    {
+      "epoch": 1.0246164823403496,
+      "grad_norm": 0.6458470821380615,
+      "learning_rate": 9.668265105308223e-05,
+      "loss": 0.4388,
+      "step": 3590
+    },
+    {
+      "epoch": 1.0274705672493756,
+      "grad_norm": 0.9359676241874695,
+      "learning_rate": 9.664686554319306e-05,
+      "loss": 0.4499,
+      "step": 3600
+    },
+    {
+      "epoch": 1.0303246521584017,
+      "grad_norm": 0.5093625783920288,
+      "learning_rate": 9.661089474657381e-05,
+      "loss": 0.3919,
+      "step": 3610
+    },
+    {
+      "epoch": 1.0331787370674277,
+      "grad_norm": 0.6743724346160889,
+      "learning_rate": 9.657473880610463e-05,
+      "loss": 0.4199,
+      "step": 3620
+    },
+    {
+      "epoch": 1.0360328219764539,
+      "grad_norm": 0.774372935295105,
+      "learning_rate": 9.653839786540114e-05,
+      "loss": 0.3975,
+      "step": 3630
+    },
+    {
+      "epoch": 1.0388869068854798,
+      "grad_norm": 0.6048170924186707,
+      "learning_rate": 9.650187206881368e-05,
+      "loss": 0.4094,
+      "step": 3640
+    },
+    {
+      "epoch": 1.041740991794506,
+      "grad_norm": 0.706270158290863,
+      "learning_rate": 9.646516156142695e-05,
+      "loss": 0.4304,
+      "step": 3650
+    },
+    {
+      "epoch": 1.044595076703532,
+      "grad_norm": 0.6507980823516846,
+      "learning_rate": 9.642826648905934e-05,
+      "loss": 0.4429,
+      "step": 3660
+    },
+    {
+      "epoch": 1.047449161612558,
+      "grad_norm": 0.7644017338752747,
+      "learning_rate": 9.639118699826231e-05,
+      "loss": 0.4192,
+      "step": 3670
+    },
+    {
+      "epoch": 1.050303246521584,
+      "grad_norm": 0.9183304905891418,
+      "learning_rate": 9.635392323631988e-05,
+      "loss": 0.4053,
+      "step": 3680
+    },
+    {
+      "epoch": 1.05315733143061,
+      "grad_norm": 0.6468967199325562,
+      "learning_rate": 9.631647535124802e-05,
+      "loss": 0.3995,
+      "step": 3690
+    },
+    {
+      "epoch": 1.0560114163396361,
+      "grad_norm": 1.1233046054840088,
+      "learning_rate": 9.627884349179402e-05,
+      "loss": 0.4151,
+      "step": 3700
+    },
+    {
+      "epoch": 1.058865501248662,
+      "grad_norm": 0.5151073932647705,
+      "learning_rate": 9.6241027807436e-05,
+      "loss": 0.4262,
+      "step": 3710
+    },
+    {
+      "epoch": 1.0617195861576882,
+      "grad_norm": 0.8094540238380432,
+      "learning_rate": 9.62030284483822e-05,
+      "loss": 0.4258,
+      "step": 3720
+    },
+    {
+      "epoch": 1.0645736710667142,
+      "grad_norm": 0.6203951239585876,
+      "learning_rate": 9.616484556557048e-05,
+      "loss": 0.4371,
+      "step": 3730
+    },
+    {
+      "epoch": 1.0674277559757404,
+      "grad_norm": 0.5027265548706055,
+      "learning_rate": 9.612647931066763e-05,
+      "loss": 0.4004,
+      "step": 3740
+    },
+    {
+      "epoch": 1.0702818408847663,
+      "grad_norm": 0.4848248362541199,
+      "learning_rate": 9.608792983606885e-05,
+      "loss": 0.3917,
+      "step": 3750
+    },
+    {
+      "epoch": 1.0731359257937925,
+      "grad_norm": 0.6018698811531067,
+      "learning_rate": 9.604919729489708e-05,
+      "loss": 0.3912,
+      "step": 3760
+    },
+    {
+      "epoch": 1.0759900107028184,
+      "grad_norm": 0.752939760684967,
+      "learning_rate": 9.601028184100246e-05,
+      "loss": 0.432,
+      "step": 3770
+    },
+    {
+      "epoch": 1.0788440956118444,
+      "grad_norm": 0.48865851759910583,
+      "learning_rate": 9.597118362896169e-05,
+      "loss": 0.416,
+      "step": 3780
+    },
+    {
+      "epoch": 1.0816981805208705,
+      "grad_norm": 0.7386459112167358,
+      "learning_rate": 9.593190281407731e-05,
+      "loss": 0.4753,
+      "step": 3790
+    },
+    {
+      "epoch": 1.0845522654298965,
+      "grad_norm": 0.9361982345581055,
+      "learning_rate": 9.589243955237731e-05,
+      "loss": 0.4187,
+      "step": 3800
+    },
+    {
+      "epoch": 1.0874063503389226,
+      "grad_norm": 0.6894406080245972,
+      "learning_rate": 9.585279400061427e-05,
+      "loss": 0.3709,
+      "step": 3810
+    },
+    {
+      "epoch": 1.0902604352479486,
+      "grad_norm": 0.6570184826850891,
+      "learning_rate": 9.581296631626493e-05,
+      "loss": 0.37,
+      "step": 3820
+    },
+    {
+      "epoch": 1.0931145201569747,
+      "grad_norm": 0.5192276835441589,
+      "learning_rate": 9.577295665752943e-05,
+      "loss": 0.3979,
+      "step": 3830
+    },
+    {
+      "epoch": 1.0959686050660007,
+      "grad_norm": 0.6113749146461487,
+      "learning_rate": 9.573276518333074e-05,
+      "loss": 0.3881,
+      "step": 3840
+    },
+    {
+      "epoch": 1.0988226899750269,
+      "grad_norm": 0.8150345683097839,
+      "learning_rate": 9.569239205331405e-05,
+      "loss": 0.4079,
+      "step": 3850
+    },
+    {
+      "epoch": 1.1016767748840528,
+      "grad_norm": 0.5030243396759033,
+      "learning_rate": 9.565183742784607e-05,
+      "loss": 0.3626,
+      "step": 3860
+    },
+    {
+      "epoch": 1.1045308597930787,
+      "grad_norm": 0.7989709973335266,
+      "learning_rate": 9.561110146801447e-05,
+      "loss": 0.4244,
+      "step": 3870
+    },
+    {
+      "epoch": 1.107384944702105,
+      "grad_norm": 0.8960319757461548,
+      "learning_rate": 9.557018433562716e-05,
+      "loss": 0.4241,
+      "step": 3880
+    },
+    {
+      "epoch": 1.1102390296111309,
+      "grad_norm": 0.7229114770889282,
+      "learning_rate": 9.552908619321173e-05,
+      "loss": 0.3973,
+      "step": 3890
+    },
+    {
+      "epoch": 1.113093114520157,
+      "grad_norm": 0.45926737785339355,
+      "learning_rate": 9.548780720401473e-05,
+      "loss": 0.4087,
+      "step": 3900
+    },
+    {
+      "epoch": 1.115947199429183,
+      "grad_norm": 0.616928219795227,
+      "learning_rate": 9.544634753200107e-05,
+      "loss": 0.4076,
+      "step": 3910
+    },
+    {
+      "epoch": 1.1188012843382091,
+      "grad_norm": 0.8416487574577332,
+      "learning_rate": 9.540470734185334e-05,
+      "loss": 0.3991,
+      "step": 3920
+    },
+    {
+      "epoch": 1.121655369247235,
+      "grad_norm": 0.5629198551177979,
+      "learning_rate": 9.536288679897118e-05,
+      "loss": 0.4197,
+      "step": 3930
+    },
+    {
+      "epoch": 1.1245094541562612,
+      "grad_norm": 0.652466356754303,
+      "learning_rate": 9.532088606947063e-05,
+      "loss": 0.3913,
+      "step": 3940
+    },
+    {
+      "epoch": 1.1273635390652872,
+      "grad_norm": 0.6805722713470459,
+      "learning_rate": 9.527870532018342e-05,
+      "loss": 0.419,
+      "step": 3950
+    },
+    {
+      "epoch": 1.1302176239743131,
+      "grad_norm": 0.8768517971038818,
+      "learning_rate": 9.523634471865632e-05,
+      "loss": 0.4223,
+      "step": 3960
+    },
+    {
+      "epoch": 1.1330717088833393,
+      "grad_norm": 0.8079743981361389,
+      "learning_rate": 9.519380443315058e-05,
+      "loss": 0.4486,
+      "step": 3970
+    },
+    {
+      "epoch": 1.1359257937923652,
+      "grad_norm": 0.6022933125495911,
+      "learning_rate": 9.515108463264109e-05,
+      "loss": 0.4504,
+      "step": 3980
+    },
+    {
+      "epoch": 1.1387798787013914,
+      "grad_norm": 0.7014849185943604,
+      "learning_rate": 9.510818548681582e-05,
+      "loss": 0.4294,
+      "step": 3990
+    },
+    {
+      "epoch": 1.1416339636104174,
+      "grad_norm": 0.8471984267234802,
+      "learning_rate": 9.506510716607513e-05,
+      "loss": 0.4264,
+      "step": 4000
+    },
+    {
+      "epoch": 1.1444880485194435,
+      "grad_norm": 0.8326864838600159,
+      "learning_rate": 9.502184984153109e-05,
+      "loss": 0.416,
+      "step": 4010
+    },
+    {
+      "epoch": 1.1473421334284695,
+      "grad_norm": 0.614597499370575,
+      "learning_rate": 9.497841368500675e-05,
+      "loss": 0.4414,
+      "step": 4020
+    },
+    {
+      "epoch": 1.1501962183374956,
+      "grad_norm": 0.8020027875900269,
+      "learning_rate": 9.493479886903558e-05,
+      "loss": 0.4162,
+      "step": 4030
+    },
+    {
+      "epoch": 1.1530503032465216,
+      "grad_norm": 0.6523504257202148,
+      "learning_rate": 9.489100556686064e-05,
+      "loss": 0.4361,
+      "step": 4040
+    },
+    {
+      "epoch": 1.1559043881555477,
+      "grad_norm": 0.4413366913795471,
+      "learning_rate": 9.484703395243396e-05,
+      "loss": 0.4152,
+      "step": 4050
+    },
+    {
+      "epoch": 1.1587584730645737,
+      "grad_norm": 0.8899474740028381,
+      "learning_rate": 9.480288420041589e-05,
+      "loss": 0.4572,
+      "step": 4060
+    },
+    {
+      "epoch": 1.1616125579735996,
+      "grad_norm": 0.8028743267059326,
+      "learning_rate": 9.475855648617432e-05,
+      "loss": 0.4541,
+      "step": 4070
+    },
+    {
+      "epoch": 1.1644666428826258,
+      "grad_norm": 0.7049791216850281,
+      "learning_rate": 9.471405098578405e-05,
+      "loss": 0.4089,
+      "step": 4080
+    },
+    {
+      "epoch": 1.1673207277916517,
+      "grad_norm": 0.7701025009155273,
+      "learning_rate": 9.466936787602607e-05,
+      "loss": 0.4132,
+      "step": 4090
+    },
+    {
+      "epoch": 1.170174812700678,
+      "grad_norm": 0.48589470982551575,
+      "learning_rate": 9.462450733438682e-05,
+      "loss": 0.4109,
+      "step": 4100
+    },
+    {
+      "epoch": 1.1730288976097039,
+      "grad_norm": 0.7822790145874023,
+      "learning_rate": 9.457946953905757e-05,
+      "loss": 0.4341,
+      "step": 4110
+    },
+    {
+      "epoch": 1.17588298251873,
+      "grad_norm": 0.7849439382553101,
+      "learning_rate": 9.453425466893362e-05,
+      "loss": 0.4129,
+      "step": 4120
+    },
+    {
+      "epoch": 1.178737067427756,
+      "grad_norm": 0.6041362285614014,
+      "learning_rate": 9.448886290361366e-05,
+      "loss": 0.4222,
+      "step": 4130
+    },
+    {
+      "epoch": 1.181591152336782,
+      "grad_norm": 0.5522482991218567,
+      "learning_rate": 9.444329442339901e-05,
+      "loss": 0.3912,
+      "step": 4140
+    },
+    {
+      "epoch": 1.184445237245808,
+      "grad_norm": 0.6508781909942627,
+      "learning_rate": 9.439754940929295e-05,
+      "loss": 0.4302,
+      "step": 4150
+    },
+    {
+      "epoch": 1.187299322154834,
+      "grad_norm": 0.568966805934906,
+      "learning_rate": 9.435162804299991e-05,
+      "loss": 0.4466,
+      "step": 4160
+    },
+    {
+      "epoch": 1.1901534070638602,
+      "grad_norm": 0.4138081669807434,
+      "learning_rate": 9.43055305069249e-05,
+      "loss": 0.3965,
+      "step": 4170
+    },
+    {
+      "epoch": 1.1930074919728861,
+      "grad_norm": 0.6189876794815063,
+      "learning_rate": 9.425925698417262e-05,
+      "loss": 0.419,
+      "step": 4180
+    },
+    {
+      "epoch": 1.1958615768819123,
+      "grad_norm": 0.6760974526405334,
+      "learning_rate": 9.421280765854685e-05,
+      "loss": 0.4319,
+      "step": 4190
+    },
+    {
+      "epoch": 1.1987156617909382,
+      "grad_norm": 0.7758780717849731,
+      "learning_rate": 9.416618271454967e-05,
+      "loss": 0.4266,
+      "step": 4200
+    },
+    {
+      "epoch": 1.2015697466999644,
+      "grad_norm": 0.6547653079032898,
+      "learning_rate": 9.411938233738073e-05,
+      "loss": 0.4228,
+      "step": 4210
+    },
+    {
+      "epoch": 1.2044238316089904,
+      "grad_norm": 1.2232414484024048,
+      "learning_rate": 9.407240671293654e-05,
+      "loss": 0.4049,
+      "step": 4220
+    },
+    {
+      "epoch": 1.2072779165180165,
+      "grad_norm": 0.5262352228164673,
+      "learning_rate": 9.402525602780968e-05,
+      "loss": 0.4029,
+      "step": 4230
+    },
+    {
+      "epoch": 1.2101320014270425,
+      "grad_norm": 0.5081900954246521,
+      "learning_rate": 9.397793046928811e-05,
+      "loss": 0.4083,
+      "step": 4240
+    },
+    {
+      "epoch": 1.2129860863360684,
+      "grad_norm": 0.5323736071586609,
+      "learning_rate": 9.393043022535442e-05,
+      "loss": 0.3798,
+      "step": 4250
+    },
+    {
+      "epoch": 1.2158401712450946,
+      "grad_norm": 0.6737754940986633,
+      "learning_rate": 9.388275548468506e-05,
+      "loss": 0.3894,
+      "step": 4260
+    },
+    {
+      "epoch": 1.2186942561541205,
+      "grad_norm": 1.1628127098083496,
+      "learning_rate": 9.383490643664959e-05,
+      "loss": 0.4055,
+      "step": 4270
+    },
+    {
+      "epoch": 1.2215483410631467,
+      "grad_norm": 0.7762293815612793,
+      "learning_rate": 9.378688327130992e-05,
+      "loss": 0.4335,
+      "step": 4280
+    },
+    {
+      "epoch": 1.2244024259721726,
+      "grad_norm": 0.7017742991447449,
+      "learning_rate": 9.373868617941966e-05,
+      "loss": 0.504,
+      "step": 4290
+    },
+    {
+      "epoch": 1.2272565108811988,
+      "grad_norm": 0.7783657312393188,
+      "learning_rate": 9.369031535242316e-05,
+      "loss": 0.3922,
+      "step": 4300
+    },
+    {
+      "epoch": 1.2301105957902247,
+      "grad_norm": 0.7561664581298828,
+      "learning_rate": 9.364177098245497e-05,
+      "loss": 0.4178,
+      "step": 4310
+    },
+    {
+      "epoch": 1.2329646806992507,
+      "grad_norm": 0.5814104080200195,
+      "learning_rate": 9.35930532623389e-05,
+      "loss": 0.3999,
+      "step": 4320
+    },
+    {
+      "epoch": 1.2358187656082769,
+      "grad_norm": 0.8447505235671997,
+      "learning_rate": 9.354416238558736e-05,
+      "loss": 0.4427,
+      "step": 4330
+    },
+    {
+      "epoch": 1.2386728505173028,
+      "grad_norm": 0.7698172330856323,
+      "learning_rate": 9.349509854640054e-05,
+      "loss": 0.422,
+      "step": 4340
+    },
+    {
+      "epoch": 1.241526935426329,
+      "grad_norm": 1.1690415143966675,
+      "learning_rate": 9.344586193966569e-05,
+      "loss": 0.4099,
+      "step": 4350
+    },
+    {
+      "epoch": 1.244381020335355,
+      "grad_norm": 0.8412760496139526,
+      "learning_rate": 9.339645276095627e-05,
+      "loss": 0.405,
+      "step": 4360
+    },
+    {
+      "epoch": 1.247235105244381,
+      "grad_norm": 0.6715117692947388,
+      "learning_rate": 9.334687120653127e-05,
+      "loss": 0.4097,
+      "step": 4370
+    },
+    {
+      "epoch": 1.250089190153407,
+      "grad_norm": 1.0793496370315552,
+      "learning_rate": 9.329711747333431e-05,
+      "loss": 0.4117,
+      "step": 4380
+    },
+    {
+      "epoch": 1.252943275062433,
+      "grad_norm": 0.8151715993881226,
+      "learning_rate": 9.324719175899297e-05,
+      "loss": 0.3923,
+      "step": 4390
+    },
+    {
+      "epoch": 1.2557973599714591,
+      "grad_norm": 0.5719575881958008,
+      "learning_rate": 9.319709426181792e-05,
+      "loss": 0.3947,
+      "step": 4400
+    },
+    {
+      "epoch": 1.2586514448804853,
+      "grad_norm": 0.5285683870315552,
+      "learning_rate": 9.314682518080225e-05,
+      "loss": 0.4248,
+      "step": 4410
+    },
+    {
+      "epoch": 1.2615055297895112,
+      "grad_norm": 0.6833191514015198,
+      "learning_rate": 9.30963847156205e-05,
+      "loss": 0.4302,
+      "step": 4420
+    },
+    {
+      "epoch": 1.2643596146985372,
+      "grad_norm": 0.7360308766365051,
+      "learning_rate": 9.304577306662803e-05,
+      "loss": 0.4655,
+      "step": 4430
+    },
+    {
+      "epoch": 1.2672136996075634,
+      "grad_norm": 1.0034043788909912,
+      "learning_rate": 9.299499043486017e-05,
+      "loss": 0.4081,
+      "step": 4440
+    },
+    {
+      "epoch": 1.2700677845165893,
+      "grad_norm": 0.5989054441452026,
+      "learning_rate": 9.294403702203135e-05,
+      "loss": 0.441,
+      "step": 4450
+    },
+    {
+      "epoch": 1.2729218694256155,
+      "grad_norm": 0.7070958614349365,
+      "learning_rate": 9.289291303053443e-05,
+      "loss": 0.4306,
+      "step": 4460
+    },
+    {
+      "epoch": 1.2757759543346414,
+      "grad_norm": 0.9389038681983948,
+      "learning_rate": 9.284161866343979e-05,
+      "loss": 0.4448,
+      "step": 4470
+    },
+    {
+      "epoch": 1.2786300392436676,
+      "grad_norm": 0.5640192031860352,
+      "learning_rate": 9.279015412449456e-05,
+      "loss": 0.4241,
+      "step": 4480
+    },
+    {
+      "epoch": 1.2814841241526935,
+      "grad_norm": 0.5926091074943542,
+      "learning_rate": 9.273851961812186e-05,
+      "loss": 0.4375,
+      "step": 4490
+    },
+    {
+      "epoch": 1.2843382090617195,
+      "grad_norm": 0.4756394922733307,
+      "learning_rate": 9.268671534941988e-05,
+      "loss": 0.4207,
+      "step": 4500
+    },
+    {
+      "epoch": 1.2871922939707456,
+      "grad_norm": 0.6925960779190063,
+      "learning_rate": 9.263474152416116e-05,
+      "loss": 0.4419,
+      "step": 4510
+    },
+    {
+      "epoch": 1.2900463788797718,
+      "grad_norm": 0.49510657787323,
+      "learning_rate": 9.258259834879173e-05,
+      "loss": 0.3977,
+      "step": 4520
+    },
+    {
+      "epoch": 1.2929004637887977,
+      "grad_norm": 0.8859143257141113,
+      "learning_rate": 9.253028603043028e-05,
+      "loss": 0.451,
+      "step": 4530
+    },
+    {
+      "epoch": 1.2957545486978237,
+      "grad_norm": 0.5511392951011658,
+      "learning_rate": 9.247780477686738e-05,
+      "loss": 0.3907,
+      "step": 4540
+    },
+    {
+      "epoch": 1.2986086336068499,
+      "grad_norm": 0.7438629865646362,
+      "learning_rate": 9.242515479656461e-05,
+      "loss": 0.3776,
+      "step": 4550
+    },
+    {
+      "epoch": 1.3014627185158758,
+      "grad_norm": 0.6656787991523743,
+      "learning_rate": 9.23723362986538e-05,
+      "loss": 0.3857,
+      "step": 4560
+    },
+    {
+      "epoch": 1.304316803424902,
+      "grad_norm": 0.5865601897239685,
+      "learning_rate": 9.23193494929361e-05,
+      "loss": 0.4073,
+      "step": 4570
+    },
+    {
+      "epoch": 1.307170888333928,
+      "grad_norm": 0.8834959268569946,
+      "learning_rate": 9.22661945898812e-05,
+      "loss": 0.4381,
+      "step": 4580
+    },
+    {
+      "epoch": 1.310024973242954,
+      "grad_norm": 0.7570575475692749,
+      "learning_rate": 9.22128718006265e-05,
+      "loss": 0.4125,
+      "step": 4590
+    },
+    {
+      "epoch": 1.31287905815198,
+      "grad_norm": 0.668268084526062,
+      "learning_rate": 9.21593813369763e-05,
+      "loss": 0.4132,
+      "step": 4600
+    },
+    {
+      "epoch": 1.315733143061006,
+      "grad_norm": 0.9750699400901794,
+      "learning_rate": 9.210572341140086e-05,
+      "loss": 0.3826,
+      "step": 4610
+    },
+    {
+      "epoch": 1.3185872279700321,
+      "grad_norm": 0.6271708607673645,
+      "learning_rate": 9.205189823703569e-05,
+      "loss": 0.4299,
+      "step": 4620
+    },
+    {
+      "epoch": 1.321441312879058,
+      "grad_norm": 0.7680832743644714,
+      "learning_rate": 9.199790602768053e-05,
+      "loss": 0.4203,
+      "step": 4630
+    },
+    {
+      "epoch": 1.3242953977880842,
+      "grad_norm": 0.4561271667480469,
+      "learning_rate": 9.194374699779871e-05,
+      "loss": 0.3668,
+      "step": 4640
+    },
+    {
+      "epoch": 1.3271494826971102,
+      "grad_norm": 0.7475771903991699,
+      "learning_rate": 9.188942136251613e-05,
+      "loss": 0.4024,
+      "step": 4650
+    },
+    {
+      "epoch": 1.3300035676061364,
+      "grad_norm": 0.5157244205474854,
+      "learning_rate": 9.183492933762048e-05,
+      "loss": 0.4042,
+      "step": 4660
+    },
+    {
+      "epoch": 1.3328576525151623,
+      "grad_norm": 0.55656498670578,
+      "learning_rate": 9.178027113956035e-05,
+      "loss": 0.4231,
+      "step": 4670
+    },
+    {
+      "epoch": 1.3357117374241883,
+      "grad_norm": 0.8296566605567932,
+      "learning_rate": 9.172544698544445e-05,
+      "loss": 0.4136,
+      "step": 4680
+    },
+    {
+      "epoch": 1.3385658223332144,
+      "grad_norm": 0.8299556970596313,
+      "learning_rate": 9.16704570930406e-05,
+      "loss": 0.422,
+      "step": 4690
+    },
+    {
+      "epoch": 1.3414199072422406,
+      "grad_norm": 0.804410994052887,
+      "learning_rate": 9.161530168077503e-05,
+      "loss": 0.3665,
+      "step": 4700
+    },
+    {
+      "epoch": 1.3442739921512665,
+      "grad_norm": 0.5750541090965271,
+      "learning_rate": 9.155998096773137e-05,
+      "loss": 0.4127,
+      "step": 4710
+    },
+    {
+      "epoch": 1.3471280770602925,
+      "grad_norm": 0.5164769887924194,
+      "learning_rate": 9.150449517364988e-05,
+      "loss": 0.4048,
+      "step": 4720
+    },
+    {
+      "epoch": 1.3499821619693186,
+      "grad_norm": 0.7072139382362366,
+      "learning_rate": 9.144884451892655e-05,
+      "loss": 0.4315,
+      "step": 4730
+    },
+    {
+      "epoch": 1.3528362468783446,
+      "grad_norm": 0.825655996799469,
+      "learning_rate": 9.139302922461218e-05,
+      "loss": 0.4242,
+      "step": 4740
+    },
+    {
+      "epoch": 1.3556903317873708,
+      "grad_norm": 0.6219303607940674,
+      "learning_rate": 9.133704951241156e-05,
+      "loss": 0.4023,
+      "step": 4750
+    },
+    {
+      "epoch": 1.3585444166963967,
+      "grad_norm": 0.5824173092842102,
+      "learning_rate": 9.128090560468256e-05,
+      "loss": 0.4088,
+      "step": 4760
+    },
+    {
+      "epoch": 1.3613985016054229,
+      "grad_norm": 0.6774200201034546,
+      "learning_rate": 9.122459772443527e-05,
+      "loss": 0.4003,
+      "step": 4770
+    },
+    {
+      "epoch": 1.3642525865144488,
+      "grad_norm": 0.8279693722724915,
+      "learning_rate": 9.116812609533107e-05,
+      "loss": 0.3765,
+      "step": 4780
+    },
+    {
+      "epoch": 1.3671066714234748,
+      "grad_norm": 0.9222491383552551,
+      "learning_rate": 9.111149094168181e-05,
+      "loss": 0.4227,
+      "step": 4790
+    },
+    {
+      "epoch": 1.369960756332501,
+      "grad_norm": 0.7028577327728271,
+      "learning_rate": 9.105469248844883e-05,
+      "loss": 0.4406,
+      "step": 4800
+    },
+    {
+      "epoch": 1.3728148412415269,
+      "grad_norm": 0.4629408121109009,
+      "learning_rate": 9.099773096124213e-05,
+      "loss": 0.3831,
+      "step": 4810
+    },
+    {
+      "epoch": 1.375668926150553,
+      "grad_norm": 0.606282651424408,
+      "learning_rate": 9.094060658631948e-05,
+      "loss": 0.3815,
+      "step": 4820
+    },
+    {
+      "epoch": 1.378523011059579,
+      "grad_norm": 0.5601187348365784,
+      "learning_rate": 9.08833195905855e-05,
+      "loss": 0.3884,
+      "step": 4830
+    },
+    {
+      "epoch": 1.3813770959686051,
+      "grad_norm": 0.5658257007598877,
+      "learning_rate": 9.082587020159073e-05,
+      "loss": 0.4015,
+      "step": 4840
+    },
+    {
+      "epoch": 1.384231180877631,
+      "grad_norm": 0.7560166120529175,
+      "learning_rate": 9.076825864753074e-05,
+      "loss": 0.3553,
+      "step": 4850
+    },
+    {
+      "epoch": 1.387085265786657,
+      "grad_norm": 0.4828161299228668,
+      "learning_rate": 9.071048515724531e-05,
+      "loss": 0.4175,
+      "step": 4860
+    },
+    {
+      "epoch": 1.3899393506956832,
+      "grad_norm": 0.8136508464813232,
+      "learning_rate": 9.06525499602174e-05,
+      "loss": 0.455,
+      "step": 4870
+    },
+    {
+      "epoch": 1.3927934356047094,
+      "grad_norm": 0.41002386808395386,
+      "learning_rate": 9.059445328657228e-05,
+      "loss": 0.3639,
+      "step": 4880
+    },
+    {
+      "epoch": 1.3956475205137353,
+      "grad_norm": 0.8927071690559387,
+      "learning_rate": 9.053619536707664e-05,
+      "loss": 0.436,
+      "step": 4890
+    },
+    {
+      "epoch": 1.3985016054227613,
+      "grad_norm": 1.0075316429138184,
+      "learning_rate": 9.047777643313767e-05,
+      "loss": 0.4339,
+      "step": 4900
+    },
+    {
+      "epoch": 1.4013556903317874,
+      "grad_norm": 0.7862810492515564,
+      "learning_rate": 9.04191967168021e-05,
+      "loss": 0.4365,
+      "step": 4910
+    },
+    {
+      "epoch": 1.4042097752408134,
+      "grad_norm": 0.9463781714439392,
+      "learning_rate": 9.036045645075532e-05,
+      "loss": 0.4307,
+      "step": 4920
+    },
+    {
+      "epoch": 1.4070638601498395,
+      "grad_norm": 0.9278760552406311,
+      "learning_rate": 9.030155586832044e-05,
+      "loss": 0.3814,
+      "step": 4930
+    },
+    {
+      "epoch": 1.4099179450588655,
+      "grad_norm": 0.7348300814628601,
+      "learning_rate": 9.024249520345738e-05,
+      "loss": 0.4019,
+      "step": 4940
+    },
+    {
+      "epoch": 1.4127720299678916,
+      "grad_norm": 0.5968047380447388,
+      "learning_rate": 9.01832746907619e-05,
+      "loss": 0.439,
+      "step": 4950
+    },
+    {
+      "epoch": 1.4156261148769176,
+      "grad_norm": 0.527391791343689,
+      "learning_rate": 9.012389456546473e-05,
+      "loss": 0.3767,
+      "step": 4960
+    },
+    {
+      "epoch": 1.4184801997859435,
+      "grad_norm": 0.5745949149131775,
+      "learning_rate": 9.006435506343054e-05,
+      "loss": 0.4271,
+      "step": 4970
+    },
+    {
+      "epoch": 1.4213342846949697,
+      "grad_norm": 0.6323638558387756,
+      "learning_rate": 9.000465642115712e-05,
+      "loss": 0.3926,
+      "step": 4980
+    },
+    {
+      "epoch": 1.4241883696039956,
+      "grad_norm": 0.5993567109107971,
+      "learning_rate": 8.994479887577436e-05,
+      "loss": 0.3986,
+      "step": 4990
+    },
+    {
+      "epoch": 1.4270424545130218,
+      "grad_norm": 0.43338602781295776,
+      "learning_rate": 8.988478266504334e-05,
+      "loss": 0.373,
+      "step": 5000
+    },
+    {
+      "epoch": 1.4298965394220478,
+      "grad_norm": 0.5481635928153992,
+      "learning_rate": 8.982460802735535e-05,
+      "loss": 0.3705,
+      "step": 5010
+    },
+    {
+      "epoch": 1.432750624331074,
+      "grad_norm": 0.7743555903434753,
+      "learning_rate": 8.976427520173102e-05,
+      "loss": 0.3915,
+      "step": 5020
+    },
+    {
+      "epoch": 1.4356047092400999,
+      "grad_norm": 0.540327787399292,
+      "learning_rate": 8.970378442781927e-05,
+      "loss": 0.4166,
+      "step": 5030
+    },
+    {
+      "epoch": 1.4384587941491258,
+      "grad_norm": 0.6502352356910706,
+      "learning_rate": 8.964313594589645e-05,
+      "loss": 0.4134,
+      "step": 5040
+    },
+    {
+      "epoch": 1.441312879058152,
+      "grad_norm": 0.5395691394805908,
+      "learning_rate": 8.958232999686531e-05,
+      "loss": 0.4514,
+      "step": 5050
+    },
+    {
+      "epoch": 1.4441669639671781,
+      "grad_norm": 0.5091240406036377,
+      "learning_rate": 8.952136682225409e-05,
+      "loss": 0.4228,
+      "step": 5060
+    },
+    {
+      "epoch": 1.447021048876204,
+      "grad_norm": 0.7273949384689331,
+      "learning_rate": 8.946024666421557e-05,
+      "loss": 0.4402,
+      "step": 5070
+    },
+    {
+      "epoch": 1.44987513378523,
+      "grad_norm": 0.7087104916572571,
+      "learning_rate": 8.939896976552605e-05,
+      "loss": 0.3857,
+      "step": 5080
+    },
+    {
+      "epoch": 1.4527292186942562,
+      "grad_norm": 0.6482439041137695,
+      "learning_rate": 8.933753636958444e-05,
+      "loss": 0.4174,
+      "step": 5090
+    },
+    {
+      "epoch": 1.4555833036032821,
+      "grad_norm": 0.5020700693130493,
+      "learning_rate": 8.927594672041128e-05,
+      "loss": 0.4244,
+      "step": 5100
+    },
+    {
+      "epoch": 1.4584373885123083,
+      "grad_norm": 0.49530014395713806,
+      "learning_rate": 8.921420106264778e-05,
+      "loss": 0.4364,
+      "step": 5110
+    },
+    {
+      "epoch": 1.4612914734213343,
+      "grad_norm": 0.5399980545043945,
+      "learning_rate": 8.915229964155476e-05,
+      "loss": 0.3733,
+      "step": 5120
+    },
+    {
+      "epoch": 1.4641455583303604,
+      "grad_norm": 0.6927512288093567,
+      "learning_rate": 8.909024270301185e-05,
+      "loss": 0.4503,
+      "step": 5130
+    },
+    {
+      "epoch": 1.4669996432393864,
+      "grad_norm": 0.6361355781555176,
+      "learning_rate": 8.902803049351635e-05,
+      "loss": 0.4405,
+      "step": 5140
+    },
+    {
+      "epoch": 1.4698537281484123,
+      "grad_norm": 0.6102903485298157,
+      "learning_rate": 8.896566326018234e-05,
+      "loss": 0.4239,
+      "step": 5150
+    },
+    {
+      "epoch": 1.4727078130574385,
+      "grad_norm": 0.8086116313934326,
+      "learning_rate": 8.890314125073966e-05,
+      "loss": 0.4172,
+      "step": 5160
+    },
+    {
+      "epoch": 1.4755618979664644,
+      "grad_norm": 0.7844595909118652,
+      "learning_rate": 8.884046471353295e-05,
+      "loss": 0.409,
+      "step": 5170
+    },
+    {
+      "epoch": 1.4784159828754906,
+      "grad_norm": 0.5432035326957703,
+      "learning_rate": 8.877763389752064e-05,
+      "loss": 0.4197,
+      "step": 5180
+    },
+    {
+      "epoch": 1.4812700677845165,
+      "grad_norm": 0.6236791014671326,
+      "learning_rate": 8.871464905227397e-05,
+      "loss": 0.3988,
+      "step": 5190
+    },
+    {
+      "epoch": 1.4841241526935427,
+      "grad_norm": 0.5851837992668152,
+      "learning_rate": 8.865151042797601e-05,
+      "loss": 0.3919,
+      "step": 5200
+    },
+    {
+      "epoch": 1.4869782376025686,
+      "grad_norm": 0.6688551902770996,
+      "learning_rate": 8.858821827542067e-05,
+      "loss": 0.4116,
+      "step": 5210
+    },
+    {
+      "epoch": 1.4898323225115946,
+      "grad_norm": 0.574059784412384,
+      "learning_rate": 8.852477284601166e-05,
+      "loss": 0.3714,
+      "step": 5220
+    },
+    {
+      "epoch": 1.4926864074206208,
+      "grad_norm": 0.695894718170166,
+      "learning_rate": 8.846117439176158e-05,
+      "loss": 0.4246,
+      "step": 5230
+    },
+    {
+      "epoch": 1.495540492329647,
+      "grad_norm": 0.6500426530838013,
+      "learning_rate": 8.839742316529079e-05,
+      "loss": 0.4179,
+      "step": 5240
+    },
+    {
+      "epoch": 1.4983945772386729,
+      "grad_norm": 0.4330879747867584,
+      "learning_rate": 8.833351941982651e-05,
+      "loss": 0.4041,
+      "step": 5250
+    },
+    {
+      "epoch": 1.5012486621476988,
+      "grad_norm": 0.749954879283905,
+      "learning_rate": 8.826946340920183e-05,
+      "loss": 0.3884,
+      "step": 5260
+    },
+    {
+      "epoch": 1.504102747056725,
+      "grad_norm": 0.5958276987075806,
+      "learning_rate": 8.820525538785458e-05,
+      "loss": 0.4179,
+      "step": 5270
+    },
+    {
+      "epoch": 1.5069568319657511,
+      "grad_norm": 0.7096928358078003,
+      "learning_rate": 8.814089561082641e-05,
+      "loss": 0.4195,
+      "step": 5280
+    },
+    {
+      "epoch": 1.5098109168747769,
+      "grad_norm": 0.4753521680831909,
+      "learning_rate": 8.807638433376183e-05,
+      "loss": 0.4302,
+      "step": 5290
+    },
+    {
+      "epoch": 1.512665001783803,
+      "grad_norm": 1.1176053285598755,
+      "learning_rate": 8.801172181290702e-05,
+      "loss": 0.4136,
+      "step": 5300
+    },
+    {
+      "epoch": 1.5155190866928292,
+      "grad_norm": 0.5208104848861694,
+      "learning_rate": 8.794690830510901e-05,
+      "loss": 0.4324,
+      "step": 5310
+    },
+    {
+      "epoch": 1.5183731716018551,
+      "grad_norm": 1.0194308757781982,
+      "learning_rate": 8.788194406781452e-05,
+      "loss": 0.4324,
+      "step": 5320
+    },
+    {
+      "epoch": 1.521227256510881,
+      "grad_norm": 0.5444753766059875,
+      "learning_rate": 8.781682935906896e-05,
+      "loss": 0.4083,
+      "step": 5330
+    },
+    {
+      "epoch": 1.5240813414199073,
+      "grad_norm": 0.5094253420829773,
+      "learning_rate": 8.775156443751549e-05,
+      "loss": 0.4186,
+      "step": 5340
+    },
+    {
+      "epoch": 1.5269354263289334,
+      "grad_norm": 0.45871055126190186,
+      "learning_rate": 8.768614956239392e-05,
+      "loss": 0.3758,
+      "step": 5350
+    },
+    {
+      "epoch": 1.5297895112379594,
+      "grad_norm": 0.5674312710762024,
+      "learning_rate": 8.762058499353967e-05,
+      "loss": 0.447,
+      "step": 5360
+    },
+    {
+      "epoch": 1.5326435961469853,
+      "grad_norm": 0.5591029524803162,
+      "learning_rate": 8.755487099138276e-05,
+      "loss": 0.4304,
+      "step": 5370
+    },
+    {
+      "epoch": 1.5354976810560115,
+      "grad_norm": 0.8171464204788208,
+      "learning_rate": 8.748900781694677e-05,
+      "loss": 0.3959,
+      "step": 5380
+    },
+    {
+      "epoch": 1.5383517659650374,
+      "grad_norm": 0.8195403814315796,
+      "learning_rate": 8.742299573184786e-05,
+      "loss": 0.3785,
+      "step": 5390
+    },
+    {
+      "epoch": 1.5412058508740634,
+      "grad_norm": 1.0446261167526245,
+      "learning_rate": 8.735683499829362e-05,
+      "loss": 0.4787,
+      "step": 5400
+    },
+    {
+      "epoch": 1.5440599357830895,
+      "grad_norm": 0.8570737242698669,
+      "learning_rate": 8.729052587908213e-05,
+      "loss": 0.4181,
+      "step": 5410
+    },
+    {
+      "epoch": 1.5469140206921157,
+      "grad_norm": 0.6217248439788818,
+      "learning_rate": 8.722406863760087e-05,
+      "loss": 0.4379,
+      "step": 5420
+    },
+    {
+      "epoch": 1.5497681056011416,
+      "grad_norm": 0.4909728467464447,
+      "learning_rate": 8.715746353782563e-05,
+      "loss": 0.4083,
+      "step": 5430
+    },
+    {
+      "epoch": 1.5526221905101676,
+      "grad_norm": 0.6557335257530212,
+      "learning_rate": 8.709071084431957e-05,
+      "loss": 0.4228,
+      "step": 5440
+    },
+    {
+      "epoch": 1.5554762754191938,
+      "grad_norm": 0.7079155445098877,
+      "learning_rate": 8.702381082223209e-05,
+      "loss": 0.4686,
+      "step": 5450
+    },
+    {
+      "epoch": 1.55833036032822,
+      "grad_norm": 0.6752623319625854,
+      "learning_rate": 8.69567637372978e-05,
+      "loss": 0.423,
+      "step": 5460
+    },
+    {
+      "epoch": 1.5611844452372456,
+      "grad_norm": 0.49642160534858704,
+      "learning_rate": 8.688956985583545e-05,
+      "loss": 0.3938,
+      "step": 5470
+    },
+    {
+      "epoch": 1.5640385301462718,
+      "grad_norm": 0.6437910199165344,
+      "learning_rate": 8.682222944474687e-05,
+      "loss": 0.3994,
+      "step": 5480
+    },
+    {
+      "epoch": 1.566892615055298,
+      "grad_norm": 0.6423130631446838,
+      "learning_rate": 8.675474277151597e-05,
+      "loss": 0.4101,
+      "step": 5490
+    },
+    {
+      "epoch": 1.569746699964324,
+      "grad_norm": 0.5386946797370911,
+      "learning_rate": 8.668711010420758e-05,
+      "loss": 0.3989,
+      "step": 5500
+    },
+    {
+      "epoch": 1.5726007848733499,
+      "grad_norm": 0.75031578540802,
+      "learning_rate": 8.661933171146645e-05,
+      "loss": 0.4626,
+      "step": 5510
+    },
+    {
+      "epoch": 1.575454869782376,
+      "grad_norm": 0.6600304245948792,
+      "learning_rate": 8.655140786251617e-05,
+      "loss": 0.4635,
+      "step": 5520
+    },
+    {
+      "epoch": 1.5783089546914022,
+      "grad_norm": 0.48744162917137146,
+      "learning_rate": 8.648333882715814e-05,
+      "loss": 0.4433,
+      "step": 5530
+    },
+    {
+      "epoch": 1.5811630396004281,
+      "grad_norm": 0.5958801507949829,
+      "learning_rate": 8.641512487577037e-05,
+      "loss": 0.42,
+      "step": 5540
+    },
+    {
+      "epoch": 1.584017124509454,
+      "grad_norm": 0.7981345653533936,
+      "learning_rate": 8.634676627930659e-05,
+      "loss": 0.4308,
+      "step": 5550
+    },
+    {
+      "epoch": 1.5868712094184803,
+      "grad_norm": 0.5025626420974731,
+      "learning_rate": 8.627826330929498e-05,
+      "loss": 0.4404,
+      "step": 5560
+    },
+    {
+      "epoch": 1.5897252943275062,
+      "grad_norm": 0.6489295363426208,
+      "learning_rate": 8.620961623783728e-05,
+      "loss": 0.4002,
+      "step": 5570
+    },
+    {
+      "epoch": 1.5925793792365321,
+      "grad_norm": 0.8079462647438049,
+      "learning_rate": 8.614082533760754e-05,
+      "loss": 0.4113,
+      "step": 5580
+    },
+    {
+      "epoch": 1.5954334641455583,
+      "grad_norm": 0.7329133749008179,
+      "learning_rate": 8.60718908818512e-05,
+      "loss": 0.399,
+      "step": 5590
+    },
+    {
+      "epoch": 1.5982875490545845,
+      "grad_norm": 0.43324145674705505,
+      "learning_rate": 8.600281314438381e-05,
+      "loss": 0.4258,
+      "step": 5600
+    },
+    {
+      "epoch": 1.6011416339636104,
+      "grad_norm": 0.5767898559570312,
+      "learning_rate": 8.593359239959014e-05,
+      "loss": 0.4256,
+      "step": 5610
+    },
+    {
+      "epoch": 1.6039957188726364,
+      "grad_norm": 0.5638619661331177,
+      "learning_rate": 8.586422892242297e-05,
+      "loss": 0.393,
+      "step": 5620
+    },
+    {
+      "epoch": 1.6068498037816625,
+      "grad_norm": 0.8214574456214905,
+      "learning_rate": 8.579472298840204e-05,
+      "loss": 0.4311,
+      "step": 5630
+    },
+    {
+      "epoch": 1.6097038886906887,
+      "grad_norm": 0.538772702217102,
+      "learning_rate": 8.572507487361292e-05,
+      "loss": 0.4372,
+      "step": 5640
+    },
+    {
+      "epoch": 1.6125579735997146,
+      "grad_norm": 0.7743338346481323,
+      "learning_rate": 8.565528485470598e-05,
+      "loss": 0.4366,
+      "step": 5650
+    },
+    {
+      "epoch": 1.6154120585087406,
+      "grad_norm": 0.4678172171115875,
+      "learning_rate": 8.55853532088952e-05,
+      "loss": 0.423,
+      "step": 5660
+    },
+    {
+      "epoch": 1.6182661434177668,
+      "grad_norm": 0.7197970747947693,
+      "learning_rate": 8.551528021395714e-05,
+      "loss": 0.4073,
+      "step": 5670
+    },
+    {
+      "epoch": 1.6211202283267927,
+      "grad_norm": 0.6703267693519592,
+      "learning_rate": 8.544506614822983e-05,
+      "loss": 0.4026,
+      "step": 5680
+    },
+    {
+      "epoch": 1.6239743132358186,
+      "grad_norm": 0.4925888478755951,
+      "learning_rate": 8.537471129061163e-05,
+      "loss": 0.3971,
+      "step": 5690
+    },
+    {
+      "epoch": 1.6268283981448448,
+      "grad_norm": 0.5355518460273743,
+      "learning_rate": 8.530421592056017e-05,
+      "loss": 0.41,
+      "step": 5700
+    },
+    {
+      "epoch": 1.629682483053871,
+      "grad_norm": 0.8152324557304382,
+      "learning_rate": 8.523358031809116e-05,
+      "loss": 0.428,
+      "step": 5710
+    },
+    {
+      "epoch": 1.632536567962897,
+      "grad_norm": 0.536670446395874,
+      "learning_rate": 8.51628047637774e-05,
+      "loss": 0.4139,
+      "step": 5720
+    },
+    {
+      "epoch": 1.6353906528719229,
+      "grad_norm": 0.5895646810531616,
+      "learning_rate": 8.509188953874749e-05,
+      "loss": 0.4086,
+      "step": 5730
+    },
+    {
+      "epoch": 1.638244737780949,
+      "grad_norm": 0.551447331905365,
+      "learning_rate": 8.502083492468495e-05,
+      "loss": 0.3999,
+      "step": 5740
+    },
+    {
+      "epoch": 1.641098822689975,
+      "grad_norm": 0.5315040946006775,
+      "learning_rate": 8.494964120382688e-05,
+      "loss": 0.4092,
+      "step": 5750
+    },
+    {
+      "epoch": 1.643952907599001,
+      "grad_norm": 0.604598879814148,
+      "learning_rate": 8.487830865896294e-05,
+      "loss": 0.4035,
+      "step": 5760
+    },
+    {
+      "epoch": 1.646806992508027,
+      "grad_norm": 0.6488718390464783,
+      "learning_rate": 8.480683757343422e-05,
+      "loss": 0.4131,
+      "step": 5770
+    },
+    {
+      "epoch": 1.6496610774170533,
+      "grad_norm": 0.8826755881309509,
+      "learning_rate": 8.473522823113213e-05,
+      "loss": 0.4191,
+      "step": 5780
+    },
+    {
+      "epoch": 1.6525151623260792,
+      "grad_norm": 0.586421012878418,
+      "learning_rate": 8.466348091649726e-05,
+      "loss": 0.4388,
+      "step": 5790
+    },
+    {
+      "epoch": 1.6553692472351051,
+      "grad_norm": 0.6341783404350281,
+      "learning_rate": 8.459159591451815e-05,
+      "loss": 0.4084,
+      "step": 5800
+    },
+    {
+      "epoch": 1.6582233321441313,
+      "grad_norm": 0.6457330584526062,
+      "learning_rate": 8.451957351073038e-05,
+      "loss": 0.4073,
+      "step": 5810
+    },
+    {
+      "epoch": 1.6610774170531575,
+      "grad_norm": 0.5750988125801086,
+      "learning_rate": 8.444741399121522e-05,
+      "loss": 0.4185,
+      "step": 5820
+    },
+    {
+      "epoch": 1.6639315019621834,
+      "grad_norm": 0.46606963872909546,
+      "learning_rate": 8.437511764259862e-05,
+      "loss": 0.3942,
+      "step": 5830
+    },
+    {
+      "epoch": 1.6667855868712094,
+      "grad_norm": 0.6799011826515198,
+      "learning_rate": 8.430268475205e-05,
+      "loss": 0.4334,
+      "step": 5840
+    },
+    {
+      "epoch": 1.6696396717802355,
+      "grad_norm": 0.45308125019073486,
+      "learning_rate": 8.423011560728118e-05,
+      "loss": 0.3851,
+      "step": 5850
+    },
+    {
+      "epoch": 1.6724937566892615,
+      "grad_norm": 0.49569767713546753,
+      "learning_rate": 8.415741049654512e-05,
+      "loss": 0.417,
+      "step": 5860
+    },
+    {
+      "epoch": 1.6753478415982874,
+      "grad_norm": 0.68571537733078,
+      "learning_rate": 8.408456970863495e-05,
+      "loss": 0.3584,
+      "step": 5870
+    },
+    {
+      "epoch": 1.6782019265073136,
+      "grad_norm": 0.64979088306427,
+      "learning_rate": 8.401159353288268e-05,
+      "loss": 0.4009,
+      "step": 5880
+    },
+    {
+      "epoch": 1.6810560114163398,
+      "grad_norm": 0.5367964506149292,
+      "learning_rate": 8.393848225915809e-05,
+      "loss": 0.4307,
+      "step": 5890
+    },
+    {
+      "epoch": 1.6839100963253657,
+      "grad_norm": 0.5272241830825806,
+      "learning_rate": 8.386523617786757e-05,
+      "loss": 0.4088,
+      "step": 5900
+    },
+    {
+      "epoch": 1.6867641812343916,
+      "grad_norm": 0.6130824685096741,
+      "learning_rate": 8.379185557995302e-05,
+      "loss": 0.3827,
+      "step": 5910
+    },
+    {
+      "epoch": 1.6896182661434178,
+      "grad_norm": 0.4380579888820648,
+      "learning_rate": 8.371834075689067e-05,
+      "loss": 0.3958,
+      "step": 5920
+    },
+    {
+      "epoch": 1.6924723510524438,
+      "grad_norm": 0.5438313484191895,
+      "learning_rate": 8.364469200068981e-05,
+      "loss": 0.4141,
+      "step": 5930
+    },
+    {
+      "epoch": 1.6953264359614697,
+      "grad_norm": 0.5417528748512268,
+      "learning_rate": 8.357090960389186e-05,
+      "loss": 0.41,
+      "step": 5940
+    },
+    {
+      "epoch": 1.6981805208704959,
+      "grad_norm": 0.4851051867008209,
+      "learning_rate": 8.349699385956894e-05,
+      "loss": 0.4063,
+      "step": 5950
+    },
+    {
+      "epoch": 1.701034605779522,
+      "grad_norm": 0.7028306126594543,
+      "learning_rate": 8.3422945061323e-05,
+      "loss": 0.3612,
+      "step": 5960
+    },
+    {
+      "epoch": 1.703888690688548,
+      "grad_norm": 0.6762276291847229,
+      "learning_rate": 8.334876350328433e-05,
+      "loss": 0.4122,
+      "step": 5970
+    },
+    {
+      "epoch": 1.706742775597574,
+      "grad_norm": 0.8307727575302124,
+      "learning_rate": 8.327444948011067e-05,
+      "loss": 0.4257,
+      "step": 5980
+    },
+    {
+      "epoch": 1.7095968605066,
+      "grad_norm": 0.6955322623252869,
+      "learning_rate": 8.32000032869859e-05,
+      "loss": 0.4208,
+      "step": 5990
+    },
+    {
+      "epoch": 1.7124509454156263,
+      "grad_norm": 0.5121544003486633,
+      "learning_rate": 8.312542521961884e-05,
+      "loss": 0.4031,
+      "step": 6000
+    },
+    {
+      "epoch": 1.7153050303246522,
+      "grad_norm": 0.5445939898490906,
+      "learning_rate": 8.305071557424222e-05,
+      "loss": 0.3946,
+      "step": 6010
+    },
+    {
+      "epoch": 1.7181591152336781,
+      "grad_norm": 0.5593860149383545,
+      "learning_rate": 8.297587464761136e-05,
+      "loss": 0.3964,
+      "step": 6020
+    },
+    {
+      "epoch": 1.7210132001427043,
+      "grad_norm": 0.6780118942260742,
+      "learning_rate": 8.290090273700304e-05,
+      "loss": 0.4243,
+      "step": 6030
+    },
+    {
+      "epoch": 1.7238672850517303,
+      "grad_norm": 0.8964153528213501,
+      "learning_rate": 8.282580014021435e-05,
+      "loss": 0.4214,
+      "step": 6040
+    },
+    {
+      "epoch": 1.7267213699607562,
+      "grad_norm": 0.658841609954834,
+      "learning_rate": 8.275056715556144e-05,
+      "loss": 0.3939,
+      "step": 6050
+    },
+    {
+      "epoch": 1.7295754548697824,
+      "grad_norm": 0.8680906891822815,
+      "learning_rate": 8.267520408187843e-05,
+      "loss": 0.4127,
+      "step": 6060
+    },
+    {
+      "epoch": 1.7324295397788085,
+      "grad_norm": 0.6496133804321289,
+      "learning_rate": 8.259971121851616e-05,
+      "loss": 0.4288,
+      "step": 6070
+    },
+    {
+      "epoch": 1.7352836246878345,
+      "grad_norm": 0.4331784248352051,
+      "learning_rate": 8.252408886534092e-05,
+      "loss": 0.4265,
+      "step": 6080
+    },
+    {
+      "epoch": 1.7381377095968604,
+      "grad_norm": 0.568143904209137,
+      "learning_rate": 8.244833732273349e-05,
+      "loss": 0.4333,
+      "step": 6090
+    },
+    {
+      "epoch": 1.7409917945058866,
+      "grad_norm": 0.5459315776824951,
+      "learning_rate": 8.237245689158768e-05,
+      "loss": 0.4249,
+      "step": 6100
+    },
+    {
+      "epoch": 1.7438458794149128,
+      "grad_norm": 0.9281138777732849,
+      "learning_rate": 8.229644787330936e-05,
+      "loss": 0.4171,
+      "step": 6110
+    },
+    {
+      "epoch": 1.7466999643239385,
+      "grad_norm": 0.7226198315620422,
+      "learning_rate": 8.22203105698151e-05,
+      "loss": 0.3746,
+      "step": 6120
+    },
+    {
+      "epoch": 1.7495540492329646,
+      "grad_norm": 0.4380105435848236,
+      "learning_rate": 8.214404528353104e-05,
+      "loss": 0.3939,
+      "step": 6130
+    },
+    {
+      "epoch": 1.7524081341419908,
+      "grad_norm": 0.5727590322494507,
+      "learning_rate": 8.206765231739168e-05,
+      "loss": 0.4035,
+      "step": 6140
+    },
+    {
+      "epoch": 1.7552622190510168,
+      "grad_norm": 0.4349416494369507,
+      "learning_rate": 8.199113197483874e-05,
+      "loss": 0.3843,
+      "step": 6150
+    },
+    {
+      "epoch": 1.7581163039600427,
+      "grad_norm": 0.5138025879859924,
+      "learning_rate": 8.191448455981982e-05,
+      "loss": 0.3897,
+      "step": 6160
+    },
+    {
+      "epoch": 1.7609703888690689,
+      "grad_norm": 0.6387166380882263,
+      "learning_rate": 8.18377103767873e-05,
+      "loss": 0.384,
+      "step": 6170
+    },
+    {
+      "epoch": 1.763824473778095,
+      "grad_norm": 0.5494593977928162,
+      "learning_rate": 8.176080973069706e-05,
+      "loss": 0.3952,
+      "step": 6180
+    },
+    {
+      "epoch": 1.766678558687121,
+      "grad_norm": 0.6428438425064087,
+      "learning_rate": 8.168378292700738e-05,
+      "loss": 0.3994,
+      "step": 6190
+    },
+    {
+      "epoch": 1.769532643596147,
+      "grad_norm": 1.2720189094543457,
+      "learning_rate": 8.160663027167756e-05,
+      "loss": 0.416,
+      "step": 6200
+    },
+    {
+      "epoch": 1.772386728505173,
+      "grad_norm": 0.539422869682312,
+      "learning_rate": 8.152935207116688e-05,
+      "loss": 0.3789,
+      "step": 6210
+    },
+    {
+      "epoch": 1.775240813414199,
+      "grad_norm": 0.5536684393882751,
+      "learning_rate": 8.145194863243328e-05,
+      "loss": 0.4329,
+      "step": 6220
+    },
+    {
+      "epoch": 1.778094898323225,
+      "grad_norm": 0.7319250702857971,
+      "learning_rate": 8.137442026293212e-05,
+      "loss": 0.4298,
+      "step": 6230
+    },
+    {
+      "epoch": 1.7809489832322511,
+      "grad_norm": 0.6316890120506287,
+      "learning_rate": 8.129676727061502e-05,
+      "loss": 0.4081,
+      "step": 6240
+    },
+    {
+      "epoch": 1.7838030681412773,
+      "grad_norm": 1.2765876054763794,
+      "learning_rate": 8.121898996392863e-05,
+      "loss": 0.4158,
+      "step": 6250
+    },
+    {
+      "epoch": 1.7866571530503033,
+      "grad_norm": 0.48435863852500916,
+      "learning_rate": 8.114108865181338e-05,
+      "loss": 0.412,
+      "step": 6260
+    },
+    {
+      "epoch": 1.7895112379593292,
+      "grad_norm": 0.7326620817184448,
+      "learning_rate": 8.106306364370228e-05,
+      "loss": 0.424,
+      "step": 6270
+    },
+    {
+      "epoch": 1.7923653228683554,
+      "grad_norm": 0.7231389284133911,
+      "learning_rate": 8.098491524951966e-05,
+      "loss": 0.4133,
+      "step": 6280
+    },
+    {
+      "epoch": 1.7952194077773815,
+      "grad_norm": 0.6374312043190002,
+      "learning_rate": 8.090664377967993e-05,
+      "loss": 0.461,
+      "step": 6290
+    },
+    {
+      "epoch": 1.7980734926864073,
+      "grad_norm": 0.5649964213371277,
+      "learning_rate": 8.082824954508642e-05,
+      "loss": 0.3768,
+      "step": 6300
+    },
+    {
+      "epoch": 1.8009275775954334,
+      "grad_norm": 0.5286874175071716,
+      "learning_rate": 8.074973285713008e-05,
+      "loss": 0.4139,
+      "step": 6310
+    },
+    {
+      "epoch": 1.8037816625044596,
+      "grad_norm": 0.5340945720672607,
+      "learning_rate": 8.067109402768822e-05,
+      "loss": 0.4288,
+      "step": 6320
+    },
+    {
+      "epoch": 1.8066357474134855,
+      "grad_norm": 0.5859851837158203,
+      "learning_rate": 8.059233336912335e-05,
+      "loss": 0.3793,
+      "step": 6330
+    },
+    {
+      "epoch": 1.8094898323225115,
+      "grad_norm": 0.6643693447113037,
+      "learning_rate": 8.051345119428191e-05,
+      "loss": 0.3761,
+      "step": 6340
+    },
+    {
+      "epoch": 1.8123439172315376,
+      "grad_norm": 0.9013630151748657,
+      "learning_rate": 8.043444781649298e-05,
+      "loss": 0.4003,
+      "step": 6350
+    },
+    {
+      "epoch": 1.8151980021405638,
+      "grad_norm": 0.514897882938385,
+      "learning_rate": 8.035532354956707e-05,
+      "loss": 0.3897,
+      "step": 6360
+    },
+    {
+      "epoch": 1.8180520870495898,
+      "grad_norm": 0.8518983125686646,
+      "learning_rate": 8.027607870779495e-05,
+      "loss": 0.3865,
+      "step": 6370
+    },
+    {
+      "epoch": 1.8209061719586157,
+      "grad_norm": 0.5759355425834656,
+      "learning_rate": 8.019671360594621e-05,
+      "loss": 0.4141,
+      "step": 6380
+    },
+    {
+      "epoch": 1.8237602568676419,
+      "grad_norm": 0.5349940061569214,
+      "learning_rate": 8.011722855926822e-05,
+      "loss": 0.4281,
+      "step": 6390
+    },
+    {
+      "epoch": 1.8266143417766678,
+      "grad_norm": 0.5943355560302734,
+      "learning_rate": 8.003762388348477e-05,
+      "loss": 0.3837,
+      "step": 6400
+    },
+    {
+      "epoch": 1.8294684266856938,
+      "grad_norm": 0.531690776348114,
+      "learning_rate": 7.995789989479477e-05,
+      "loss": 0.3673,
+      "step": 6410
+    },
+    {
+      "epoch": 1.83232251159472,
+      "grad_norm": 0.6260749101638794,
+      "learning_rate": 7.987805690987114e-05,
+      "loss": 0.4121,
+      "step": 6420
+    },
+    {
+      "epoch": 1.835176596503746,
+      "grad_norm": 0.5923037528991699,
+      "learning_rate": 7.979809524585937e-05,
+      "loss": 0.4019,
+      "step": 6430
+    },
+    {
+      "epoch": 1.838030681412772,
+      "grad_norm": 0.693532407283783,
+      "learning_rate": 7.971801522037647e-05,
+      "loss": 0.4394,
+      "step": 6440
+    },
+    {
+      "epoch": 1.840884766321798,
+      "grad_norm": 0.9910332560539246,
+      "learning_rate": 7.963781715150951e-05,
+      "loss": 0.4668,
+      "step": 6450
+    },
+    {
+      "epoch": 1.8437388512308241,
+      "grad_norm": 0.5126996636390686,
+      "learning_rate": 7.955750135781447e-05,
+      "loss": 0.4084,
+      "step": 6460
+    },
+    {
+      "epoch": 1.8465929361398503,
+      "grad_norm": 0.6543841361999512,
+      "learning_rate": 7.947706815831497e-05,
+      "loss": 0.3998,
+      "step": 6470
+    },
+    {
+      "epoch": 1.849447021048876,
+      "grad_norm": 0.5038157105445862,
+      "learning_rate": 7.939651787250091e-05,
+      "loss": 0.4045,
+      "step": 6480
+    },
+    {
+      "epoch": 1.8523011059579022,
+      "grad_norm": 0.5962367057800293,
+      "learning_rate": 7.931585082032737e-05,
+      "loss": 0.4243,
+      "step": 6490
+    },
+    {
+      "epoch": 1.8551551908669284,
+      "grad_norm": 0.5447099804878235,
+      "learning_rate": 7.923506732221314e-05,
+      "loss": 0.4105,
+      "step": 6500
+    },
+    {
+      "epoch": 1.8580092757759543,
+      "grad_norm": 0.6572823524475098,
+      "learning_rate": 7.91541676990396e-05,
+      "loss": 0.4253,
+      "step": 6510
+    },
+    {
+      "epoch": 1.8608633606849803,
+      "grad_norm": 0.48391202092170715,
+      "learning_rate": 7.907315227214939e-05,
+      "loss": 0.4065,
+      "step": 6520
+    },
+    {
+      "epoch": 1.8637174455940064,
+      "grad_norm": 0.599360466003418,
+      "learning_rate": 7.899202136334514e-05,
+      "loss": 0.4022,
+      "step": 6530
+    },
+    {
+      "epoch": 1.8665715305030326,
+      "grad_norm": 0.6030710935592651,
+      "learning_rate": 7.891077529488817e-05,
+      "loss": 0.3849,
+      "step": 6540
+    },
+    {
+      "epoch": 1.8694256154120585,
+      "grad_norm": 0.6563034653663635,
+      "learning_rate": 7.882941438949723e-05,
+      "loss": 0.4047,
+      "step": 6550
+    },
+    {
+      "epoch": 1.8722797003210845,
+      "grad_norm": 0.7676080465316772,
+      "learning_rate": 7.874793897034722e-05,
+      "loss": 0.416,
+      "step": 6560
+    },
+    {
+      "epoch": 1.8751337852301107,
+      "grad_norm": 0.9442748427391052,
+      "learning_rate": 7.866634936106792e-05,
+      "loss": 0.3855,
+      "step": 6570
+    },
+    {
+      "epoch": 1.8779878701391366,
+      "grad_norm": 0.5683465600013733,
+      "learning_rate": 7.858464588574264e-05,
+      "loss": 0.3895,
+      "step": 6580
+    },
+    {
+      "epoch": 1.8808419550481625,
+      "grad_norm": 0.6074008941650391,
+      "learning_rate": 7.850282886890702e-05,
+      "loss": 0.3592,
+      "step": 6590
+    },
+    {
+      "epoch": 1.8836960399571887,
+      "grad_norm": 0.44344863295555115,
+      "learning_rate": 7.84208986355477e-05,
+      "loss": 0.3667,
+      "step": 6600
+    },
+    {
+      "epoch": 1.8865501248662149,
+      "grad_norm": 0.5034730434417725,
+      "learning_rate": 7.8338855511101e-05,
+      "loss": 0.3899,
+      "step": 6610
+    },
+    {
+      "epoch": 1.8894042097752408,
+      "grad_norm": 0.5974074006080627,
+      "learning_rate": 7.825669982145169e-05,
+      "loss": 0.4063,
+      "step": 6620
+    },
+    {
+      "epoch": 1.8922582946842668,
+      "grad_norm": 0.6027740836143494,
+      "learning_rate": 7.817443189293162e-05,
+      "loss": 0.378,
+      "step": 6630
+    },
+    {
+      "epoch": 1.895112379593293,
+      "grad_norm": 0.6006380319595337,
+      "learning_rate": 7.809205205231851e-05,
+      "loss": 0.4143,
+      "step": 6640
+    },
+    {
+      "epoch": 1.897966464502319,
+      "grad_norm": 0.4824349284172058,
+      "learning_rate": 7.800956062683458e-05,
+      "loss": 0.4059,
+      "step": 6650
+    },
+    {
+      "epoch": 1.900820549411345,
+      "grad_norm": 0.4044560194015503,
+      "learning_rate": 7.792695794414528e-05,
+      "loss": 0.3845,
+      "step": 6660
+    },
+    {
+      "epoch": 1.903674634320371,
+      "grad_norm": 0.5848098397254944,
+      "learning_rate": 7.784424433235803e-05,
+      "loss": 0.4118,
+      "step": 6670
+    },
+    {
+      "epoch": 1.9065287192293972,
+      "grad_norm": 0.7342092990875244,
+      "learning_rate": 7.776142012002077e-05,
+      "loss": 0.4457,
+      "step": 6680
+    },
+    {
+      "epoch": 1.909382804138423,
+      "grad_norm": 0.6541807651519775,
+      "learning_rate": 7.767848563612087e-05,
+      "loss": 0.4013,
+      "step": 6690
+    },
+    {
+      "epoch": 1.912236889047449,
+      "grad_norm": 0.6325574517250061,
+      "learning_rate": 7.759544121008368e-05,
+      "loss": 0.3985,
+      "step": 6700
+    },
+    {
+      "epoch": 1.9150909739564752,
+      "grad_norm": 0.6047912836074829,
+      "learning_rate": 7.75122871717712e-05,
+      "loss": 0.3748,
+      "step": 6710
+    },
+    {
+      "epoch": 1.9179450588655014,
+      "grad_norm": 0.8499783277511597,
+      "learning_rate": 7.742902385148087e-05,
+      "loss": 0.4232,
+      "step": 6720
+    },
+    {
+      "epoch": 1.9207991437745273,
+      "grad_norm": 0.6166467070579529,
+      "learning_rate": 7.734565157994423e-05,
+      "loss": 0.3894,
+      "step": 6730
+    },
+    {
+      "epoch": 1.9236532286835533,
+      "grad_norm": 0.6198390126228333,
+      "learning_rate": 7.726217068832551e-05,
+      "loss": 0.4304,
+      "step": 6740
+    },
+    {
+      "epoch": 1.9265073135925794,
+      "grad_norm": 0.5126845240592957,
+      "learning_rate": 7.717858150822048e-05,
+      "loss": 0.3671,
+      "step": 6750
+    },
+    {
+      "epoch": 1.9293613985016054,
+      "grad_norm": 0.6374401450157166,
+      "learning_rate": 7.7094884371655e-05,
+      "loss": 0.4355,
+      "step": 6760
+    },
+    {
+      "epoch": 1.9322154834106313,
+      "grad_norm": 0.5290126204490662,
+      "learning_rate": 7.701107961108374e-05,
+      "loss": 0.3938,
+      "step": 6770
+    },
+    {
+      "epoch": 1.9350695683196575,
+      "grad_norm": 0.5920684933662415,
+      "learning_rate": 7.692716755938884e-05,
+      "loss": 0.4078,
+      "step": 6780
+    },
+    {
+      "epoch": 1.9379236532286837,
+      "grad_norm": 0.8244432210922241,
+      "learning_rate": 7.684314854987873e-05,
+      "loss": 0.4081,
+      "step": 6790
+    },
+    {
+      "epoch": 1.9407777381377096,
+      "grad_norm": 0.7835015654563904,
+      "learning_rate": 7.675902291628653e-05,
+      "loss": 0.4596,
+      "step": 6800
+    },
+    {
+      "epoch": 1.9436318230467355,
+      "grad_norm": 0.7511467337608337,
+      "learning_rate": 7.6674790992769e-05,
+      "loss": 0.4151,
+      "step": 6810
+    },
+    {
+      "epoch": 1.9464859079557617,
+      "grad_norm": 0.7354687452316284,
+      "learning_rate": 7.659045311390505e-05,
+      "loss": 0.3712,
+      "step": 6820
+    },
+    {
+      "epoch": 1.9493399928647879,
+      "grad_norm": 0.7676160931587219,
+      "learning_rate": 7.650600961469445e-05,
+      "loss": 0.4129,
+      "step": 6830
+    },
+    {
+      "epoch": 1.9521940777738138,
+      "grad_norm": 0.5947080850601196,
+      "learning_rate": 7.642146083055653e-05,
+      "loss": 0.4174,
+      "step": 6840
+    },
+    {
+      "epoch": 1.9550481626828398,
+      "grad_norm": 0.7656364440917969,
+      "learning_rate": 7.633680709732878e-05,
+      "loss": 0.4157,
+      "step": 6850
+    },
+    {
+      "epoch": 1.957902247591866,
+      "grad_norm": 0.6544439792633057,
+      "learning_rate": 7.625204875126564e-05,
+      "loss": 0.3855,
+      "step": 6860
+    },
+    {
+      "epoch": 1.9607563325008919,
+      "grad_norm": 0.8954373002052307,
+      "learning_rate": 7.616718612903702e-05,
+      "loss": 0.4537,
+      "step": 6870
+    },
+    {
+      "epoch": 1.9636104174099178,
+      "grad_norm": 0.9667245745658875,
+      "learning_rate": 7.608221956772704e-05,
+      "loss": 0.3826,
+      "step": 6880
+    },
+    {
+      "epoch": 1.966464502318944,
+      "grad_norm": 0.584932804107666,
+      "learning_rate": 7.599714940483267e-05,
+      "loss": 0.446,
+      "step": 6890
+    },
+    {
+      "epoch": 1.9693185872279702,
+      "grad_norm": 0.568055272102356,
+      "learning_rate": 7.591197597826243e-05,
+      "loss": 0.4048,
+      "step": 6900
+    },
+    {
+      "epoch": 1.972172672136996,
+      "grad_norm": 0.5002406239509583,
+      "learning_rate": 7.582669962633498e-05,
+      "loss": 0.4238,
+      "step": 6910
+    },
+    {
+      "epoch": 1.975026757046022,
+      "grad_norm": 1.0992151498794556,
+      "learning_rate": 7.574132068777781e-05,
+      "loss": 0.4496,
+      "step": 6920
+    },
+    {
+      "epoch": 1.9778808419550482,
+      "grad_norm": 0.5405780076980591,
+      "learning_rate": 7.565583950172595e-05,
+      "loss": 0.3914,
+      "step": 6930
+    },
+    {
+      "epoch": 1.9807349268640742,
+      "grad_norm": 0.6248615980148315,
+      "learning_rate": 7.55702564077205e-05,
+      "loss": 0.4425,
+      "step": 6940
+    },
+    {
+      "epoch": 1.9835890117731,
+      "grad_norm": 0.5931448340415955,
+      "learning_rate": 7.54845717457074e-05,
+      "loss": 0.3759,
+      "step": 6950
+    },
+    {
+      "epoch": 1.9864430966821263,
+      "grad_norm": 0.6648136377334595,
+      "learning_rate": 7.539878585603599e-05,
+      "loss": 0.4167,
+      "step": 6960
+    },
+    {
+      "epoch": 1.9892971815911524,
+      "grad_norm": 0.6499849557876587,
+      "learning_rate": 7.531289907945773e-05,
+      "loss": 0.4238,
+      "step": 6970
+    },
+    {
+      "epoch": 1.9921512665001784,
+      "grad_norm": 0.5630407929420471,
+      "learning_rate": 7.522691175712482e-05,
+      "loss": 0.405,
+      "step": 6980
+    },
+    {
+      "epoch": 1.9950053514092043,
+      "grad_norm": 0.6160908341407776,
+      "learning_rate": 7.514082423058879e-05,
+      "loss": 0.4276,
+      "step": 6990
+    },
+    {
+      "epoch": 1.9978594363182305,
+      "grad_norm": 0.6159669160842896,
+      "learning_rate": 7.505463684179923e-05,
+      "loss": 0.4513,
+      "step": 7000
+    },
+    {
+      "epoch": 2.0007135212272567,
+      "grad_norm": 0.5455153584480286,
+      "learning_rate": 7.49683499331024e-05,
+      "loss": 0.3841,
+      "step": 7010
+    },
+    {
+      "epoch": 2.0035676061362824,
+      "grad_norm": 0.5316396355628967,
+      "learning_rate": 7.488196384723986e-05,
+      "loss": 0.4091,
+      "step": 7020
+    },
+    {
+      "epoch": 2.0064216910453085,
+      "grad_norm": 0.5328727960586548,
+      "learning_rate": 7.479547892734707e-05,
+      "loss": 0.3826,
+      "step": 7030
+    },
+    {
+      "epoch": 2.0092757759543347,
+      "grad_norm": 0.6331178545951843,
+      "learning_rate": 7.470889551695215e-05,
+      "loss": 0.3912,
+      "step": 7040
+    },
+    {
+      "epoch": 2.012129860863361,
+      "grad_norm": 0.6090428829193115,
+      "learning_rate": 7.462221395997439e-05,
+      "loss": 0.3951,
+      "step": 7050
+    },
+    {
+      "epoch": 2.0149839457723866,
+      "grad_norm": 0.7349817156791687,
+      "learning_rate": 7.45354346007229e-05,
+      "loss": 0.3798,
+      "step": 7060
+    },
+    {
+      "epoch": 2.0178380306814128,
+      "grad_norm": 0.4755643904209137,
+      "learning_rate": 7.444855778389533e-05,
+      "loss": 0.3885,
+      "step": 7070
+    },
+    {
+      "epoch": 2.020692115590439,
+      "grad_norm": 0.6862337589263916,
+      "learning_rate": 7.43615838545764e-05,
+      "loss": 0.3959,
+      "step": 7080
+    },
+    {
+      "epoch": 2.0235462004994647,
+      "grad_norm": 0.9478683471679688,
+      "learning_rate": 7.427451315823661e-05,
+      "loss": 0.3998,
+      "step": 7090
+    },
+    {
+      "epoch": 2.026400285408491,
+      "grad_norm": 1.3362914323806763,
+      "learning_rate": 7.418734604073078e-05,
+      "loss": 0.3649,
+      "step": 7100
+    },
+    {
+      "epoch": 2.029254370317517,
+      "grad_norm": 0.9035010933876038,
+      "learning_rate": 7.410008284829677e-05,
+      "loss": 0.393,
+      "step": 7110
+    },
+    {
+      "epoch": 2.032108455226543,
+      "grad_norm": 0.6594682335853577,
+      "learning_rate": 7.401272392755405e-05,
+      "loss": 0.3629,
+      "step": 7120
+    },
+    {
+      "epoch": 2.034962540135569,
+      "grad_norm": 0.7762911915779114,
+      "learning_rate": 7.392526962550233e-05,
+      "loss": 0.3763,
+      "step": 7130
+    },
+    {
+      "epoch": 2.037816625044595,
+      "grad_norm": 0.9021196365356445,
+      "learning_rate": 7.383772028952016e-05,
+      "loss": 0.4282,
+      "step": 7140
+    },
+    {
+      "epoch": 2.040670709953621,
+      "grad_norm": 0.5962260365486145,
+      "learning_rate": 7.375007626736363e-05,
+      "loss": 0.4137,
+      "step": 7150
+    },
+    {
+      "epoch": 2.0435247948626474,
+      "grad_norm": 0.6486306190490723,
+      "learning_rate": 7.36623379071649e-05,
+      "loss": 0.4109,
+      "step": 7160
+    },
+    {
+      "epoch": 2.046378879771673,
+      "grad_norm": 0.8934798836708069,
+      "learning_rate": 7.357450555743082e-05,
+      "loss": 0.4047,
+      "step": 7170
+    },
+    {
+      "epoch": 2.0492329646806993,
+      "grad_norm": 1.0776855945587158,
+      "learning_rate": 7.348657956704166e-05,
+      "loss": 0.4144,
+      "step": 7180
+    },
+    {
+      "epoch": 2.0520870495897254,
+      "grad_norm": 1.1209388971328735,
+      "learning_rate": 7.339856028524958e-05,
+      "loss": 0.3987,
+      "step": 7190
+    },
+    {
+      "epoch": 2.054941134498751,
+      "grad_norm": 0.7486249208450317,
+      "learning_rate": 7.331044806167733e-05,
+      "loss": 0.3858,
+      "step": 7200
+    },
+    {
+      "epoch": 2.0577952194077773,
+      "grad_norm": 0.8954799771308899,
+      "learning_rate": 7.322224324631679e-05,
+      "loss": 0.396,
+      "step": 7210
+    },
+    {
+      "epoch": 2.0606493043168035,
+      "grad_norm": 0.9029479622840881,
+      "learning_rate": 7.31339461895277e-05,
+      "loss": 0.3833,
+      "step": 7220
+    },
+    {
+      "epoch": 2.0635033892258297,
+      "grad_norm": 0.9437860250473022,
+      "learning_rate": 7.30455572420361e-05,
+      "loss": 0.3884,
+      "step": 7230
+    },
+    {
+      "epoch": 2.0663574741348554,
+      "grad_norm": 0.7004602551460266,
+      "learning_rate": 7.295707675493315e-05,
+      "loss": 0.3728,
+      "step": 7240
+    },
+    {
+      "epoch": 2.0692115590438815,
+      "grad_norm": 0.812864363193512,
+      "learning_rate": 7.286850507967351e-05,
+      "loss": 0.3388,
+      "step": 7250
+    },
+    {
+      "epoch": 2.0720656439529077,
+      "grad_norm": 0.65233314037323,
+      "learning_rate": 7.277984256807411e-05,
+      "loss": 0.3755,
+      "step": 7260
+    },
+    {
+      "epoch": 2.0749197288619334,
+      "grad_norm": 1.1088277101516724,
+      "learning_rate": 7.269108957231263e-05,
+      "loss": 0.4505,
+      "step": 7270
+    },
+    {
+      "epoch": 2.0777738137709596,
+      "grad_norm": 0.8338673114776611,
+      "learning_rate": 7.260224644492627e-05,
+      "loss": 0.3966,
+      "step": 7280
+    },
+    {
+      "epoch": 2.0806278986799858,
+      "grad_norm": 0.7986013889312744,
+      "learning_rate": 7.251331353881012e-05,
+      "loss": 0.3803,
+      "step": 7290
+    },
+    {
+      "epoch": 2.083481983589012,
+      "grad_norm": 0.8974848985671997,
+      "learning_rate": 7.242429120721597e-05,
+      "loss": 0.3404,
+      "step": 7300
+    },
+    {
+      "epoch": 2.0863360684980377,
+      "grad_norm": 0.6625577807426453,
+      "learning_rate": 7.23351798037508e-05,
+      "loss": 0.3767,
+      "step": 7310
+    },
+    {
+      "epoch": 2.089190153407064,
+      "grad_norm": 0.6558323502540588,
+      "learning_rate": 7.224597968237535e-05,
+      "loss": 0.3747,
+      "step": 7320
+    },
+    {
+      "epoch": 2.09204423831609,
+      "grad_norm": 0.9012863039970398,
+      "learning_rate": 7.215669119740283e-05,
+      "loss": 0.3747,
+      "step": 7330
+    },
+    {
+      "epoch": 2.094898323225116,
+      "grad_norm": 1.1862736940383911,
+      "learning_rate": 7.206731470349735e-05,
+      "loss": 0.3755,
+      "step": 7340
+    },
+    {
+      "epoch": 2.097752408134142,
+      "grad_norm": 0.825542151927948,
+      "learning_rate": 7.197785055567269e-05,
+      "loss": 0.3909,
+      "step": 7350
+    },
+    {
+      "epoch": 2.100606493043168,
+      "grad_norm": 0.9518323540687561,
+      "learning_rate": 7.188829910929076e-05,
+      "loss": 0.3965,
+      "step": 7360
+    },
+    {
+      "epoch": 2.103460577952194,
+      "grad_norm": 0.6145418286323547,
+      "learning_rate": 7.179866072006023e-05,
+      "loss": 0.3869,
+      "step": 7370
+    },
+    {
+      "epoch": 2.10631466286122,
+      "grad_norm": 0.6919503211975098,
+      "learning_rate": 7.17089357440351e-05,
+      "loss": 0.353,
+      "step": 7380
+    },
+    {
+      "epoch": 2.109168747770246,
+      "grad_norm": 0.8245930671691895,
+      "learning_rate": 7.161912453761336e-05,
+      "loss": 0.4093,
+      "step": 7390
+    },
+    {
+      "epoch": 2.1120228326792723,
+      "grad_norm": 0.8390684723854065,
+      "learning_rate": 7.152922745753544e-05,
+      "loss": 0.3907,
+      "step": 7400
+    },
+    {
+      "epoch": 2.1148769175882984,
+      "grad_norm": 1.182658314704895,
+      "learning_rate": 7.143924486088294e-05,
+      "loss": 0.4122,
+      "step": 7410
+    },
+    {
+      "epoch": 2.117731002497324,
+      "grad_norm": 1.051106333732605,
+      "learning_rate": 7.134917710507713e-05,
+      "loss": 0.3924,
+      "step": 7420
+    },
+    {
+      "epoch": 2.1205850874063503,
+      "grad_norm": 0.7197733521461487,
+      "learning_rate": 7.125902454787747e-05,
+      "loss": 0.4191,
+      "step": 7430
+    },
+    {
+      "epoch": 2.1234391723153765,
+      "grad_norm": 0.7944600582122803,
+      "learning_rate": 7.116878754738036e-05,
+      "loss": 0.3655,
+      "step": 7440
+    },
+    {
+      "epoch": 2.126293257224402,
+      "grad_norm": 1.1898823976516724,
+      "learning_rate": 7.107846646201753e-05,
+      "loss": 0.3806,
+      "step": 7450
+    },
+    {
+      "epoch": 2.1291473421334284,
+      "grad_norm": 0.6632446646690369,
+      "learning_rate": 7.098806165055481e-05,
+      "loss": 0.4042,
+      "step": 7460
+    },
+    {
+      "epoch": 2.1320014270424545,
+      "grad_norm": 0.9864144325256348,
+      "learning_rate": 7.089757347209049e-05,
+      "loss": 0.3935,
+      "step": 7470
+    },
+    {
+      "epoch": 2.1348555119514807,
+      "grad_norm": 0.6241956949234009,
+      "learning_rate": 7.080700228605408e-05,
+      "loss": 0.3616,
+      "step": 7480
+    },
+    {
+      "epoch": 2.1377095968605064,
+      "grad_norm": 0.7167524099349976,
+      "learning_rate": 7.071634845220474e-05,
+      "loss": 0.3819,
+      "step": 7490
+    },
+    {
+      "epoch": 2.1405636817695326,
+      "grad_norm": 0.9943698644638062,
+      "learning_rate": 7.062561233062999e-05,
+      "loss": 0.387,
+      "step": 7500
+    },
+    {
+      "epoch": 2.1434177666785588,
+      "grad_norm": 0.8617923855781555,
+      "learning_rate": 7.053479428174415e-05,
+      "loss": 0.3646,
+      "step": 7510
+    },
+    {
+      "epoch": 2.146271851587585,
+      "grad_norm": 0.8294242024421692,
+      "learning_rate": 7.044389466628701e-05,
+      "loss": 0.3491,
+      "step": 7520
+    },
+    {
+      "epoch": 2.1491259364966107,
+      "grad_norm": 0.9514858722686768,
+      "learning_rate": 7.035291384532228e-05,
+      "loss": 0.4077,
+      "step": 7530
+    },
+    {
+      "epoch": 2.151980021405637,
+      "grad_norm": 0.7234044671058655,
+      "learning_rate": 7.026185218023633e-05,
+      "loss": 0.3728,
+      "step": 7540
+    },
+    {
+      "epoch": 2.154834106314663,
+      "grad_norm": 0.6389915943145752,
+      "learning_rate": 7.017071003273656e-05,
+      "loss": 0.3878,
+      "step": 7550
+    },
+    {
+      "epoch": 2.1576881912236887,
+      "grad_norm": 0.9411191940307617,
+      "learning_rate": 7.007948776485008e-05,
+      "loss": 0.3869,
+      "step": 7560
+    },
+    {
+      "epoch": 2.160542276132715,
+      "grad_norm": 0.597533643245697,
+      "learning_rate": 6.998818573892232e-05,
+      "loss": 0.3764,
+      "step": 7570
+    },
+    {
+      "epoch": 2.163396361041741,
+      "grad_norm": 0.8050810098648071,
+      "learning_rate": 6.989680431761538e-05,
+      "loss": 0.3956,
+      "step": 7580
+    },
+    {
+      "epoch": 2.166250445950767,
+      "grad_norm": 0.6903899908065796,
+      "learning_rate": 6.980534386390687e-05,
+      "loss": 0.3759,
+      "step": 7590
+    },
+    {
+      "epoch": 2.169104530859793,
+      "grad_norm": 0.9158540964126587,
+      "learning_rate": 6.971380474108824e-05,
+      "loss": 0.3651,
+      "step": 7600
+    },
+    {
+      "epoch": 2.171958615768819,
+      "grad_norm": 0.9668444395065308,
+      "learning_rate": 6.962218731276342e-05,
+      "loss": 0.3794,
+      "step": 7610
+    },
+    {
+      "epoch": 2.1748127006778453,
+      "grad_norm": 0.6928943395614624,
+      "learning_rate": 6.953049194284741e-05,
+      "loss": 0.3682,
+      "step": 7620
+    },
+    {
+      "epoch": 2.1776667855868714,
+      "grad_norm": 0.8554291129112244,
+      "learning_rate": 6.943871899556485e-05,
+      "loss": 0.3844,
+      "step": 7630
+    },
+    {
+      "epoch": 2.180520870495897,
+      "grad_norm": 1.2635998725891113,
+      "learning_rate": 6.934686883544838e-05,
+      "loss": 0.354,
+      "step": 7640
+    },
+    {
+      "epoch": 2.1833749554049233,
+      "grad_norm": 0.7581480145454407,
+      "learning_rate": 6.92549418273375e-05,
+      "loss": 0.3698,
+      "step": 7650
+    },
+    {
+      "epoch": 2.1862290403139495,
+      "grad_norm": 0.9354470372200012,
+      "learning_rate": 6.916293833637686e-05,
+      "loss": 0.3865,
+      "step": 7660
+    },
+    {
+      "epoch": 2.189083125222975,
+      "grad_norm": 0.8501242995262146,
+      "learning_rate": 6.907085872801495e-05,
+      "loss": 0.3892,
+      "step": 7670
+    },
+    {
+      "epoch": 2.1919372101320014,
+      "grad_norm": 0.6645202040672302,
+      "learning_rate": 6.897870336800258e-05,
+      "loss": 0.3712,
+      "step": 7680
+    },
+    {
+      "epoch": 2.1947912950410275,
+      "grad_norm": 0.8390302658081055,
+      "learning_rate": 6.888647262239148e-05,
+      "loss": 0.4012,
+      "step": 7690
+    },
+    {
+      "epoch": 2.1976453799500537,
+      "grad_norm": 0.5222529768943787,
+      "learning_rate": 6.879416685753281e-05,
+      "loss": 0.3852,
+      "step": 7700
+    },
+    {
+      "epoch": 2.2004994648590794,
+      "grad_norm": 0.6421943306922913,
+      "learning_rate": 6.87017864400757e-05,
+      "loss": 0.3836,
+      "step": 7710
+    },
+    {
+      "epoch": 2.2033535497681056,
+      "grad_norm": 1.3301259279251099,
+      "learning_rate": 6.860933173696584e-05,
+      "loss": 0.3573,
+      "step": 7720
+    },
+    {
+      "epoch": 2.2062076346771318,
+      "grad_norm": 0.7202717661857605,
+      "learning_rate": 6.851680311544397e-05,
+      "loss": 0.3981,
+      "step": 7730
+    },
+    {
+      "epoch": 2.2090617195861575,
+      "grad_norm": 0.8301630020141602,
+      "learning_rate": 6.842420094304447e-05,
+      "loss": 0.3652,
+      "step": 7740
+    },
+    {
+      "epoch": 2.2119158044951837,
+      "grad_norm": 0.9000500440597534,
+      "learning_rate": 6.833152558759377e-05,
+      "loss": 0.3655,
+      "step": 7750
+    },
+    {
+      "epoch": 2.21476988940421,
+      "grad_norm": 0.8134086728096008,
+      "learning_rate": 6.823877741720916e-05,
+      "loss": 0.3717,
+      "step": 7760
+    },
+    {
+      "epoch": 2.217623974313236,
+      "grad_norm": 0.8952001929283142,
+      "learning_rate": 6.814595680029701e-05,
+      "loss": 0.3841,
+      "step": 7770
+    },
+    {
+      "epoch": 2.2204780592222617,
+      "grad_norm": 1.0003637075424194,
+      "learning_rate": 6.805306410555156e-05,
+      "loss": 0.4214,
+      "step": 7780
+    },
+    {
+      "epoch": 2.223332144131288,
+      "grad_norm": 0.7732775211334229,
+      "learning_rate": 6.79600997019533e-05,
+      "loss": 0.3606,
+      "step": 7790
+    },
+    {
+      "epoch": 2.226186229040314,
+      "grad_norm": 0.7188279628753662,
+      "learning_rate": 6.786706395876754e-05,
+      "loss": 0.3716,
+      "step": 7800
+    },
+    {
+      "epoch": 2.2290403139493398,
+      "grad_norm": 0.8993227481842041,
+      "learning_rate": 6.777395724554298e-05,
+      "loss": 0.4104,
+      "step": 7810
+    },
+    {
+      "epoch": 2.231894398858366,
+      "grad_norm": 0.7618519067764282,
+      "learning_rate": 6.768077993211024e-05,
+      "loss": 0.3861,
+      "step": 7820
+    },
+    {
+      "epoch": 2.234748483767392,
+      "grad_norm": 0.7557815313339233,
+      "learning_rate": 6.758753238858038e-05,
+      "loss": 0.4025,
+      "step": 7830
+    },
+    {
+      "epoch": 2.2376025686764183,
+      "grad_norm": 1.1683645248413086,
+      "learning_rate": 6.749421498534336e-05,
+      "loss": 0.3799,
+      "step": 7840
+    },
+    {
+      "epoch": 2.240456653585444,
+      "grad_norm": 0.8091493844985962,
+      "learning_rate": 6.740082809306666e-05,
+      "loss": 0.3742,
+      "step": 7850
+    },
+    {
+      "epoch": 2.24331073849447,
+      "grad_norm": 0.9847107529640198,
+      "learning_rate": 6.730737208269382e-05,
+      "loss": 0.3502,
+      "step": 7860
+    },
+    {
+      "epoch": 2.2461648234034963,
+      "grad_norm": 1.1543819904327393,
+      "learning_rate": 6.721384732544289e-05,
+      "loss": 0.4312,
+      "step": 7870
+    },
+    {
+      "epoch": 2.2490189083125225,
+      "grad_norm": 0.9270452857017517,
+      "learning_rate": 6.712025419280497e-05,
+      "loss": 0.4035,
+      "step": 7880
+    },
+    {
+      "epoch": 2.251872993221548,
+      "grad_norm": 1.2209489345550537,
+      "learning_rate": 6.702659305654279e-05,
+      "loss": 0.3877,
+      "step": 7890
+    },
+    {
+      "epoch": 2.2547270781305744,
+      "grad_norm": 1.0458225011825562,
+      "learning_rate": 6.693286428868919e-05,
+      "loss": 0.4109,
+      "step": 7900
+    },
+    {
+      "epoch": 2.2575811630396005,
+      "grad_norm": 0.9096805453300476,
+      "learning_rate": 6.683906826154564e-05,
+      "loss": 0.4066,
+      "step": 7910
+    },
+    {
+      "epoch": 2.2604352479486263,
+      "grad_norm": 1.024956226348877,
+      "learning_rate": 6.674520534768077e-05,
+      "loss": 0.4015,
+      "step": 7920
+    },
+    {
+      "epoch": 2.2632893328576524,
+      "grad_norm": 1.0131760835647583,
+      "learning_rate": 6.665127591992892e-05,
+      "loss": 0.3277,
+      "step": 7930
+    },
+    {
+      "epoch": 2.2661434177666786,
+      "grad_norm": 0.6009324193000793,
+      "learning_rate": 6.65572803513886e-05,
+      "loss": 0.3848,
+      "step": 7940
+    },
+    {
+      "epoch": 2.2689975026757048,
+      "grad_norm": 0.8868556618690491,
+      "learning_rate": 6.646321901542105e-05,
+      "loss": 0.3915,
+      "step": 7950
+    },
+    {
+      "epoch": 2.2718515875847305,
+      "grad_norm": 0.8528206944465637,
+      "learning_rate": 6.636909228564874e-05,
+      "loss": 0.3419,
+      "step": 7960
+    },
+    {
+      "epoch": 2.2747056724937567,
+      "grad_norm": 1.19645357131958,
+      "learning_rate": 6.627490053595391e-05,
+      "loss": 0.38,
+      "step": 7970
+    },
+    {
+      "epoch": 2.277559757402783,
+      "grad_norm": 0.7129682898521423,
+      "learning_rate": 6.618064414047705e-05,
+      "loss": 0.3765,
+      "step": 7980
+    },
+    {
+      "epoch": 2.280413842311809,
+      "grad_norm": 0.7732852101325989,
+      "learning_rate": 6.608632347361541e-05,
+      "loss": 0.4087,
+      "step": 7990
+    },
+    {
+      "epoch": 2.2832679272208347,
+      "grad_norm": 0.8368834257125854,
+      "learning_rate": 6.599193891002157e-05,
+      "loss": 0.3987,
+      "step": 8000
+    },
+    {
+      "epoch": 2.286122012129861,
+      "grad_norm": 1.0063822269439697,
+      "learning_rate": 6.589749082460195e-05,
+      "loss": 0.3933,
+      "step": 8010
+    },
+    {
+      "epoch": 2.288976097038887,
+      "grad_norm": 1.067052960395813,
+      "learning_rate": 6.580297959251518e-05,
+      "loss": 0.3943,
+      "step": 8020
+    },
+    {
+      "epoch": 2.2918301819479128,
+      "grad_norm": 0.8661326766014099,
+      "learning_rate": 6.570840558917077e-05,
+      "loss": 0.3847,
+      "step": 8030
+    },
+    {
+      "epoch": 2.294684266856939,
+      "grad_norm": 0.9897991418838501,
+      "learning_rate": 6.561376919022762e-05,
+      "loss": 0.3876,
+      "step": 8040
+    },
+    {
+      "epoch": 2.297538351765965,
+      "grad_norm": 0.8766522407531738,
+      "learning_rate": 6.551907077159238e-05,
+      "loss": 0.4062,
+      "step": 8050
+    },
+    {
+      "epoch": 2.3003924366749913,
+      "grad_norm": 0.7486727237701416,
+      "learning_rate": 6.542431070941813e-05,
+      "loss": 0.4079,
+      "step": 8060
+    },
+    {
+      "epoch": 2.303246521584017,
+      "grad_norm": 0.877748429775238,
+      "learning_rate": 6.532948938010271e-05,
+      "loss": 0.3774,
+      "step": 8070
+    },
+    {
+      "epoch": 2.306100606493043,
+      "grad_norm": 0.9829452633857727,
+      "learning_rate": 6.523460716028743e-05,
+      "loss": 0.4179,
+      "step": 8080
+    },
+    {
+      "epoch": 2.3089546914020693,
+      "grad_norm": 0.5957035422325134,
+      "learning_rate": 6.513966442685537e-05,
+      "loss": 0.3501,
+      "step": 8090
+    },
+    {
+      "epoch": 2.3118087763110955,
+      "grad_norm": 0.8652012944221497,
+      "learning_rate": 6.504466155693004e-05,
+      "loss": 0.3715,
+      "step": 8100
+    },
+    {
+      "epoch": 2.314662861220121,
+      "grad_norm": 1.0595470666885376,
+      "learning_rate": 6.494959892787381e-05,
+      "loss": 0.3591,
+      "step": 8110
+    },
+    {
+      "epoch": 2.3175169461291474,
+      "grad_norm": 0.9246557354927063,
+      "learning_rate": 6.485447691728637e-05,
+      "loss": 0.4321,
+      "step": 8120
+    },
+    {
+      "epoch": 2.3203710310381735,
+      "grad_norm": 0.5890812873840332,
+      "learning_rate": 6.475929590300335e-05,
+      "loss": 0.3867,
+      "step": 8130
+    },
+    {
+      "epoch": 2.3232251159471993,
+      "grad_norm": 0.6471204161643982,
+      "learning_rate": 6.466405626309469e-05,
+      "loss": 0.3735,
+      "step": 8140
+    },
+    {
+      "epoch": 2.3260792008562254,
+      "grad_norm": 0.7558359503746033,
+      "learning_rate": 6.456875837586324e-05,
+      "loss": 0.3897,
+      "step": 8150
+    },
+    {
+      "epoch": 2.3289332857652516,
+      "grad_norm": 0.7475236654281616,
+      "learning_rate": 6.447340261984323e-05,
+      "loss": 0.3864,
+      "step": 8160
+    },
+    {
+      "epoch": 2.3317873706742773,
+      "grad_norm": 0.7441720366477966,
+      "learning_rate": 6.437798937379865e-05,
+      "loss": 0.4012,
+      "step": 8170
+    },
+    {
+      "epoch": 2.3346414555833035,
+      "grad_norm": 0.7266222834587097,
+      "learning_rate": 6.428251901672197e-05,
+      "loss": 0.3948,
+      "step": 8180
+    },
+    {
+      "epoch": 2.3374955404923297,
+      "grad_norm": 1.0292062759399414,
+      "learning_rate": 6.418699192783245e-05,
+      "loss": 0.4245,
+      "step": 8190
+    },
+    {
+      "epoch": 2.340349625401356,
+      "grad_norm": 0.809492826461792,
+      "learning_rate": 6.409140848657469e-05,
+      "loss": 0.4159,
+      "step": 8200
+    },
+    {
+      "epoch": 2.3432037103103815,
+      "grad_norm": 0.8932891488075256,
+      "learning_rate": 6.399576907261717e-05,
+      "loss": 0.3748,
+      "step": 8210
+    },
+    {
+      "epoch": 2.3460577952194077,
+      "grad_norm": 1.2191537618637085,
+      "learning_rate": 6.390007406585064e-05,
+      "loss": 0.4075,
+      "step": 8220
+    },
+    {
+      "epoch": 2.348911880128434,
+      "grad_norm": 0.950178325176239,
+      "learning_rate": 6.380432384638672e-05,
+      "loss": 0.4026,
+      "step": 8230
+    },
+    {
+      "epoch": 2.35176596503746,
+      "grad_norm": 1.0088558197021484,
+      "learning_rate": 6.370851879455635e-05,
+      "loss": 0.393,
+      "step": 8240
+    },
+    {
+      "epoch": 2.3546200499464858,
+      "grad_norm": 0.860796332359314,
+      "learning_rate": 6.361265929090819e-05,
+      "loss": 0.389,
+      "step": 8250
+    },
+    {
+      "epoch": 2.357474134855512,
+      "grad_norm": 0.9812953472137451,
+      "learning_rate": 6.351674571620731e-05,
+      "loss": 0.4121,
+      "step": 8260
+    },
+    {
+      "epoch": 2.360328219764538,
+      "grad_norm": 0.8625460267066956,
+      "learning_rate": 6.342077845143344e-05,
+      "loss": 0.4058,
+      "step": 8270
+    },
+    {
+      "epoch": 2.363182304673564,
+      "grad_norm": 1.086666464805603,
+      "learning_rate": 6.332475787777965e-05,
+      "loss": 0.4004,
+      "step": 8280
+    },
+    {
+      "epoch": 2.36603638958259,
+      "grad_norm": 0.8118520379066467,
+      "learning_rate": 6.322868437665071e-05,
+      "loss": 0.3974,
+      "step": 8290
+    },
+    {
+      "epoch": 2.368890474491616,
+      "grad_norm": 1.0880775451660156,
+      "learning_rate": 6.313255832966168e-05,
+      "loss": 0.4076,
+      "step": 8300
+    },
+    {
+      "epoch": 2.3717445594006423,
+      "grad_norm": 0.7525433897972107,
+      "learning_rate": 6.303638011863629e-05,
+      "loss": 0.383,
+      "step": 8310
+    },
+    {
+      "epoch": 2.374598644309668,
+      "grad_norm": 0.865863025188446,
+      "learning_rate": 6.294015012560547e-05,
+      "loss": 0.3633,
+      "step": 8320
+    },
+    {
+      "epoch": 2.377452729218694,
+      "grad_norm": 0.8211045861244202,
+      "learning_rate": 6.284386873280589e-05,
+      "loss": 0.3665,
+      "step": 8330
+    },
+    {
+      "epoch": 2.3803068141277204,
+      "grad_norm": 0.8563920855522156,
+      "learning_rate": 6.27475363226783e-05,
+      "loss": 0.3984,
+      "step": 8340
+    },
+    {
+      "epoch": 2.3831608990367465,
+      "grad_norm": 0.7572536468505859,
+      "learning_rate": 6.265115327786618e-05,
+      "loss": 0.3898,
+      "step": 8350
+    },
+    {
+      "epoch": 2.3860149839457723,
+      "grad_norm": 0.7613979578018188,
+      "learning_rate": 6.255471998121408e-05,
+      "loss": 0.3699,
+      "step": 8360
+    },
+    {
+      "epoch": 2.3888690688547984,
+      "grad_norm": 0.941780149936676,
+      "learning_rate": 6.245823681576619e-05,
+      "loss": 0.4069,
+      "step": 8370
+    },
+    {
+      "epoch": 2.3917231537638246,
+      "grad_norm": 1.2750709056854248,
+      "learning_rate": 6.236170416476478e-05,
+      "loss": 0.3912,
+      "step": 8380
+    },
+    {
+      "epoch": 2.3945772386728503,
+      "grad_norm": 1.0106887817382812,
+      "learning_rate": 6.226512241164866e-05,
+      "loss": 0.3606,
+      "step": 8390
+    },
+    {
+      "epoch": 2.3974313235818765,
+      "grad_norm": 1.0778266191482544,
+      "learning_rate": 6.216849194005166e-05,
+      "loss": 0.3923,
+      "step": 8400
+    },
+    {
+      "epoch": 2.4002854084909027,
+      "grad_norm": 0.8825225234031677,
+      "learning_rate": 6.207181313380123e-05,
+      "loss": 0.3739,
+      "step": 8410
+    },
+    {
+      "epoch": 2.403139493399929,
+      "grad_norm": 0.9134084582328796,
+      "learning_rate": 6.197508637691668e-05,
+      "loss": 0.4142,
+      "step": 8420
+    },
+    {
+      "epoch": 2.4059935783089546,
+      "grad_norm": 0.9855687022209167,
+      "learning_rate": 6.187831205360785e-05,
+      "loss": 0.3932,
+      "step": 8430
+    },
+    {
+      "epoch": 2.4088476632179807,
+      "grad_norm": 1.0128726959228516,
+      "learning_rate": 6.178149054827356e-05,
+      "loss": 0.385,
+      "step": 8440
+    },
+    {
+      "epoch": 2.411701748127007,
+      "grad_norm": 0.6451798677444458,
+      "learning_rate": 6.168462224549996e-05,
+      "loss": 0.355,
+      "step": 8450
+    },
+    {
+      "epoch": 2.414555833036033,
+      "grad_norm": 0.6846132874488831,
+      "learning_rate": 6.158770753005911e-05,
+      "loss": 0.4573,
+      "step": 8460
+    },
+    {
+      "epoch": 2.4174099179450588,
+      "grad_norm": 0.8105731010437012,
+      "learning_rate": 6.149074678690746e-05,
+      "loss": 0.3789,
+      "step": 8470
+    },
+    {
+      "epoch": 2.420264002854085,
+      "grad_norm": 0.7642350196838379,
+      "learning_rate": 6.139374040118426e-05,
+      "loss": 0.4019,
+      "step": 8480
+    },
+    {
+      "epoch": 2.423118087763111,
+      "grad_norm": 0.7982140183448792,
+      "learning_rate": 6.129668875821007e-05,
+      "loss": 0.3762,
+      "step": 8490
+    },
+    {
+      "epoch": 2.425972172672137,
+      "grad_norm": 0.8663055896759033,
+      "learning_rate": 6.119959224348519e-05,
+      "loss": 0.3672,
+      "step": 8500
+    },
+    {
+      "epoch": 2.428826257581163,
+      "grad_norm": 1.0429260730743408,
+      "learning_rate": 6.110245124268818e-05,
+      "loss": 0.4031,
+      "step": 8510
+    },
+    {
+      "epoch": 2.431680342490189,
+      "grad_norm": 1.0244431495666504,
+      "learning_rate": 6.100526614167432e-05,
+      "loss": 0.3805,
+      "step": 8520
+    },
+    {
+      "epoch": 2.434534427399215,
+      "grad_norm": 0.8987639546394348,
+      "learning_rate": 6.090803732647402e-05,
+      "loss": 0.3469,
+      "step": 8530
+    },
+    {
+      "epoch": 2.437388512308241,
+      "grad_norm": 0.9394255876541138,
+      "learning_rate": 6.081076518329137e-05,
+      "loss": 0.3863,
+      "step": 8540
+    },
+    {
+      "epoch": 2.440242597217267,
+      "grad_norm": 1.006545901298523,
+      "learning_rate": 6.071345009850251e-05,
+      "loss": 0.4063,
+      "step": 8550
+    },
+    {
+      "epoch": 2.4430966821262934,
+      "grad_norm": 0.7682129144668579,
+      "learning_rate": 6.0616092458654194e-05,
+      "loss": 0.3752,
+      "step": 8560
+    },
+    {
+      "epoch": 2.445950767035319,
+      "grad_norm": 0.7381885051727295,
+      "learning_rate": 6.0518692650462206e-05,
+      "loss": 0.3717,
+      "step": 8570
+    },
+    {
+      "epoch": 2.4488048519443453,
+      "grad_norm": 0.8551028966903687,
+      "learning_rate": 6.04212510608098e-05,
+      "loss": 0.3834,
+      "step": 8580
+    },
+    {
+      "epoch": 2.4516589368533714,
+      "grad_norm": 0.8665489554405212,
+      "learning_rate": 6.032376807674623e-05,
+      "loss": 0.3742,
+      "step": 8590
+    },
+    {
+      "epoch": 2.4545130217623976,
+      "grad_norm": 0.8613978028297424,
+      "learning_rate": 6.0226244085485136e-05,
+      "loss": 0.3688,
+      "step": 8600
+    },
+    {
+      "epoch": 2.4573671066714233,
+      "grad_norm": 1.0067691802978516,
+      "learning_rate": 6.0128679474403084e-05,
+      "loss": 0.4014,
+      "step": 8610
+    },
+    {
+      "epoch": 2.4602211915804495,
+      "grad_norm": 0.8539415001869202,
+      "learning_rate": 6.003107463103792e-05,
+      "loss": 0.335,
+      "step": 8620
+    },
+    {
+      "epoch": 2.4630752764894757,
+      "grad_norm": 0.759884774684906,
+      "learning_rate": 5.993342994308737e-05,
+      "loss": 0.362,
+      "step": 8630
+    },
+    {
+      "epoch": 2.4659293613985014,
+      "grad_norm": 0.7755060195922852,
+      "learning_rate": 5.9835745798407405e-05,
+      "loss": 0.3872,
+      "step": 8640
+    },
+    {
+      "epoch": 2.4687834463075276,
+      "grad_norm": 0.7879381775856018,
+      "learning_rate": 5.973802258501069e-05,
+      "loss": 0.3754,
+      "step": 8650
+    },
+    {
+      "epoch": 2.4716375312165537,
+      "grad_norm": 0.7982614636421204,
+      "learning_rate": 5.96402606910651e-05,
+      "loss": 0.3608,
+      "step": 8660
+    },
+    {
+      "epoch": 2.47449161612558,
+      "grad_norm": 1.1680744886398315,
+      "learning_rate": 5.954246050489216e-05,
+      "loss": 0.3877,
+      "step": 8670
+    },
+    {
+      "epoch": 2.4773457010346056,
+      "grad_norm": 1.4971110820770264,
+      "learning_rate": 5.944462241496549e-05,
+      "loss": 0.3924,
+      "step": 8680
+    },
+    {
+      "epoch": 2.4801997859436318,
+      "grad_norm": 1.1304658651351929,
+      "learning_rate": 5.934674680990928e-05,
+      "loss": 0.3722,
+      "step": 8690
+    },
+    {
+      "epoch": 2.483053870852658,
+      "grad_norm": 1.0444248914718628,
+      "learning_rate": 5.924883407849672e-05,
+      "loss": 0.3809,
+      "step": 8700
+    },
+    {
+      "epoch": 2.485907955761684,
+      "grad_norm": 1.4929407835006714,
+      "learning_rate": 5.9150884609648446e-05,
+      "loss": 0.3922,
+      "step": 8710
+    },
+    {
+      "epoch": 2.48876204067071,
+      "grad_norm": 0.7926280498504639,
+      "learning_rate": 5.905289879243107e-05,
+      "loss": 0.3578,
+      "step": 8720
+    },
+    {
+      "epoch": 2.491616125579736,
+      "grad_norm": 0.7616733908653259,
+      "learning_rate": 5.8954877016055534e-05,
+      "loss": 0.3914,
+      "step": 8730
+    },
+    {
+      "epoch": 2.494470210488762,
+      "grad_norm": 0.8602902293205261,
+      "learning_rate": 5.885681966987568e-05,
+      "loss": 0.3923,
+      "step": 8740
+    },
+    {
+      "epoch": 2.497324295397788,
+      "grad_norm": 1.0538463592529297,
+      "learning_rate": 5.875872714338657e-05,
+      "loss": 0.3908,
+      "step": 8750
+    },
+    {
+      "epoch": 2.500178380306814,
+      "grad_norm": 1.175032377243042,
+      "learning_rate": 5.866059982622304e-05,
+      "loss": 0.3953,
+      "step": 8760
+    },
+    {
+      "epoch": 2.50303246521584,
+      "grad_norm": 0.7533116936683655,
+      "learning_rate": 5.85624381081581e-05,
+      "loss": 0.4135,
+      "step": 8770
+    },
+    {
+      "epoch": 2.505886550124866,
+      "grad_norm": 0.8190605640411377,
+      "learning_rate": 5.846424237910143e-05,
+      "loss": 0.384,
+      "step": 8780
+    },
+    {
+      "epoch": 2.508740635033892,
+      "grad_norm": 1.0056654214859009,
+      "learning_rate": 5.8366013029097776e-05,
+      "loss": 0.4046,
+      "step": 8790
+    },
+    {
+      "epoch": 2.5115947199429183,
+      "grad_norm": 0.893796980381012,
+      "learning_rate": 5.826775044832547e-05,
+      "loss": 0.3767,
+      "step": 8800
+    },
+    {
+      "epoch": 2.5144488048519444,
+      "grad_norm": 1.019010066986084,
+      "learning_rate": 5.81694550270948e-05,
+      "loss": 0.4287,
+      "step": 8810
+    },
+    {
+      "epoch": 2.5173028897609706,
+      "grad_norm": 0.836607813835144,
+      "learning_rate": 5.8071127155846514e-05,
+      "loss": 0.3467,
+      "step": 8820
+    },
+    {
+      "epoch": 2.5201569746699963,
+      "grad_norm": 0.9914632439613342,
+      "learning_rate": 5.797276722515026e-05,
+      "loss": 0.3973,
+      "step": 8830
+    },
+    {
+      "epoch": 2.5230110595790225,
+      "grad_norm": 1.2152297496795654,
+      "learning_rate": 5.7874375625703036e-05,
+      "loss": 0.3497,
+      "step": 8840
+    },
+    {
+      "epoch": 2.5258651444880487,
+      "grad_norm": 1.1097205877304077,
+      "learning_rate": 5.77759527483276e-05,
+      "loss": 0.3637,
+      "step": 8850
+    },
+    {
+      "epoch": 2.5287192293970744,
+      "grad_norm": 0.9172791242599487,
+      "learning_rate": 5.7677498983971e-05,
+      "loss": 0.4027,
+      "step": 8860
+    },
+    {
+      "epoch": 2.5315733143061006,
+      "grad_norm": 0.9831005334854126,
+      "learning_rate": 5.757901472370294e-05,
+      "loss": 0.359,
+      "step": 8870
+    },
+    {
+      "epoch": 2.5344273992151267,
+      "grad_norm": 0.9835144281387329,
+      "learning_rate": 5.748050035871424e-05,
+      "loss": 0.3866,
+      "step": 8880
+    },
+    {
+      "epoch": 2.5372814841241524,
+      "grad_norm": 0.8125594854354858,
+      "learning_rate": 5.738195628031534e-05,
+      "loss": 0.4012,
+      "step": 8890
+    },
+    {
+      "epoch": 2.5401355690331786,
+      "grad_norm": 1.133615255355835,
+      "learning_rate": 5.728338287993467e-05,
+      "loss": 0.3997,
+      "step": 8900
+    },
+    {
+      "epoch": 2.5429896539422048,
+      "grad_norm": 1.015852928161621,
+      "learning_rate": 5.718478054911717e-05,
+      "loss": 0.3674,
+      "step": 8910
+    },
+    {
+      "epoch": 2.545843738851231,
+      "grad_norm": 1.3895350694656372,
+      "learning_rate": 5.708614967952265e-05,
+      "loss": 0.3711,
+      "step": 8920
+    },
+    {
+      "epoch": 2.548697823760257,
+      "grad_norm": 0.8260318040847778,
+      "learning_rate": 5.69874906629243e-05,
+      "loss": 0.3899,
+      "step": 8930
+    },
+    {
+      "epoch": 2.551551908669283,
+      "grad_norm": 0.8439449071884155,
+      "learning_rate": 5.688880389120711e-05,
+      "loss": 0.3935,
+      "step": 8940
+    },
+    {
+      "epoch": 2.554405993578309,
+      "grad_norm": 0.9614757299423218,
+      "learning_rate": 5.679008975636635e-05,
+      "loss": 0.3754,
+      "step": 8950
+    },
+    {
+      "epoch": 2.557260078487335,
+      "grad_norm": 0.9397571682929993,
+      "learning_rate": 5.6691348650505896e-05,
+      "loss": 0.4068,
+      "step": 8960
+    },
+    {
+      "epoch": 2.560114163396361,
+      "grad_norm": 0.6689046621322632,
+      "learning_rate": 5.659258096583687e-05,
+      "loss": 0.3683,
+      "step": 8970
+    },
+    {
+      "epoch": 2.562968248305387,
+      "grad_norm": 0.9342771172523499,
+      "learning_rate": 5.6493787094675856e-05,
+      "loss": 0.425,
+      "step": 8980
+    },
+    {
+      "epoch": 2.565822333214413,
+      "grad_norm": 0.8137280941009521,
+      "learning_rate": 5.639496742944352e-05,
+      "loss": 0.4074,
+      "step": 8990
+    },
+    {
+      "epoch": 2.568676418123439,
+      "grad_norm": 0.8159138560295105,
+      "learning_rate": 5.6296122362662996e-05,
+      "loss": 0.3363,
+      "step": 9000
+    },
+    {
+      "epoch": 2.571530503032465,
+      "grad_norm": 0.939554750919342,
+      "learning_rate": 5.619725228695825e-05,
+      "loss": 0.36,
+      "step": 9010
+    },
+    {
+      "epoch": 2.5743845879414913,
+      "grad_norm": 1.2181627750396729,
+      "learning_rate": 5.609835759505267e-05,
+      "loss": 0.3942,
+      "step": 9020
+    },
+    {
+      "epoch": 2.5772386728505174,
+      "grad_norm": 1.0984028577804565,
+      "learning_rate": 5.599943867976733e-05,
+      "loss": 0.4121,
+      "step": 9030
+    },
+    {
+      "epoch": 2.5800927577595436,
+      "grad_norm": 0.941370964050293,
+      "learning_rate": 5.590049593401959e-05,
+      "loss": 0.3771,
+      "step": 9040
+    },
+    {
+      "epoch": 2.5829468426685693,
+      "grad_norm": 0.7422721982002258,
+      "learning_rate": 5.580152975082146e-05,
+      "loss": 0.3635,
+      "step": 9050
+    },
+    {
+      "epoch": 2.5858009275775955,
+      "grad_norm": 1.314098596572876,
+      "learning_rate": 5.5702540523278036e-05,
+      "loss": 0.3989,
+      "step": 9060
+    },
+    {
+      "epoch": 2.5886550124866217,
+      "grad_norm": 1.0986342430114746,
+      "learning_rate": 5.560352864458592e-05,
+      "loss": 0.39,
+      "step": 9070
+    },
+    {
+      "epoch": 2.5915090973956474,
+      "grad_norm": 1.1221035718917847,
+      "learning_rate": 5.550449450803175e-05,
+      "loss": 0.4027,
+      "step": 9080
+    },
+    {
+      "epoch": 2.5943631823046736,
+      "grad_norm": 0.909200131893158,
+      "learning_rate": 5.54054385069905e-05,
+      "loss": 0.3396,
+      "step": 9090
+    },
+    {
+      "epoch": 2.5972172672136997,
+      "grad_norm": 0.7582272291183472,
+      "learning_rate": 5.530636103492407e-05,
+      "loss": 0.3925,
+      "step": 9100
+    },
+    {
+      "epoch": 2.6000713521227254,
+      "grad_norm": 1.0211666822433472,
+      "learning_rate": 5.5207262485379564e-05,
+      "loss": 0.4029,
+      "step": 9110
+    },
+    {
+      "epoch": 2.6029254370317516,
+      "grad_norm": 1.0995800495147705,
+      "learning_rate": 5.51081432519879e-05,
+      "loss": 0.3668,
+      "step": 9120
+    },
+    {
+      "epoch": 2.6057795219407778,
+      "grad_norm": 0.9261085987091064,
+      "learning_rate": 5.500900372846206e-05,
+      "loss": 0.3792,
+      "step": 9130
+    },
+    {
+      "epoch": 2.608633606849804,
+      "grad_norm": 0.9360781908035278,
+      "learning_rate": 5.490984430859568e-05,
+      "loss": 0.3989,
+      "step": 9140
+    },
+    {
+      "epoch": 2.6114876917588297,
+      "grad_norm": 0.9865853190422058,
+      "learning_rate": 5.481066538626144e-05,
+      "loss": 0.3892,
+      "step": 9150
+    },
+    {
+      "epoch": 2.614341776667856,
+      "grad_norm": 1.0475095510482788,
+      "learning_rate": 5.4711467355409417e-05,
+      "loss": 0.4068,
+      "step": 9160
+    },
+    {
+      "epoch": 2.617195861576882,
+      "grad_norm": 0.7886587977409363,
+      "learning_rate": 5.4612250610065664e-05,
+      "loss": 0.3695,
+      "step": 9170
+    },
+    {
+      "epoch": 2.620049946485908,
+      "grad_norm": 1.201371192932129,
+      "learning_rate": 5.451301554433052e-05,
+      "loss": 0.3933,
+      "step": 9180
+    },
+    {
+      "epoch": 2.622904031394934,
+      "grad_norm": 0.7625223994255066,
+      "learning_rate": 5.4413762552377114e-05,
+      "loss": 0.3658,
+      "step": 9190
+    },
+    {
+      "epoch": 2.62575811630396,
+      "grad_norm": 1.2264348268508911,
+      "learning_rate": 5.431449202844976e-05,
+      "loss": 0.4036,
+      "step": 9200
+    },
+    {
+      "epoch": 2.628612201212986,
+      "grad_norm": 1.4466603994369507,
+      "learning_rate": 5.421520436686246e-05,
+      "loss": 0.3718,
+      "step": 9210
+    },
+    {
+      "epoch": 2.631466286122012,
+      "grad_norm": 0.7877131700515747,
+      "learning_rate": 5.411589996199724e-05,
+      "loss": 0.4057,
+      "step": 9220
+    },
+    {
+      "epoch": 2.634320371031038,
+      "grad_norm": 0.8216263055801392,
+      "learning_rate": 5.401657920830265e-05,
+      "loss": 0.3954,
+      "step": 9230
+    },
+    {
+      "epoch": 2.6371744559400643,
+      "grad_norm": 0.8210817575454712,
+      "learning_rate": 5.39172425002922e-05,
+      "loss": 0.3599,
+      "step": 9240
+    },
+    {
+      "epoch": 2.64002854084909,
+      "grad_norm": 0.896376371383667,
+      "learning_rate": 5.381789023254272e-05,
+      "loss": 0.3637,
+      "step": 9250
+    },
+    {
+      "epoch": 2.642882625758116,
+      "grad_norm": 0.9731716513633728,
+      "learning_rate": 5.37185227996929e-05,
+      "loss": 0.4001,
+      "step": 9260
+    },
+    {
+      "epoch": 2.6457367106671423,
+      "grad_norm": 1.031147837638855,
+      "learning_rate": 5.361914059644162e-05,
+      "loss": 0.3686,
+      "step": 9270
+    },
+    {
+      "epoch": 2.6485907955761685,
+      "grad_norm": 0.7797552347183228,
+      "learning_rate": 5.351974401754648e-05,
+      "loss": 0.3762,
+      "step": 9280
+    },
+    {
+      "epoch": 2.6514448804851947,
+      "grad_norm": 1.0218497514724731,
+      "learning_rate": 5.3420333457822146e-05,
+      "loss": 0.376,
+      "step": 9290
+    },
+    {
+      "epoch": 2.6542989653942204,
+      "grad_norm": 0.913997232913971,
+      "learning_rate": 5.332090931213882e-05,
+      "loss": 0.3854,
+      "step": 9300
+    },
+    {
+      "epoch": 2.6571530503032466,
+      "grad_norm": 0.8870353698730469,
+      "learning_rate": 5.322147197542068e-05,
+      "loss": 0.3896,
+      "step": 9310
+    },
+    {
+      "epoch": 2.6600071352122727,
+      "grad_norm": 1.233062744140625,
+      "learning_rate": 5.31220218426443e-05,
+      "loss": 0.4101,
+      "step": 9320
+    },
+    {
+      "epoch": 2.6628612201212984,
+      "grad_norm": 0.8647748827934265,
+      "learning_rate": 5.302255930883706e-05,
+      "loss": 0.3443,
+      "step": 9330
+    },
+    {
+      "epoch": 2.6657153050303246,
+      "grad_norm": 1.381943941116333,
+      "learning_rate": 5.292308476907566e-05,
+      "loss": 0.3944,
+      "step": 9340
+    },
+    {
+      "epoch": 2.6685693899393508,
+      "grad_norm": 0.845750629901886,
+      "learning_rate": 5.282359861848439e-05,
+      "loss": 0.3792,
+      "step": 9350
+    },
+    {
+      "epoch": 2.6714234748483765,
+      "grad_norm": 0.7501406669616699,
+      "learning_rate": 5.272410125223376e-05,
+      "loss": 0.3303,
+      "step": 9360
+    },
+    {
+      "epoch": 2.6742775597574027,
+      "grad_norm": 0.8577442169189453,
+      "learning_rate": 5.262459306553875e-05,
+      "loss": 0.3723,
+      "step": 9370
+    },
+    {
+      "epoch": 2.677131644666429,
+      "grad_norm": 1.1972564458847046,
+      "learning_rate": 5.2525074453657374e-05,
+      "loss": 0.3973,
+      "step": 9380
+    },
+    {
+      "epoch": 2.679985729575455,
+      "grad_norm": 0.7672300338745117,
+      "learning_rate": 5.242554581188903e-05,
+      "loss": 0.3579,
+      "step": 9390
+    },
+    {
+      "epoch": 2.682839814484481,
+      "grad_norm": 1.2907828092575073,
+      "learning_rate": 5.232600753557298e-05,
+      "loss": 0.4231,
+      "step": 9400
+    },
+    {
+      "epoch": 2.685693899393507,
+      "grad_norm": 0.7546185255050659,
+      "learning_rate": 5.222646002008672e-05,
+      "loss": 0.3878,
+      "step": 9410
+    },
+    {
+      "epoch": 2.688547984302533,
+      "grad_norm": 0.8427802324295044,
+      "learning_rate": 5.212690366084445e-05,
+      "loss": 0.3504,
+      "step": 9420
+    },
+    {
+      "epoch": 2.6914020692115592,
+      "grad_norm": 1.0579659938812256,
+      "learning_rate": 5.2027338853295536e-05,
+      "loss": 0.3878,
+      "step": 9430
+    },
+    {
+      "epoch": 2.694256154120585,
+      "grad_norm": 0.9414021372795105,
+      "learning_rate": 5.192776599292286e-05,
+      "loss": 0.3508,
+      "step": 9440
+    },
+    {
+      "epoch": 2.697110239029611,
+      "grad_norm": 1.111647129058838,
+      "learning_rate": 5.182818547524134e-05,
+      "loss": 0.3704,
+      "step": 9450
+    },
+    {
+      "epoch": 2.6999643239386373,
+      "grad_norm": 1.0562825202941895,
+      "learning_rate": 5.1728597695796234e-05,
+      "loss": 0.3811,
+      "step": 9460
+    },
+    {
+      "epoch": 2.702818408847663,
+      "grad_norm": 1.2730356454849243,
+      "learning_rate": 5.1629003050161716e-05,
+      "loss": 0.3799,
+      "step": 9470
+    },
+    {
+      "epoch": 2.705672493756689,
+      "grad_norm": 0.8649618029594421,
+      "learning_rate": 5.152940193393919e-05,
+      "loss": 0.3863,
+      "step": 9480
+    },
+    {
+      "epoch": 2.7085265786657153,
+      "grad_norm": 0.9491603970527649,
+      "learning_rate": 5.14297947427558e-05,
+      "loss": 0.386,
+      "step": 9490
+    },
+    {
+      "epoch": 2.7113806635747415,
+      "grad_norm": 0.9528137445449829,
+      "learning_rate": 5.133018187226277e-05,
+      "loss": 0.3943,
+      "step": 9500
+    },
+    {
+      "epoch": 2.7142347484837677,
+      "grad_norm": 0.7841125726699829,
+      "learning_rate": 5.1230563718133915e-05,
+      "loss": 0.3487,
+      "step": 9510
+    },
+    {
+      "epoch": 2.7170888333927934,
+      "grad_norm": 1.0835132598876953,
+      "learning_rate": 5.113094067606407e-05,
+      "loss": 0.4113,
+      "step": 9520
+    },
+    {
+      "epoch": 2.7199429183018196,
+      "grad_norm": 1.1187998056411743,
+      "learning_rate": 5.103131314176739e-05,
+      "loss": 0.375,
+      "step": 9530
+    },
+    {
+      "epoch": 2.7227970032108457,
+      "grad_norm": 0.9432478547096252,
+      "learning_rate": 5.093168151097598e-05,
+      "loss": 0.3962,
+      "step": 9540
+    },
+    {
+      "epoch": 2.7256510881198714,
+      "grad_norm": 1.0071675777435303,
+      "learning_rate": 5.083204617943814e-05,
+      "loss": 0.3828,
+      "step": 9550
+    },
+    {
+      "epoch": 2.7285051730288976,
+      "grad_norm": 1.2895485162734985,
+      "learning_rate": 5.073240754291693e-05,
+      "loss": 0.3916,
+      "step": 9560
+    },
+    {
+      "epoch": 2.731359257937924,
+      "grad_norm": 0.9218833446502686,
+      "learning_rate": 5.063276599718846e-05,
+      "loss": 0.3465,
+      "step": 9570
+    },
+    {
+      "epoch": 2.7342133428469495,
+      "grad_norm": 1.1650524139404297,
+      "learning_rate": 5.053312193804048e-05,
+      "loss": 0.3797,
+      "step": 9580
+    },
+    {
+      "epoch": 2.7370674277559757,
+      "grad_norm": 1.2409383058547974,
+      "learning_rate": 5.043347576127067e-05,
+      "loss": 0.3746,
+      "step": 9590
+    },
+    {
+      "epoch": 2.739921512665002,
+      "grad_norm": 1.321643352508545,
+      "learning_rate": 5.0333827862685137e-05,
+      "loss": 0.3716,
+      "step": 9600
+    },
+    {
+      "epoch": 2.7427755975740276,
+      "grad_norm": 1.0632143020629883,
+      "learning_rate": 5.023417863809684e-05,
+      "loss": 0.39,
+      "step": 9610
+    },
+    {
+      "epoch": 2.7456296824830537,
+      "grad_norm": 1.0268093347549438,
+      "learning_rate": 5.013452848332398e-05,
+      "loss": 0.3529,
+      "step": 9620
+    },
+    {
+      "epoch": 2.74848376739208,
+      "grad_norm": 0.9910897612571716,
+      "learning_rate": 5.003487779418846e-05,
+      "loss": 0.381,
+      "step": 9630
+    },
+    {
+      "epoch": 2.751337852301106,
+      "grad_norm": 0.8758265376091003,
+      "learning_rate": 4.993522696651432e-05,
+      "loss": 0.4047,
+      "step": 9640
+    },
+    {
+      "epoch": 2.7541919372101322,
+      "grad_norm": 0.7053985595703125,
+      "learning_rate": 4.983557639612614e-05,
+      "loss": 0.3733,
+      "step": 9650
+    },
+    {
+      "epoch": 2.757046022119158,
+      "grad_norm": 1.3241455554962158,
+      "learning_rate": 4.9735926478847495e-05,
+      "loss": 0.3889,
+      "step": 9660
+    },
+    {
+      "epoch": 2.759900107028184,
+      "grad_norm": 1.361720323562622,
+      "learning_rate": 4.9636277610499324e-05,
+      "loss": 0.3571,
+      "step": 9670
+    },
+    {
+      "epoch": 2.7627541919372103,
+      "grad_norm": 1.1579484939575195,
+      "learning_rate": 4.953663018689843e-05,
+      "loss": 0.3754,
+      "step": 9680
+    },
+    {
+      "epoch": 2.765608276846236,
+      "grad_norm": 0.9311447143554688,
+      "learning_rate": 4.943698460385587e-05,
+      "loss": 0.3587,
+      "step": 9690
+    },
+    {
+      "epoch": 2.768462361755262,
+      "grad_norm": 1.0010428428649902,
+      "learning_rate": 4.933734125717542e-05,
+      "loss": 0.3761,
+      "step": 9700
+    },
+    {
+      "epoch": 2.7713164466642883,
+      "grad_norm": 0.920684814453125,
+      "learning_rate": 4.923770054265191e-05,
+      "loss": 0.3585,
+      "step": 9710
+    },
+    {
+      "epoch": 2.774170531573314,
+      "grad_norm": 0.9885991811752319,
+      "learning_rate": 4.913806285606978e-05,
+      "loss": 0.355,
+      "step": 9720
+    },
+    {
+      "epoch": 2.7770246164823402,
+      "grad_norm": 0.9228259921073914,
+      "learning_rate": 4.903842859320138e-05,
+      "loss": 0.4018,
+      "step": 9730
+    },
+    {
+      "epoch": 2.7798787013913664,
+      "grad_norm": 0.97159743309021,
+      "learning_rate": 4.893879814980554e-05,
+      "loss": 0.3712,
+      "step": 9740
+    },
+    {
+      "epoch": 2.7827327863003926,
+      "grad_norm": 1.0281260013580322,
+      "learning_rate": 4.8839171921625825e-05,
+      "loss": 0.3581,
+      "step": 9750
+    },
+    {
+      "epoch": 2.7855868712094187,
+      "grad_norm": 1.0322412252426147,
+      "learning_rate": 4.873955030438912e-05,
+      "loss": 0.3793,
+      "step": 9760
+    },
+    {
+      "epoch": 2.7884409561184444,
+      "grad_norm": 1.138863444328308,
+      "learning_rate": 4.8639933693803994e-05,
+      "loss": 0.3549,
+      "step": 9770
+    },
+    {
+      "epoch": 2.7912950410274706,
+      "grad_norm": 1.1630406379699707,
+      "learning_rate": 4.8540322485559096e-05,
+      "loss": 0.3615,
+      "step": 9780
+    },
+    {
+      "epoch": 2.794149125936497,
+      "grad_norm": 0.7852999567985535,
+      "learning_rate": 4.8440717075321665e-05,
+      "loss": 0.3761,
+      "step": 9790
+    },
+    {
+      "epoch": 2.7970032108455225,
+      "grad_norm": 1.1240308284759521,
+      "learning_rate": 4.834111785873582e-05,
+      "loss": 0.3764,
+      "step": 9800
+    },
+    {
+      "epoch": 2.7998572957545487,
+      "grad_norm": 1.4110418558120728,
+      "learning_rate": 4.824152523142119e-05,
+      "loss": 0.3532,
+      "step": 9810
+    },
+    {
+      "epoch": 2.802711380663575,
+      "grad_norm": 1.2288501262664795,
+      "learning_rate": 4.8141939588971154e-05,
+      "loss": 0.3798,
+      "step": 9820
+    },
+    {
+      "epoch": 2.8055654655726006,
+      "grad_norm": 0.9042624235153198,
+      "learning_rate": 4.804236132695139e-05,
+      "loss": 0.3658,
+      "step": 9830
+    },
+    {
+      "epoch": 2.8084195504816267,
+      "grad_norm": 1.4888442754745483,
+      "learning_rate": 4.79427908408982e-05,
+      "loss": 0.3814,
+      "step": 9840
+    },
+    {
+      "epoch": 2.811273635390653,
+      "grad_norm": 0.7844000458717346,
+      "learning_rate": 4.7843228526317074e-05,
+      "loss": 0.3733,
+      "step": 9850
+    },
+    {
+      "epoch": 2.814127720299679,
+      "grad_norm": 1.0347716808319092,
+      "learning_rate": 4.7743674778680984e-05,
+      "loss": 0.3868,
+      "step": 9860
+    },
+    {
+      "epoch": 2.8169818052087052,
+      "grad_norm": 0.7579706311225891,
+      "learning_rate": 4.7644129993428944e-05,
+      "loss": 0.3729,
+      "step": 9870
+    },
+    {
+      "epoch": 2.819835890117731,
+      "grad_norm": 1.2491960525512695,
+      "learning_rate": 4.754459456596428e-05,
+      "loss": 0.4339,
+      "step": 9880
+    },
+    {
+      "epoch": 2.822689975026757,
+      "grad_norm": 0.8434522151947021,
+      "learning_rate": 4.744506889165318e-05,
+      "loss": 0.3851,
+      "step": 9890
+    },
+    {
+      "epoch": 2.8255440599357833,
+      "grad_norm": 0.8665650486946106,
+      "learning_rate": 4.734555336582312e-05,
+      "loss": 0.3667,
+      "step": 9900
+    },
+    {
+      "epoch": 2.828398144844809,
+      "grad_norm": 0.9557413458824158,
+      "learning_rate": 4.724604838376126e-05,
+      "loss": 0.3558,
+      "step": 9910
+    },
+    {
+      "epoch": 2.831252229753835,
+      "grad_norm": 1.1994484663009644,
+      "learning_rate": 4.714655434071287e-05,
+      "loss": 0.3667,
+      "step": 9920
+    },
+    {
+      "epoch": 2.8341063146628613,
+      "grad_norm": 0.9534554481506348,
+      "learning_rate": 4.704707163187974e-05,
+      "loss": 0.3491,
+      "step": 9930
+    },
+    {
+      "epoch": 2.836960399571887,
+      "grad_norm": 1.180798888206482,
+      "learning_rate": 4.6947600652418655e-05,
+      "loss": 0.3774,
+      "step": 9940
+    },
+    {
+      "epoch": 2.8398144844809132,
+      "grad_norm": 0.9711904525756836,
+      "learning_rate": 4.6848141797439834e-05,
+      "loss": 0.3651,
+      "step": 9950
+    },
+    {
+      "epoch": 2.8426685693899394,
+      "grad_norm": 1.1422544717788696,
+      "learning_rate": 4.674869546200535e-05,
+      "loss": 0.3946,
+      "step": 9960
+    },
+    {
+      "epoch": 2.8455226542989656,
+      "grad_norm": 1.1134331226348877,
+      "learning_rate": 4.6649262041127465e-05,
+      "loss": 0.3915,
+      "step": 9970
+    },
+    {
+      "epoch": 2.8483767392079913,
+      "grad_norm": 0.7959815859794617,
+      "learning_rate": 4.654984192976722e-05,
+      "loss": 0.3912,
+      "step": 9980
+    },
+    {
+      "epoch": 2.8512308241170174,
+      "grad_norm": 0.79036545753479,
+      "learning_rate": 4.645043552283276e-05,
+      "loss": 0.3957,
+      "step": 9990
+    },
+    {
+      "epoch": 2.8540849090260436,
+      "grad_norm": 1.0974929332733154,
+      "learning_rate": 4.635104321517782e-05,
+      "loss": 0.4073,
+      "step": 10000
+    },
+    {
+      "epoch": 2.85693899393507,
+      "grad_norm": 0.7789895534515381,
+      "learning_rate": 4.625166540160007e-05,
+      "loss": 0.3923,
+      "step": 10010
+    },
+    {
+      "epoch": 2.8597930788440955,
+      "grad_norm": 1.0822783708572388,
+      "learning_rate": 4.615230247683967e-05,
+      "loss": 0.3846,
+      "step": 10020
+    },
+    {
+      "epoch": 2.8626471637531217,
+      "grad_norm": 0.8098913431167603,
+      "learning_rate": 4.605295483557762e-05,
+      "loss": 0.3786,
+      "step": 10030
+    },
+    {
+      "epoch": 2.865501248662148,
+      "grad_norm": 0.8834393620491028,
+      "learning_rate": 4.59536228724342e-05,
+      "loss": 0.3595,
+      "step": 10040
+    },
+    {
+      "epoch": 2.8683553335711736,
+      "grad_norm": 1.1877049207687378,
+      "learning_rate": 4.5854306981967446e-05,
+      "loss": 0.3634,
+      "step": 10050
+    },
+    {
+      "epoch": 2.8712094184801997,
+      "grad_norm": 0.8630797266960144,
+      "learning_rate": 4.57550075586715e-05,
+      "loss": 0.3752,
+      "step": 10060
+    },
+    {
+      "epoch": 2.874063503389226,
+      "grad_norm": 0.8884981274604797,
+      "learning_rate": 4.565572499697513e-05,
+      "loss": 0.3759,
+      "step": 10070
+    },
+    {
+      "epoch": 2.8769175882982516,
+      "grad_norm": 1.0692553520202637,
+      "learning_rate": 4.555645969124015e-05,
+      "loss": 0.3636,
+      "step": 10080
+    },
+    {
+      "epoch": 2.879771673207278,
+      "grad_norm": 0.8800570964813232,
+      "learning_rate": 4.5457212035759775e-05,
+      "loss": 0.3749,
+      "step": 10090
+    },
+    {
+      "epoch": 2.882625758116304,
+      "grad_norm": 1.0624586343765259,
+      "learning_rate": 4.5357982424757146e-05,
+      "loss": 0.397,
+      "step": 10100
+    },
+    {
+      "epoch": 2.88547984302533,
+      "grad_norm": 1.155393362045288,
+      "learning_rate": 4.5258771252383714e-05,
+      "loss": 0.4057,
+      "step": 10110
+    },
+    {
+      "epoch": 2.8883339279343563,
+      "grad_norm": 1.4139622449874878,
+      "learning_rate": 4.5159578912717717e-05,
+      "loss": 0.3651,
+      "step": 10120
+    },
+    {
+      "epoch": 2.891188012843382,
+      "grad_norm": 1.1728215217590332,
+      "learning_rate": 4.506040579976257e-05,
+      "loss": 0.4041,
+      "step": 10130
+    },
+    {
+      "epoch": 2.894042097752408,
+      "grad_norm": 0.8503699898719788,
+      "learning_rate": 4.4961252307445294e-05,
+      "loss": 0.3583,
+      "step": 10140
+    },
+    {
+      "epoch": 2.8968961826614343,
+      "grad_norm": 1.1832642555236816,
+      "learning_rate": 4.486211882961502e-05,
+      "loss": 0.377,
+      "step": 10150
+    },
+    {
+      "epoch": 2.89975026757046,
+      "grad_norm": 1.144201636314392,
+      "learning_rate": 4.476300576004133e-05,
+      "loss": 0.3639,
+      "step": 10160
+    },
+    {
+      "epoch": 2.9026043524794862,
+      "grad_norm": 1.2758240699768066,
+      "learning_rate": 4.4663913492412785e-05,
+      "loss": 0.3883,
+      "step": 10170
+    },
+    {
+      "epoch": 2.9054584373885124,
+      "grad_norm": 1.354203462600708,
+      "learning_rate": 4.4564842420335315e-05,
+      "loss": 0.3932,
+      "step": 10180
+    },
+    {
+      "epoch": 2.908312522297538,
+      "grad_norm": 1.103074550628662,
+      "learning_rate": 4.446579293733061e-05,
+      "loss": 0.3934,
+      "step": 10190
+    },
+    {
+      "epoch": 2.9111666072065643,
+      "grad_norm": 1.5048420429229736,
+      "learning_rate": 4.436676543683467e-05,
+      "loss": 0.4066,
+      "step": 10200
+    },
+    {
+      "epoch": 2.9140206921155904,
+      "grad_norm": 0.9282875657081604,
+      "learning_rate": 4.426776031219613e-05,
+      "loss": 0.412,
+      "step": 10210
+    },
+    {
+      "epoch": 2.9168747770246166,
+      "grad_norm": 1.1087204217910767,
+      "learning_rate": 4.4168777956674796e-05,
+      "loss": 0.3809,
+      "step": 10220
+    },
+    {
+      "epoch": 2.919728861933643,
+      "grad_norm": 1.0893741846084595,
+      "learning_rate": 4.406981876343995e-05,
+      "loss": 0.3597,
+      "step": 10230
+    },
+    {
+      "epoch": 2.9225829468426685,
+      "grad_norm": 0.7606377005577087,
+      "learning_rate": 4.397088312556895e-05,
+      "loss": 0.3517,
+      "step": 10240
+    },
+    {
+      "epoch": 2.9254370317516947,
+      "grad_norm": 0.9937264919281006,
+      "learning_rate": 4.3871971436045565e-05,
+      "loss": 0.3784,
+      "step": 10250
+    },
+    {
+      "epoch": 2.928291116660721,
+      "grad_norm": 1.194831132888794,
+      "learning_rate": 4.377308408775843e-05,
+      "loss": 0.3924,
+      "step": 10260
+    },
+    {
+      "epoch": 2.9311452015697466,
+      "grad_norm": 0.8780274987220764,
+      "learning_rate": 4.36742214734995e-05,
+      "loss": 0.3572,
+      "step": 10270
+    },
+    {
+      "epoch": 2.9339992864787727,
+      "grad_norm": 0.943505585193634,
+      "learning_rate": 4.357538398596246e-05,
+      "loss": 0.3673,
+      "step": 10280
+    },
+    {
+      "epoch": 2.936853371387799,
+      "grad_norm": 0.8081468343734741,
+      "learning_rate": 4.347657201774123e-05,
+      "loss": 0.3551,
+      "step": 10290
+    },
+    {
+      "epoch": 2.9397074562968246,
+      "grad_norm": 0.843436062335968,
+      "learning_rate": 4.337778596132834e-05,
+      "loss": 0.3678,
+      "step": 10300
+    },
+    {
+      "epoch": 2.942561541205851,
+      "grad_norm": 0.8084463477134705,
+      "learning_rate": 4.3279026209113415e-05,
+      "loss": 0.3802,
+      "step": 10310
+    },
+    {
+      "epoch": 2.945415626114877,
+      "grad_norm": 0.9112711548805237,
+      "learning_rate": 4.3180293153381564e-05,
+      "loss": 0.3639,
+      "step": 10320
+    },
+    {
+      "epoch": 2.948269711023903,
+      "grad_norm": 1.0327959060668945,
+      "learning_rate": 4.308158718631186e-05,
+      "loss": 0.3545,
+      "step": 10330
+    },
+    {
+      "epoch": 2.951123795932929,
+      "grad_norm": 1.144468903541565,
+      "learning_rate": 4.29829086999758e-05,
+      "loss": 0.3848,
+      "step": 10340
+    },
+    {
+      "epoch": 2.953977880841955,
+      "grad_norm": 1.114212989807129,
+      "learning_rate": 4.288425808633575e-05,
+      "loss": 0.3352,
+      "step": 10350
+    },
+    {
+      "epoch": 2.956831965750981,
+      "grad_norm": 0.8838599324226379,
+      "learning_rate": 4.278563573724327e-05,
+      "loss": 0.3602,
+      "step": 10360
+    },
+    {
+      "epoch": 2.9596860506600073,
+      "grad_norm": 0.6728760600090027,
+      "learning_rate": 4.2687042044437735e-05,
+      "loss": 0.3774,
+      "step": 10370
+    },
+    {
+      "epoch": 2.962540135569033,
+      "grad_norm": 0.8785340189933777,
+      "learning_rate": 4.2588477399544646e-05,
+      "loss": 0.376,
+      "step": 10380
+    },
+    {
+      "epoch": 2.9653942204780592,
+      "grad_norm": 0.8221905827522278,
+      "learning_rate": 4.248994219407418e-05,
+      "loss": 0.3985,
+      "step": 10390
+    },
+    {
+      "epoch": 2.9682483053870854,
+      "grad_norm": 0.64462810754776,
+      "learning_rate": 4.2391436819419484e-05,
+      "loss": 0.3705,
+      "step": 10400
+    },
+    {
+      "epoch": 2.971102390296111,
+      "grad_norm": 1.144260048866272,
+      "learning_rate": 4.2292961666855293e-05,
+      "loss": 0.3775,
+      "step": 10410
+    },
+    {
+      "epoch": 2.9739564752051373,
+      "grad_norm": 0.7611860632896423,
+      "learning_rate": 4.2194517127536284e-05,
+      "loss": 0.3941,
+      "step": 10420
+    },
+    {
+      "epoch": 2.9768105601141635,
+      "grad_norm": 1.127455472946167,
+      "learning_rate": 4.209610359249549e-05,
+      "loss": 0.3842,
+      "step": 10430
+    },
+    {
+      "epoch": 2.979664645023189,
+      "grad_norm": 1.066175103187561,
+      "learning_rate": 4.199772145264287e-05,
+      "loss": 0.3654,
+      "step": 10440
+    },
+    {
+      "epoch": 2.9825187299322153,
+      "grad_norm": 1.0229511260986328,
+      "learning_rate": 4.189937109876358e-05,
+      "loss": 0.3797,
+      "step": 10450
+    },
+    {
+      "epoch": 2.9853728148412415,
+      "grad_norm": 1.1939435005187988,
+      "learning_rate": 4.180105292151659e-05,
+      "loss": 0.4035,
+      "step": 10460
+    },
+    {
+      "epoch": 2.9882268997502677,
+      "grad_norm": 1.0425211191177368,
+      "learning_rate": 4.170276731143306e-05,
+      "loss": 0.3845,
+      "step": 10470
+    },
+    {
+      "epoch": 2.991080984659294,
+      "grad_norm": 1.2975025177001953,
+      "learning_rate": 4.160451465891476e-05,
+      "loss": 0.4128,
+      "step": 10480
+    },
+    {
+      "epoch": 2.9939350695683196,
+      "grad_norm": 1.103302001953125,
+      "learning_rate": 4.150629535423252e-05,
+      "loss": 0.3899,
+      "step": 10490
+    },
+    {
+      "epoch": 2.9967891544773457,
+      "grad_norm": 0.9552080631256104,
+      "learning_rate": 4.140810978752478e-05,
+      "loss": 0.3792,
+      "step": 10500
+    },
+    {
+      "epoch": 2.999643239386372,
+      "grad_norm": 0.7346426844596863,
+      "learning_rate": 4.130995834879593e-05,
+      "loss": 0.3453,
+      "step": 10510
+    },
+    {
+      "epoch": 3.0024973242953976,
+      "grad_norm": 1.463599681854248,
+      "learning_rate": 4.121184142791482e-05,
+      "loss": 0.3457,
+      "step": 10520
+    },
+    {
+      "epoch": 3.005351409204424,
+      "grad_norm": 0.9227272272109985,
+      "learning_rate": 4.111375941461316e-05,
+      "loss": 0.3224,
+      "step": 10530
+    },
+    {
+      "epoch": 3.00820549411345,
+      "grad_norm": 1.2124933004379272,
+      "learning_rate": 4.101571269848401e-05,
+      "loss": 0.3045,
+      "step": 10540
+    },
+    {
+      "epoch": 3.011059579022476,
+      "grad_norm": 0.9984824657440186,
+      "learning_rate": 4.0917701668980244e-05,
+      "loss": 0.3294,
+      "step": 10550
+    },
+    {
+      "epoch": 3.013913663931502,
+      "grad_norm": 1.624333143234253,
+      "learning_rate": 4.081972671541298e-05,
+      "loss": 0.299,
+      "step": 10560
+    },
+    {
+      "epoch": 3.016767748840528,
+      "grad_norm": 1.5054949522018433,
+      "learning_rate": 4.072178822695005e-05,
+      "loss": 0.3535,
+      "step": 10570
+    },
+    {
+      "epoch": 3.019621833749554,
+      "grad_norm": 1.4161115884780884,
+      "learning_rate": 4.06238865926144e-05,
+      "loss": 0.3128,
+      "step": 10580
+    },
+    {
+      "epoch": 3.02247591865858,
+      "grad_norm": 0.9349290132522583,
+      "learning_rate": 4.052602220128261e-05,
+      "loss": 0.3096,
+      "step": 10590
+    },
+    {
+      "epoch": 3.025330003567606,
+      "grad_norm": 1.5315011739730835,
+      "learning_rate": 4.0428195441683345e-05,
+      "loss": 0.335,
+      "step": 10600
+    },
+    {
+      "epoch": 3.0281840884766322,
+      "grad_norm": 1.9867864847183228,
+      "learning_rate": 4.0330406702395796e-05,
+      "loss": 0.3268,
+      "step": 10610
+    },
+    {
+      "epoch": 3.0310381733856584,
+      "grad_norm": 1.2554751634597778,
+      "learning_rate": 4.023265637184806e-05,
+      "loss": 0.3149,
+      "step": 10620
+    },
+    {
+      "epoch": 3.033892258294684,
+      "grad_norm": 1.5094444751739502,
+      "learning_rate": 4.0134944838315767e-05,
+      "loss": 0.3482,
+      "step": 10630
+    },
+    {
+      "epoch": 3.0367463432037103,
+      "grad_norm": 1.3767508268356323,
+      "learning_rate": 4.003727248992039e-05,
+      "loss": 0.3439,
+      "step": 10640
+    },
+    {
+      "epoch": 3.0396004281127365,
+      "grad_norm": 1.811305046081543,
+      "learning_rate": 3.9939639714627765e-05,
+      "loss": 0.341,
+      "step": 10650
+    },
+    {
+      "epoch": 3.042454513021762,
+      "grad_norm": 1.2566548585891724,
+      "learning_rate": 3.984204690024651e-05,
+      "loss": 0.2984,
+      "step": 10660
+    },
+    {
+      "epoch": 3.0453085979307883,
+      "grad_norm": 1.4624361991882324,
+      "learning_rate": 3.9744494434426565e-05,
+      "loss": 0.3178,
+      "step": 10670
+    },
+    {
+      "epoch": 3.0481626828398145,
+      "grad_norm": 1.7333046197891235,
+      "learning_rate": 3.9646982704657584e-05,
+      "loss": 0.3347,
+      "step": 10680
+    },
+    {
+      "epoch": 3.0510167677488407,
+      "grad_norm": 1.8150200843811035,
+      "learning_rate": 3.954951209826737e-05,
+      "loss": 0.353,
+      "step": 10690
+    },
+    {
+      "epoch": 3.0538708526578664,
+      "grad_norm": 1.6588340997695923,
+      "learning_rate": 3.945208300242046e-05,
+      "loss": 0.3009,
+      "step": 10700
+    },
+    {
+      "epoch": 3.0567249375668926,
+      "grad_norm": 2.271573781967163,
+      "learning_rate": 3.9354695804116414e-05,
+      "loss": 0.315,
+      "step": 10710
+    },
+    {
+      "epoch": 3.0595790224759187,
+      "grad_norm": 2.114109516143799,
+      "learning_rate": 3.925735089018844e-05,
+      "loss": 0.3362,
+      "step": 10720
+    },
+    {
+      "epoch": 3.062433107384945,
+      "grad_norm": 1.960115671157837,
+      "learning_rate": 3.916004864730177e-05,
+      "loss": 0.3145,
+      "step": 10730
+    },
+    {
+      "epoch": 3.0652871922939706,
+      "grad_norm": 1.7500094175338745,
+      "learning_rate": 3.906278946195212e-05,
+      "loss": 0.3517,
+      "step": 10740
+    },
+    {
+      "epoch": 3.068141277202997,
+      "grad_norm": 2.0695393085479736,
+      "learning_rate": 3.89655737204642e-05,
+      "loss": 0.3374,
+      "step": 10750
+    },
+    {
+      "epoch": 3.070995362112023,
+      "grad_norm": 1.843173861503601,
+      "learning_rate": 3.886840180899013e-05,
+      "loss": 0.3207,
+      "step": 10760
+    },
+    {
+      "epoch": 3.0738494470210487,
+      "grad_norm": 2.033423662185669,
+      "learning_rate": 3.877127411350797e-05,
+      "loss": 0.3568,
+      "step": 10770
+    },
+    {
+      "epoch": 3.076703531930075,
+      "grad_norm": 1.9005411863327026,
+      "learning_rate": 3.867419101982014e-05,
+      "loss": 0.2956,
+      "step": 10780
+    },
+    {
+      "epoch": 3.079557616839101,
+      "grad_norm": 2.1244897842407227,
+      "learning_rate": 3.857715291355186e-05,
+      "loss": 0.3034,
+      "step": 10790
+    },
+    {
+      "epoch": 3.082411701748127,
+      "grad_norm": 2.2284629344940186,
+      "learning_rate": 3.8480160180149686e-05,
+      "loss": 0.3295,
+      "step": 10800
+    },
+    {
+      "epoch": 3.085265786657153,
+      "grad_norm": 2.1775922775268555,
+      "learning_rate": 3.838321320487994e-05,
+      "loss": 0.3,
+      "step": 10810
+    },
+    {
+      "epoch": 3.088119871566179,
+      "grad_norm": 1.6091368198394775,
+      "learning_rate": 3.8286312372827196e-05,
+      "loss": 0.3063,
+      "step": 10820
+    },
+    {
+      "epoch": 3.0909739564752052,
+      "grad_norm": 2.2415101528167725,
+      "learning_rate": 3.8189458068892757e-05,
+      "loss": 0.3292,
+      "step": 10830
+    },
+    {
+      "epoch": 3.0938280413842314,
+      "grad_norm": 1.6763960123062134,
+      "learning_rate": 3.809265067779304e-05,
+      "loss": 0.3093,
+      "step": 10840
+    },
+    {
+      "epoch": 3.096682126293257,
+      "grad_norm": 1.8144205808639526,
+      "learning_rate": 3.7995890584058205e-05,
+      "loss": 0.2954,
+      "step": 10850
+    },
+    {
+      "epoch": 3.0995362112022833,
+      "grad_norm": 3.186918020248413,
+      "learning_rate": 3.7899178172030485e-05,
+      "loss": 0.3172,
+      "step": 10860
+    },
+    {
+      "epoch": 3.1023902961113095,
+      "grad_norm": 1.9650551080703735,
+      "learning_rate": 3.7802513825862766e-05,
+      "loss": 0.3272,
+      "step": 10870
+    },
+    {
+      "epoch": 3.105244381020335,
+      "grad_norm": 2.2329864501953125,
+      "learning_rate": 3.7705897929516934e-05,
+      "loss": 0.2984,
+      "step": 10880
+    },
+    {
+      "epoch": 3.1080984659293613,
+      "grad_norm": 1.8501965999603271,
+      "learning_rate": 3.760933086676251e-05,
+      "loss": 0.3237,
+      "step": 10890
+    },
+    {
+      "epoch": 3.1109525508383875,
+      "grad_norm": 1.8563834428787231,
+      "learning_rate": 3.751281302117498e-05,
+      "loss": 0.3094,
+      "step": 10900
+    },
+    {
+      "epoch": 3.1138066357474137,
+      "grad_norm": 2.7098913192749023,
+      "learning_rate": 3.7416344776134396e-05,
+      "loss": 0.2995,
+      "step": 10910
+    },
+    {
+      "epoch": 3.1166607206564394,
+      "grad_norm": 2.422917127609253,
+      "learning_rate": 3.73199265148237e-05,
+      "loss": 0.3277,
+      "step": 10920
+    },
+    {
+      "epoch": 3.1195148055654656,
+      "grad_norm": 2.2790322303771973,
+      "learning_rate": 3.722355862022737e-05,
+      "loss": 0.3023,
+      "step": 10930
+    },
+    {
+      "epoch": 3.1223688904744917,
+      "grad_norm": 2.1287717819213867,
+      "learning_rate": 3.7127241475129814e-05,
+      "loss": 0.3018,
+      "step": 10940
+    },
+    {
+      "epoch": 3.1252229753835175,
+      "grad_norm": 1.9214931726455688,
+      "learning_rate": 3.703097546211384e-05,
+      "loss": 0.2942,
+      "step": 10950
+    },
+    {
+      "epoch": 3.1280770602925436,
+      "grad_norm": 1.8557084798812866,
+      "learning_rate": 3.693476096355914e-05,
+      "loss": 0.3244,
+      "step": 10960
+    },
+    {
+      "epoch": 3.13093114520157,
+      "grad_norm": 2.056940793991089,
+      "learning_rate": 3.6838598361640796e-05,
+      "loss": 0.3145,
+      "step": 10970
+    },
+    {
+      "epoch": 3.133785230110596,
+      "grad_norm": 2.423170566558838,
+      "learning_rate": 3.674248803832776e-05,
+      "loss": 0.3193,
+      "step": 10980
+    },
+    {
+      "epoch": 3.1366393150196217,
+      "grad_norm": 2.183138370513916,
+      "learning_rate": 3.664643037538132e-05,
+      "loss": 0.3031,
+      "step": 10990
+    },
+    {
+      "epoch": 3.139493399928648,
+      "grad_norm": 2.5854146480560303,
+      "learning_rate": 3.655042575435363e-05,
+      "loss": 0.2895,
+      "step": 11000
+    },
+    {
+      "epoch": 3.142347484837674,
+      "grad_norm": 2.0022974014282227,
+      "learning_rate": 3.645447455658606e-05,
+      "loss": 0.3058,
+      "step": 11010
+    },
+    {
+      "epoch": 3.1452015697466997,
+      "grad_norm": 2.34773588180542,
+      "learning_rate": 3.635857716320785e-05,
+      "loss": 0.3446,
+      "step": 11020
+    },
+    {
+      "epoch": 3.148055654655726,
+      "grad_norm": 2.164949417114258,
+      "learning_rate": 3.626273395513452e-05,
+      "loss": 0.3309,
+      "step": 11030
+    },
+    {
+      "epoch": 3.150909739564752,
+      "grad_norm": 2.035012722015381,
+      "learning_rate": 3.6166945313066356e-05,
+      "loss": 0.3344,
+      "step": 11040
+    },
+    {
+      "epoch": 3.1537638244737782,
+      "grad_norm": 1.8895570039749146,
+      "learning_rate": 3.6071211617486854e-05,
+      "loss": 0.2638,
+      "step": 11050
+    },
+    {
+      "epoch": 3.156617909382804,
+      "grad_norm": 2.0327749252319336,
+      "learning_rate": 3.597553324866133e-05,
+      "loss": 0.3092,
+      "step": 11060
+    },
+    {
+      "epoch": 3.15947199429183,
+      "grad_norm": 2.889387845993042,
+      "learning_rate": 3.587991058663529e-05,
+      "loss": 0.3272,
+      "step": 11070
+    },
+    {
+      "epoch": 3.1623260792008563,
+      "grad_norm": 2.0560340881347656,
+      "learning_rate": 3.578434401123296e-05,
+      "loss": 0.2967,
+      "step": 11080
+    },
+    {
+      "epoch": 3.1651801641098825,
+      "grad_norm": 2.2152063846588135,
+      "learning_rate": 3.568883390205585e-05,
+      "loss": 0.3319,
+      "step": 11090
+    },
+    {
+      "epoch": 3.168034249018908,
+      "grad_norm": 2.943876266479492,
+      "learning_rate": 3.559338063848106e-05,
+      "loss": 0.2784,
+      "step": 11100
+    },
+    {
+      "epoch": 3.1708883339279343,
+      "grad_norm": 1.6368544101715088,
+      "learning_rate": 3.549798459966002e-05,
+      "loss": 0.2907,
+      "step": 11110
+    },
+    {
+      "epoch": 3.1737424188369605,
+      "grad_norm": 2.72891902923584,
+      "learning_rate": 3.540264616451675e-05,
+      "loss": 0.3296,
+      "step": 11120
+    },
+    {
+      "epoch": 3.1765965037459862,
+      "grad_norm": 2.542816162109375,
+      "learning_rate": 3.5307365711746564e-05,
+      "loss": 0.2827,
+      "step": 11130
+    },
+    {
+      "epoch": 3.1794505886550124,
+      "grad_norm": 2.4247114658355713,
+      "learning_rate": 3.5212143619814366e-05,
+      "loss": 0.3002,
+      "step": 11140
+    },
+    {
+      "epoch": 3.1823046735640386,
+      "grad_norm": 2.8057472705841064,
+      "learning_rate": 3.5116980266953304e-05,
+      "loss": 0.3219,
+      "step": 11150
+    },
+    {
+      "epoch": 3.1851587584730647,
+      "grad_norm": 2.1139702796936035,
+      "learning_rate": 3.5021876031163194e-05,
+      "loss": 0.2866,
+      "step": 11160
+    },
+    {
+      "epoch": 3.1880128433820905,
+      "grad_norm": 2.5530476570129395,
+      "learning_rate": 3.492683129020903e-05,
+      "loss": 0.2849,
+      "step": 11170
+    },
+    {
+      "epoch": 3.1908669282911166,
+      "grad_norm": 1.8995988368988037,
+      "learning_rate": 3.4831846421619486e-05,
+      "loss": 0.3049,
+      "step": 11180
+    },
+    {
+      "epoch": 3.193721013200143,
+      "grad_norm": 3.222698450088501,
+      "learning_rate": 3.47369218026854e-05,
+      "loss": 0.3299,
+      "step": 11190
+    },
+    {
+      "epoch": 3.196575098109169,
+      "grad_norm": 1.9907785654067993,
+      "learning_rate": 3.464205781045831e-05,
+      "loss": 0.2956,
+      "step": 11200
+    },
+    {
+      "epoch": 3.1994291830181947,
+      "grad_norm": 2.1759512424468994,
+      "learning_rate": 3.454725482174897e-05,
+      "loss": 0.3172,
+      "step": 11210
+    },
+    {
+      "epoch": 3.202283267927221,
+      "grad_norm": 2.066030979156494,
+      "learning_rate": 3.445251321312574e-05,
+      "loss": 0.3302,
+      "step": 11220
+    },
+    {
+      "epoch": 3.205137352836247,
+      "grad_norm": 2.9806997776031494,
+      "learning_rate": 3.4357833360913246e-05,
+      "loss": 0.3138,
+      "step": 11230
+    },
+    {
+      "epoch": 3.2079914377452727,
+      "grad_norm": 2.3913772106170654,
+      "learning_rate": 3.4263215641190756e-05,
+      "loss": 0.3021,
+      "step": 11240
+    },
+    {
+      "epoch": 3.210845522654299,
+      "grad_norm": 1.8535680770874023,
+      "learning_rate": 3.416866042979078e-05,
+      "loss": 0.313,
+      "step": 11250
+    },
+    {
+      "epoch": 3.213699607563325,
+      "grad_norm": 2.4189822673797607,
+      "learning_rate": 3.4074168102297544e-05,
+      "loss": 0.3239,
+      "step": 11260
+    },
+    {
+      "epoch": 3.2165536924723512,
+      "grad_norm": 2.500195026397705,
+      "learning_rate": 3.3979739034045435e-05,
+      "loss": 0.3204,
+      "step": 11270
+    },
+    {
+      "epoch": 3.219407777381377,
+      "grad_norm": 2.023677349090576,
+      "learning_rate": 3.388537360011762e-05,
+      "loss": 0.3206,
+      "step": 11280
+    },
+    {
+      "epoch": 3.222261862290403,
+      "grad_norm": 2.198542833328247,
+      "learning_rate": 3.379107217534448e-05,
+      "loss": 0.3118,
+      "step": 11290
+    },
+    {
+      "epoch": 3.2251159471994293,
+      "grad_norm": 2.5365757942199707,
+      "learning_rate": 3.369683513430217e-05,
+      "loss": 0.332,
+      "step": 11300
+    },
+    {
+      "epoch": 3.2279700321084555,
+      "grad_norm": 1.946198582649231,
+      "learning_rate": 3.360266285131106e-05,
+      "loss": 0.3098,
+      "step": 11310
+    },
+    {
+      "epoch": 3.230824117017481,
+      "grad_norm": 1.6458300352096558,
+      "learning_rate": 3.350855570043433e-05,
+      "loss": 0.3331,
+      "step": 11320
+    },
+    {
+      "epoch": 3.2336782019265073,
+      "grad_norm": 2.673943042755127,
+      "learning_rate": 3.341451405547645e-05,
+      "loss": 0.3067,
+      "step": 11330
+    },
+    {
+      "epoch": 3.2365322868355335,
+      "grad_norm": 2.8760769367218018,
+      "learning_rate": 3.332053828998166e-05,
+      "loss": 0.3178,
+      "step": 11340
+    },
+    {
+      "epoch": 3.2393863717445592,
+      "grad_norm": 2.138685703277588,
+      "learning_rate": 3.322662877723257e-05,
+      "loss": 0.3086,
+      "step": 11350
+    },
+    {
+      "epoch": 3.2422404566535854,
+      "grad_norm": 2.8188040256500244,
+      "learning_rate": 3.313278589024855e-05,
+      "loss": 0.3044,
+      "step": 11360
+    },
+    {
+      "epoch": 3.2450945415626116,
+      "grad_norm": 1.6736277341842651,
+      "learning_rate": 3.30390100017844e-05,
+      "loss": 0.3213,
+      "step": 11370
+    },
+    {
+      "epoch": 3.2479486264716377,
+      "grad_norm": 2.6807808876037598,
+      "learning_rate": 3.2945301484328764e-05,
+      "loss": 0.3046,
+      "step": 11380
+    },
+    {
+      "epoch": 3.2508027113806635,
+      "grad_norm": 2.0274837017059326,
+      "learning_rate": 3.285166071010269e-05,
+      "loss": 0.3262,
+      "step": 11390
+    },
+    {
+      "epoch": 3.2536567962896896,
+      "grad_norm": 2.1135919094085693,
+      "learning_rate": 3.27580880510581e-05,
+      "loss": 0.3095,
+      "step": 11400
+    },
+    {
+      "epoch": 3.256510881198716,
+      "grad_norm": 2.0542471408843994,
+      "learning_rate": 3.266458387887641e-05,
+      "loss": 0.2779,
+      "step": 11410
+    },
+    {
+      "epoch": 3.2593649661077415,
+      "grad_norm": 2.2782020568847656,
+      "learning_rate": 3.257114856496698e-05,
+      "loss": 0.3028,
+      "step": 11420
+    },
+    {
+      "epoch": 3.2622190510167677,
+      "grad_norm": 1.8134151697158813,
+      "learning_rate": 3.247778248046567e-05,
+      "loss": 0.3195,
+      "step": 11430
+    },
+    {
+      "epoch": 3.265073135925794,
+      "grad_norm": 1.9497913122177124,
+      "learning_rate": 3.23844859962333e-05,
+      "loss": 0.2959,
+      "step": 11440
+    },
+    {
+      "epoch": 3.26792722083482,
+      "grad_norm": 3.1539499759674072,
+      "learning_rate": 3.229125948285428e-05,
+      "loss": 0.3114,
+      "step": 11450
+    },
+    {
+      "epoch": 3.2707813057438457,
+      "grad_norm": 1.8426719903945923,
+      "learning_rate": 3.2198103310635063e-05,
+      "loss": 0.2976,
+      "step": 11460
+    },
+    {
+      "epoch": 3.273635390652872,
+      "grad_norm": 1.764742374420166,
+      "learning_rate": 3.210501784960273e-05,
+      "loss": 0.2567,
+      "step": 11470
+    },
+    {
+      "epoch": 3.276489475561898,
+      "grad_norm": 1.9132665395736694,
+      "learning_rate": 3.2012003469503484e-05,
+      "loss": 0.2783,
+      "step": 11480
+    },
+    {
+      "epoch": 3.279343560470924,
+      "grad_norm": 2.686553955078125,
+      "learning_rate": 3.191906053980113e-05,
+      "loss": 0.2941,
+      "step": 11490
+    },
+    {
+      "epoch": 3.28219764537995,
+      "grad_norm": 2.312584638595581,
+      "learning_rate": 3.182618942967571e-05,
+      "loss": 0.2951,
+      "step": 11500
+    },
+    {
+      "epoch": 3.285051730288976,
+      "grad_norm": 2.8648900985717773,
+      "learning_rate": 3.1733390508022e-05,
+      "loss": 0.2977,
+      "step": 11510
+    },
+    {
+      "epoch": 3.2879058151980023,
+      "grad_norm": 2.399113416671753,
+      "learning_rate": 3.1640664143448037e-05,
+      "loss": 0.3222,
+      "step": 11520
+    },
+    {
+      "epoch": 3.290759900107028,
+      "grad_norm": 1.9362565279006958,
+      "learning_rate": 3.154801070427359e-05,
+      "loss": 0.3305,
+      "step": 11530
+    },
+    {
+      "epoch": 3.293613985016054,
+      "grad_norm": 2.5253491401672363,
+      "learning_rate": 3.145543055852884e-05,
+      "loss": 0.307,
+      "step": 11540
+    },
+    {
+      "epoch": 3.2964680699250803,
+      "grad_norm": 1.9359861612319946,
+      "learning_rate": 3.136292407395279e-05,
+      "loss": 0.3074,
+      "step": 11550
+    },
+    {
+      "epoch": 3.2993221548341065,
+      "grad_norm": 2.321005344390869,
+      "learning_rate": 3.127049161799189e-05,
+      "loss": 0.3112,
+      "step": 11560
+    },
+    {
+      "epoch": 3.3021762397431322,
+      "grad_norm": 2.295661687850952,
+      "learning_rate": 3.1178133557798486e-05,
+      "loss": 0.2751,
+      "step": 11570
+    },
+    {
+      "epoch": 3.3050303246521584,
+      "grad_norm": 2.0653467178344727,
+      "learning_rate": 3.108585026022944e-05,
+      "loss": 0.3113,
+      "step": 11580
+    },
+    {
+      "epoch": 3.3078844095611846,
+      "grad_norm": 2.26777720451355,
+      "learning_rate": 3.099364209184469e-05,
+      "loss": 0.2819,
+      "step": 11590
+    },
+    {
+      "epoch": 3.3107384944702103,
+      "grad_norm": 2.4750428199768066,
+      "learning_rate": 3.090150941890569e-05,
+      "loss": 0.2958,
+      "step": 11600
+    },
+    {
+      "epoch": 3.3135925793792365,
+      "grad_norm": 2.1771957874298096,
+      "learning_rate": 3.080945260737407e-05,
+      "loss": 0.3228,
+      "step": 11610
+    },
+    {
+      "epoch": 3.3164466642882626,
+      "grad_norm": 2.8841826915740967,
+      "learning_rate": 3.0717472022910055e-05,
+      "loss": 0.3001,
+      "step": 11620
+    },
+    {
+      "epoch": 3.319300749197289,
+      "grad_norm": 2.6921682357788086,
+      "learning_rate": 3.062556803087116e-05,
+      "loss": 0.2934,
+      "step": 11630
+    },
+    {
+      "epoch": 3.3221548341063145,
+      "grad_norm": 3.178048610687256,
+      "learning_rate": 3.053374099631066e-05,
+      "loss": 0.3086,
+      "step": 11640
+    },
+    {
+      "epoch": 3.3250089190153407,
+      "grad_norm": 2.0060882568359375,
+      "learning_rate": 3.0441991283976106e-05,
+      "loss": 0.2972,
+      "step": 11650
+    },
+    {
+      "epoch": 3.327863003924367,
+      "grad_norm": 2.5764193534851074,
+      "learning_rate": 3.0350319258307947e-05,
+      "loss": 0.3102,
+      "step": 11660
+    },
+    {
+      "epoch": 3.330717088833393,
+      "grad_norm": 1.9401841163635254,
+      "learning_rate": 3.0258725283438028e-05,
+      "loss": 0.2917,
+      "step": 11670
+    },
+    {
+      "epoch": 3.3335711737424187,
+      "grad_norm": 1.859510064125061,
+      "learning_rate": 3.0167209723188195e-05,
+      "loss": 0.2994,
+      "step": 11680
+    },
+    {
+      "epoch": 3.336425258651445,
+      "grad_norm": 2.2214395999908447,
+      "learning_rate": 3.0075772941068818e-05,
+      "loss": 0.3188,
+      "step": 11690
+    },
+    {
+      "epoch": 3.339279343560471,
+      "grad_norm": 2.056652784347534,
+      "learning_rate": 2.9984415300277315e-05,
+      "loss": 0.2965,
+      "step": 11700
+    },
+    {
+      "epoch": 3.342133428469497,
+      "grad_norm": 1.6663153171539307,
+      "learning_rate": 2.9893137163696793e-05,
+      "loss": 0.3096,
+      "step": 11710
+    },
+    {
+      "epoch": 3.344987513378523,
+      "grad_norm": 2.426206111907959,
+      "learning_rate": 2.980193889389452e-05,
+      "loss": 0.2987,
+      "step": 11720
+    },
+    {
+      "epoch": 3.347841598287549,
+      "grad_norm": 3.1257927417755127,
+      "learning_rate": 2.9710820853120558e-05,
+      "loss": 0.3251,
+      "step": 11730
+    },
+    {
+      "epoch": 3.3506956831965753,
+      "grad_norm": 2.4305126667022705,
+      "learning_rate": 2.9619783403306285e-05,
+      "loss": 0.3115,
+      "step": 11740
+    },
+    {
+      "epoch": 3.353549768105601,
+      "grad_norm": 2.008829355239868,
+      "learning_rate": 2.9528826906062916e-05,
+      "loss": 0.2892,
+      "step": 11750
+    },
+    {
+      "epoch": 3.356403853014627,
+      "grad_norm": 2.515057325363159,
+      "learning_rate": 2.943795172268018e-05,
+      "loss": 0.334,
+      "step": 11760
+    },
+    {
+      "epoch": 3.3592579379236533,
+      "grad_norm": 3.7486140727996826,
+      "learning_rate": 2.934715821412477e-05,
+      "loss": 0.3152,
+      "step": 11770
+    },
+    {
+      "epoch": 3.3621120228326795,
+      "grad_norm": 2.83160662651062,
+      "learning_rate": 2.9256446741038994e-05,
+      "loss": 0.3077,
+      "step": 11780
+    },
+    {
+      "epoch": 3.3649661077417052,
+      "grad_norm": 3.645878791809082,
+      "learning_rate": 2.9165817663739264e-05,
+      "loss": 0.2973,
+      "step": 11790
+    },
+    {
+      "epoch": 3.3678201926507314,
+      "grad_norm": 2.012726306915283,
+      "learning_rate": 2.9075271342214728e-05,
+      "loss": 0.2954,
+      "step": 11800
+    },
+    {
+      "epoch": 3.3706742775597576,
+      "grad_norm": 1.7031292915344238,
+      "learning_rate": 2.8984808136125817e-05,
+      "loss": 0.2891,
+      "step": 11810
+    },
+    {
+      "epoch": 3.3735283624687833,
+      "grad_norm": 1.867417573928833,
+      "learning_rate": 2.8894428404802855e-05,
+      "loss": 0.2935,
+      "step": 11820
+    },
+    {
+      "epoch": 3.3763824473778095,
+      "grad_norm": 2.5113019943237305,
+      "learning_rate": 2.8804132507244507e-05,
+      "loss": 0.3482,
+      "step": 11830
+    },
+    {
+      "epoch": 3.3792365322868356,
+      "grad_norm": 2.145963430404663,
+      "learning_rate": 2.8713920802116522e-05,
+      "loss": 0.3005,
+      "step": 11840
+    },
+    {
+      "epoch": 3.3820906171958613,
+      "grad_norm": 2.3954710960388184,
+      "learning_rate": 2.8623793647750197e-05,
+      "loss": 0.2868,
+      "step": 11850
+    },
+    {
+      "epoch": 3.3849447021048875,
+      "grad_norm": 2.1182644367218018,
+      "learning_rate": 2.8533751402140974e-05,
+      "loss": 0.2984,
+      "step": 11860
+    },
+    {
+      "epoch": 3.3877987870139137,
+      "grad_norm": 3.194291830062866,
+      "learning_rate": 2.8443794422947068e-05,
+      "loss": 0.3183,
+      "step": 11870
+    },
+    {
+      "epoch": 3.39065287192294,
+      "grad_norm": 1.8101868629455566,
+      "learning_rate": 2.8353923067487932e-05,
+      "loss": 0.312,
+      "step": 11880
+    },
+    {
+      "epoch": 3.3935069568319656,
+      "grad_norm": 1.9580895900726318,
+      "learning_rate": 2.8264137692742986e-05,
+      "loss": 0.34,
+      "step": 11890
+    },
+    {
+      "epoch": 3.3963610417409917,
+      "grad_norm": 2.1564316749572754,
+      "learning_rate": 2.817443865535008e-05,
+      "loss": 0.3317,
+      "step": 11900
+    },
+    {
+      "epoch": 3.399215126650018,
+      "grad_norm": 2.4529192447662354,
+      "learning_rate": 2.808482631160414e-05,
+      "loss": 0.2773,
+      "step": 11910
+    },
+    {
+      "epoch": 3.402069211559044,
+      "grad_norm": 1.9759066104888916,
+      "learning_rate": 2.799530101745572e-05,
+      "loss": 0.303,
+      "step": 11920
+    },
+    {
+      "epoch": 3.40492329646807,
+      "grad_norm": 2.720956802368164,
+      "learning_rate": 2.790586312850963e-05,
+      "loss": 0.2848,
+      "step": 11930
+    },
+    {
+      "epoch": 3.407777381377096,
+      "grad_norm": 2.919189453125,
+      "learning_rate": 2.781651300002347e-05,
+      "loss": 0.3052,
+      "step": 11940
+    },
+    {
+      "epoch": 3.410631466286122,
+      "grad_norm": 4.829803466796875,
+      "learning_rate": 2.7727250986906284e-05,
+      "loss": 0.3654,
+      "step": 11950
+    },
+    {
+      "epoch": 3.413485551195148,
+      "grad_norm": 1.6773247718811035,
+      "learning_rate": 2.7638077443717036e-05,
+      "loss": 0.3399,
+      "step": 11960
+    },
+    {
+      "epoch": 3.416339636104174,
+      "grad_norm": 2.4335241317749023,
+      "learning_rate": 2.7548992724663346e-05,
+      "loss": 0.3144,
+      "step": 11970
+    },
+    {
+      "epoch": 3.4191937210132,
+      "grad_norm": 2.437046766281128,
+      "learning_rate": 2.7459997183600005e-05,
+      "loss": 0.3266,
+      "step": 11980
+    },
+    {
+      "epoch": 3.4220478059222263,
+      "grad_norm": 1.9997532367706299,
+      "learning_rate": 2.7371091174027576e-05,
+      "loss": 0.2772,
+      "step": 11990
+    },
+    {
+      "epoch": 3.424901890831252,
+      "grad_norm": 3.084336519241333,
+      "learning_rate": 2.7282275049090956e-05,
+      "loss": 0.3475,
+      "step": 12000
+    },
+    {
+      "epoch": 3.4277559757402782,
+      "grad_norm": 1.6730353832244873,
+      "learning_rate": 2.719354916157803e-05,
+      "loss": 0.3183,
+      "step": 12010
+    },
+    {
+      "epoch": 3.4306100606493044,
+      "grad_norm": 2.198012113571167,
+      "learning_rate": 2.7104913863918275e-05,
+      "loss": 0.3118,
+      "step": 12020
+    },
+    {
+      "epoch": 3.4334641455583306,
+      "grad_norm": 1.9634590148925781,
+      "learning_rate": 2.701636950818131e-05,
+      "loss": 0.3381,
+      "step": 12030
+    },
+    {
+      "epoch": 3.4363182304673563,
+      "grad_norm": 2.6610710620880127,
+      "learning_rate": 2.692791644607555e-05,
+      "loss": 0.3357,
+      "step": 12040
+    },
+    {
+      "epoch": 3.4391723153763825,
+      "grad_norm": 2.748284339904785,
+      "learning_rate": 2.6839555028946727e-05,
+      "loss": 0.3098,
+      "step": 12050
+    },
+    {
+      "epoch": 3.4420264002854086,
+      "grad_norm": 2.4703245162963867,
+      "learning_rate": 2.6751285607776594e-05,
+      "loss": 0.3273,
+      "step": 12060
+    },
+    {
+      "epoch": 3.4448804851944343,
+      "grad_norm": 2.2269365787506104,
+      "learning_rate": 2.6663108533181474e-05,
+      "loss": 0.2881,
+      "step": 12070
+    },
+    {
+      "epoch": 3.4477345701034605,
+      "grad_norm": 2.5148026943206787,
+      "learning_rate": 2.65750241554109e-05,
+      "loss": 0.3288,
+      "step": 12080
+    },
+    {
+      "epoch": 3.4505886550124867,
+      "grad_norm": 3.1650230884552,
+      "learning_rate": 2.6487032824346135e-05,
+      "loss": 0.3113,
+      "step": 12090
+    },
+    {
+      "epoch": 3.453442739921513,
+      "grad_norm": 2.1970267295837402,
+      "learning_rate": 2.6399134889498933e-05,
+      "loss": 0.3235,
+      "step": 12100
+    },
+    {
+      "epoch": 3.4562968248305386,
+      "grad_norm": 3.0943925380706787,
+      "learning_rate": 2.6311330700010002e-05,
+      "loss": 0.3188,
+      "step": 12110
+    },
+    {
+      "epoch": 3.4591509097395647,
+      "grad_norm": 2.365020751953125,
+      "learning_rate": 2.6223620604647716e-05,
+      "loss": 0.3049,
+      "step": 12120
+    },
+    {
+      "epoch": 3.462004994648591,
+      "grad_norm": 2.4323387145996094,
+      "learning_rate": 2.613600495180672e-05,
+      "loss": 0.3242,
+      "step": 12130
+    },
+    {
+      "epoch": 3.464859079557617,
+      "grad_norm": 2.0959761142730713,
+      "learning_rate": 2.6048484089506442e-05,
+      "loss": 0.2706,
+      "step": 12140
+    },
+    {
+      "epoch": 3.467713164466643,
+      "grad_norm": 2.6008434295654297,
+      "learning_rate": 2.5961058365389874e-05,
+      "loss": 0.2771,
+      "step": 12150
+    },
+    {
+      "epoch": 3.470567249375669,
+      "grad_norm": 2.3737094402313232,
+      "learning_rate": 2.5873728126722064e-05,
+      "loss": 0.299,
+      "step": 12160
+    },
+    {
+      "epoch": 3.473421334284695,
+      "grad_norm": 2.3306820392608643,
+      "learning_rate": 2.5786493720388815e-05,
+      "loss": 0.3158,
+      "step": 12170
+    },
+    {
+      "epoch": 3.476275419193721,
+      "grad_norm": 2.8044018745422363,
+      "learning_rate": 2.569935549289522e-05,
+      "loss": 0.2881,
+      "step": 12180
+    },
+    {
+      "epoch": 3.479129504102747,
+      "grad_norm": 3.1712582111358643,
+      "learning_rate": 2.5612313790364377e-05,
+      "loss": 0.2862,
+      "step": 12190
+    },
+    {
+      "epoch": 3.481983589011773,
+      "grad_norm": 2.3961644172668457,
+      "learning_rate": 2.5525368958536e-05,
+      "loss": 0.316,
+      "step": 12200
+    },
+    {
+      "epoch": 3.484837673920799,
+      "grad_norm": 1.7763845920562744,
+      "learning_rate": 2.543852134276493e-05,
+      "loss": 0.3134,
+      "step": 12210
+    },
+    {
+      "epoch": 3.487691758829825,
+      "grad_norm": 1.8748339414596558,
+      "learning_rate": 2.535177128801997e-05,
+      "loss": 0.3043,
+      "step": 12220
+    },
+    {
+      "epoch": 3.4905458437388512,
+      "grad_norm": 1.7089003324508667,
+      "learning_rate": 2.52651191388823e-05,
+      "loss": 0.2531,
+      "step": 12230
+    },
+    {
+      "epoch": 3.4933999286478774,
+      "grad_norm": 2.649825096130371,
+      "learning_rate": 2.5178565239544266e-05,
+      "loss": 0.2945,
+      "step": 12240
+    },
+    {
+      "epoch": 3.496254013556903,
+      "grad_norm": 1.940625548362732,
+      "learning_rate": 2.509210993380794e-05,
+      "loss": 0.2861,
+      "step": 12250
+    },
+    {
+      "epoch": 3.4991080984659293,
+      "grad_norm": 2.8728575706481934,
+      "learning_rate": 2.5005753565083796e-05,
+      "loss": 0.2772,
+      "step": 12260
+    },
+    {
+      "epoch": 3.5019621833749555,
+      "grad_norm": 2.242058038711548,
+      "learning_rate": 2.4919496476389236e-05,
+      "loss": 0.3029,
+      "step": 12270
+    },
+    {
+      "epoch": 3.5048162682839816,
+      "grad_norm": 2.4839067459106445,
+      "learning_rate": 2.48333390103474e-05,
+      "loss": 0.324,
+      "step": 12280
+    },
+    {
+      "epoch": 3.5076703531930074,
+      "grad_norm": 2.9785611629486084,
+      "learning_rate": 2.474728150918567e-05,
+      "loss": 0.3117,
+      "step": 12290
+    },
+    {
+      "epoch": 3.5105244381020335,
+      "grad_norm": 2.490342617034912,
+      "learning_rate": 2.466132431473439e-05,
+      "loss": 0.3132,
+      "step": 12300
+    },
+    {
+      "epoch": 3.5133785230110597,
+      "grad_norm": 2.2015674114227295,
+      "learning_rate": 2.4575467768425415e-05,
+      "loss": 0.3106,
+      "step": 12310
+    },
+    {
+      "epoch": 3.5162326079200854,
+      "grad_norm": 2.942949056625366,
+      "learning_rate": 2.4489712211290878e-05,
+      "loss": 0.305,
+      "step": 12320
+    },
+    {
+      "epoch": 3.5190866928291116,
+      "grad_norm": 2.5716145038604736,
+      "learning_rate": 2.440405798396171e-05,
+      "loss": 0.2831,
+      "step": 12330
+    },
+    {
+      "epoch": 3.5219407777381377,
+      "grad_norm": 2.0549802780151367,
+      "learning_rate": 2.431850542666642e-05,
+      "loss": 0.3033,
+      "step": 12340
+    },
+    {
+      "epoch": 3.524794862647164,
+      "grad_norm": 2.684563636779785,
+      "learning_rate": 2.42330548792296e-05,
+      "loss": 0.2892,
+      "step": 12350
+    },
+    {
+      "epoch": 3.52764894755619,
+      "grad_norm": 2.8911728858947754,
+      "learning_rate": 2.414770668107069e-05,
+      "loss": 0.3056,
+      "step": 12360
+    },
+    {
+      "epoch": 3.530503032465216,
+      "grad_norm": 2.5336625576019287,
+      "learning_rate": 2.4062461171202576e-05,
+      "loss": 0.3225,
+      "step": 12370
+    },
+    {
+      "epoch": 3.533357117374242,
+      "grad_norm": 3.059535503387451,
+      "learning_rate": 2.3977318688230266e-05,
+      "loss": 0.2905,
+      "step": 12380
+    },
+    {
+      "epoch": 3.536211202283268,
+      "grad_norm": 1.9596600532531738,
+      "learning_rate": 2.3892279570349534e-05,
+      "loss": 0.3125,
+      "step": 12390
+    },
+    {
+      "epoch": 3.539065287192294,
+      "grad_norm": 2.3087189197540283,
+      "learning_rate": 2.380734415534553e-05,
+      "loss": 0.3131,
+      "step": 12400
+    },
+    {
+      "epoch": 3.54191937210132,
+      "grad_norm": 3.6303563117980957,
+      "learning_rate": 2.372251278059153e-05,
+      "loss": 0.303,
+      "step": 12410
+    },
+    {
+      "epoch": 3.544773457010346,
+      "grad_norm": 2.220292568206787,
+      "learning_rate": 2.363778578304757e-05,
+      "loss": 0.3129,
+      "step": 12420
+    },
+    {
+      "epoch": 3.547627541919372,
+      "grad_norm": 2.5933589935302734,
+      "learning_rate": 2.355316349925902e-05,
+      "loss": 0.3324,
+      "step": 12430
+    },
+    {
+      "epoch": 3.550481626828398,
+      "grad_norm": 2.0918591022491455,
+      "learning_rate": 2.3468646265355348e-05,
+      "loss": 0.3376,
+      "step": 12440
+    },
+    {
+      "epoch": 3.5533357117374242,
+      "grad_norm": 2.7997231483459473,
+      "learning_rate": 2.3384234417048766e-05,
+      "loss": 0.3197,
+      "step": 12450
+    },
+    {
+      "epoch": 3.55618979664645,
+      "grad_norm": 3.100090265274048,
+      "learning_rate": 2.3299928289632878e-05,
+      "loss": 0.3183,
+      "step": 12460
+    },
+    {
+      "epoch": 3.559043881555476,
+      "grad_norm": 2.517780303955078,
+      "learning_rate": 2.3215728217981363e-05,
+      "loss": 0.3308,
+      "step": 12470
+    },
+    {
+      "epoch": 3.5618979664645023,
+      "grad_norm": 2.264176845550537,
+      "learning_rate": 2.313163453654657e-05,
+      "loss": 0.2907,
+      "step": 12480
+    },
+    {
+      "epoch": 3.5647520513735285,
+      "grad_norm": 3.2383158206939697,
+      "learning_rate": 2.3047647579358333e-05,
+      "loss": 0.3332,
+      "step": 12490
+    },
+    {
+      "epoch": 3.5676061362825546,
+      "grad_norm": 3.331801176071167,
+      "learning_rate": 2.2963767680022524e-05,
+      "loss": 0.3318,
+      "step": 12500
+    },
+    {
+      "epoch": 3.5704602211915804,
+      "grad_norm": 2.4765217304229736,
+      "learning_rate": 2.2879995171719783e-05,
+      "loss": 0.3036,
+      "step": 12510
+    },
+    {
+      "epoch": 3.5733143061006065,
+      "grad_norm": 3.2066502571105957,
+      "learning_rate": 2.2796330387204184e-05,
+      "loss": 0.2884,
+      "step": 12520
+    },
+    {
+      "epoch": 3.5761683910096327,
+      "grad_norm": 2.805443286895752,
+      "learning_rate": 2.2712773658801894e-05,
+      "loss": 0.3057,
+      "step": 12530
+    },
+    {
+      "epoch": 3.5790224759186584,
+      "grad_norm": 2.041646718978882,
+      "learning_rate": 2.262932531840985e-05,
+      "loss": 0.3151,
+      "step": 12540
+    },
+    {
+      "epoch": 3.5818765608276846,
+      "grad_norm": 2.6316616535186768,
+      "learning_rate": 2.2545985697494508e-05,
+      "loss": 0.3205,
+      "step": 12550
+    },
+    {
+      "epoch": 3.5847306457367107,
+      "grad_norm": 1.8530291318893433,
+      "learning_rate": 2.2462755127090468e-05,
+      "loss": 0.2762,
+      "step": 12560
+    },
+    {
+      "epoch": 3.5875847306457365,
+      "grad_norm": 2.0995118618011475,
+      "learning_rate": 2.2379633937799123e-05,
+      "loss": 0.3052,
+      "step": 12570
+    },
+    {
+      "epoch": 3.5904388155547626,
+      "grad_norm": 3.2132577896118164,
+      "learning_rate": 2.2296622459787437e-05,
+      "loss": 0.3253,
+      "step": 12580
+    },
+    {
+      "epoch": 3.593292900463789,
+      "grad_norm": 2.3871312141418457,
+      "learning_rate": 2.2213721022786577e-05,
+      "loss": 0.314,
+      "step": 12590
+    },
+    {
+      "epoch": 3.596146985372815,
+      "grad_norm": 2.5007219314575195,
+      "learning_rate": 2.2130929956090636e-05,
+      "loss": 0.304,
+      "step": 12600
+    },
+    {
+      "epoch": 3.599001070281841,
+      "grad_norm": 2.291391372680664,
+      "learning_rate": 2.204824958855523e-05,
+      "loss": 0.2768,
+      "step": 12610
+    },
+    {
+      "epoch": 3.601855155190867,
+      "grad_norm": 3.370358467102051,
+      "learning_rate": 2.1965680248596348e-05,
+      "loss": 0.3409,
+      "step": 12620
+    },
+    {
+      "epoch": 3.604709240099893,
+      "grad_norm": 2.82300066947937,
+      "learning_rate": 2.188322226418893e-05,
+      "loss": 0.3146,
+      "step": 12630
+    },
+    {
+      "epoch": 3.607563325008919,
+      "grad_norm": 1.9803205728530884,
+      "learning_rate": 2.180087596286561e-05,
+      "loss": 0.2822,
+      "step": 12640
+    },
+    {
+      "epoch": 3.610417409917945,
+      "grad_norm": 2.2831294536590576,
+      "learning_rate": 2.1718641671715377e-05,
+      "loss": 0.2818,
+      "step": 12650
+    },
+    {
+      "epoch": 3.613271494826971,
+      "grad_norm": 3.3273935317993164,
+      "learning_rate": 2.1636519717382307e-05,
+      "loss": 0.3059,
+      "step": 12660
+    },
+    {
+      "epoch": 3.6161255797359972,
+      "grad_norm": 2.60036301612854,
+      "learning_rate": 2.1554510426064294e-05,
+      "loss": 0.3313,
+      "step": 12670
+    },
+    {
+      "epoch": 3.618979664645023,
+      "grad_norm": 2.8558053970336914,
+      "learning_rate": 2.14726141235117e-05,
+      "loss": 0.2998,
+      "step": 12680
+    },
+    {
+      "epoch": 3.621833749554049,
+      "grad_norm": 3.08219313621521,
+      "learning_rate": 2.13908311350261e-05,
+      "loss": 0.2666,
+      "step": 12690
+    },
+    {
+      "epoch": 3.6246878344630753,
+      "grad_norm": 2.550483465194702,
+      "learning_rate": 2.1309161785458926e-05,
+      "loss": 0.3001,
+      "step": 12700
+    },
+    {
+      "epoch": 3.6275419193721015,
+      "grad_norm": 2.2145779132843018,
+      "learning_rate": 2.1227606399210263e-05,
+      "loss": 0.2923,
+      "step": 12710
+    },
+    {
+      "epoch": 3.6303960042811276,
+      "grad_norm": 3.1179816722869873,
+      "learning_rate": 2.1146165300227516e-05,
+      "loss": 0.2903,
+      "step": 12720
+    },
+    {
+      "epoch": 3.6332500891901534,
+      "grad_norm": 3.5646145343780518,
+      "learning_rate": 2.1064838812004146e-05,
+      "loss": 0.3002,
+      "step": 12730
+    },
+    {
+      "epoch": 3.6361041740991795,
+      "grad_norm": 2.551764726638794,
+      "learning_rate": 2.0983627257578302e-05,
+      "loss": 0.29,
+      "step": 12740
+    },
+    {
+      "epoch": 3.6389582590082057,
+      "grad_norm": 3.1751394271850586,
+      "learning_rate": 2.0902530959531686e-05,
+      "loss": 0.2964,
+      "step": 12750
+    },
+    {
+      "epoch": 3.6418123439172314,
+      "grad_norm": 1.7176988124847412,
+      "learning_rate": 2.0821550239988107e-05,
+      "loss": 0.3258,
+      "step": 12760
+    },
+    {
+      "epoch": 3.6446664288262576,
+      "grad_norm": 2.1163158416748047,
+      "learning_rate": 2.0740685420612345e-05,
+      "loss": 0.3198,
+      "step": 12770
+    },
+    {
+      "epoch": 3.6475205137352837,
+      "grad_norm": 2.264244556427002,
+      "learning_rate": 2.0659936822608807e-05,
+      "loss": 0.3184,
+      "step": 12780
+    },
+    {
+      "epoch": 3.6503745986443095,
+      "grad_norm": 2.034036636352539,
+      "learning_rate": 2.0579304766720192e-05,
+      "loss": 0.3204,
+      "step": 12790
+    },
+    {
+      "epoch": 3.6532286835533356,
+      "grad_norm": 2.2870633602142334,
+      "learning_rate": 2.0498789573226356e-05,
+      "loss": 0.3102,
+      "step": 12800
+    },
+    {
+      "epoch": 3.656082768462362,
+      "grad_norm": 2.5353150367736816,
+      "learning_rate": 2.0418391561942934e-05,
+      "loss": 0.3145,
+      "step": 12810
+    },
+    {
+      "epoch": 3.658936853371388,
+      "grad_norm": 2.442257881164551,
+      "learning_rate": 2.033811105222011e-05,
+      "loss": 0.2931,
+      "step": 12820
+    },
+    {
+      "epoch": 3.6617909382804137,
+      "grad_norm": 2.978790283203125,
+      "learning_rate": 2.025794836294129e-05,
+      "loss": 0.3047,
+      "step": 12830
+    },
+    {
+      "epoch": 3.66464502318944,
+      "grad_norm": 2.0821356773376465,
+      "learning_rate": 2.0177903812521952e-05,
+      "loss": 0.2902,
+      "step": 12840
+    },
+    {
+      "epoch": 3.667499108098466,
+      "grad_norm": 2.5075106620788574,
+      "learning_rate": 2.009797771890829e-05,
+      "loss": 0.3131,
+      "step": 12850
+    },
+    {
+      "epoch": 3.670353193007492,
+      "grad_norm": 2.6231086254119873,
+      "learning_rate": 2.0018170399575937e-05,
+      "loss": 0.3344,
+      "step": 12860
+    },
+    {
+      "epoch": 3.673207277916518,
+      "grad_norm": 3.2441518306732178,
+      "learning_rate": 1.99384821715288e-05,
+      "loss": 0.281,
+      "step": 12870
+    },
+    {
+      "epoch": 3.676061362825544,
+      "grad_norm": 3.035707950592041,
+      "learning_rate": 1.9858913351297675e-05,
+      "loss": 0.2973,
+      "step": 12880
+    },
+    {
+      "epoch": 3.6789154477345702,
+      "grad_norm": 1.7633196115493774,
+      "learning_rate": 1.9779464254939123e-05,
+      "loss": 0.2888,
+      "step": 12890
+    },
+    {
+      "epoch": 3.681769532643596,
+      "grad_norm": 3.896420955657959,
+      "learning_rate": 1.9700135198034105e-05,
+      "loss": 0.3169,
+      "step": 12900
+    },
+    {
+      "epoch": 3.684623617552622,
+      "grad_norm": 1.9077520370483398,
+      "learning_rate": 1.962092649568681e-05,
+      "loss": 0.2893,
+      "step": 12910
+    },
+    {
+      "epoch": 3.6874777024616483,
+      "grad_norm": 3.17303466796875,
+      "learning_rate": 1.95418384625233e-05,
+      "loss": 0.3275,
+      "step": 12920
+    },
+    {
+      "epoch": 3.690331787370674,
+      "grad_norm": 2.737929105758667,
+      "learning_rate": 1.946287141269038e-05,
+      "loss": 0.2909,
+      "step": 12930
+    },
+    {
+      "epoch": 3.6931858722797,
+      "grad_norm": 1.771128535270691,
+      "learning_rate": 1.9384025659854286e-05,
+      "loss": 0.288,
+      "step": 12940
+    },
+    {
+      "epoch": 3.6960399571887264,
+      "grad_norm": 2.433246612548828,
+      "learning_rate": 1.930530151719946e-05,
+      "loss": 0.338,
+      "step": 12950
+    },
+    {
+      "epoch": 3.6988940420977525,
+      "grad_norm": 4.06203556060791,
+      "learning_rate": 1.9226699297427238e-05,
+      "loss": 0.2982,
+      "step": 12960
+    },
+    {
+      "epoch": 3.7017481270067787,
+      "grad_norm": 2.2410807609558105,
+      "learning_rate": 1.9148219312754738e-05,
+      "loss": 0.3118,
+      "step": 12970
+    },
+    {
+      "epoch": 3.7046022119158044,
+      "grad_norm": 3.3286681175231934,
+      "learning_rate": 1.9069861874913476e-05,
+      "loss": 0.3091,
+      "step": 12980
+    },
+    {
+      "epoch": 3.7074562968248306,
+      "grad_norm": 1.8656421899795532,
+      "learning_rate": 1.899162729514826e-05,
+      "loss": 0.2938,
+      "step": 12990
+    },
+    {
+      "epoch": 3.7103103817338567,
+      "grad_norm": 2.353891372680664,
+      "learning_rate": 1.891351588421585e-05,
+      "loss": 0.2995,
+      "step": 13000
+    },
+    {
+      "epoch": 3.7131644666428825,
+      "grad_norm": 2.059058904647827,
+      "learning_rate": 1.8835527952383775e-05,
+      "loss": 0.2934,
+      "step": 13010
+    },
+    {
+      "epoch": 3.7160185515519086,
+      "grad_norm": 3.2514071464538574,
+      "learning_rate": 1.875766380942911e-05,
+      "loss": 0.2733,
+      "step": 13020
+    },
+    {
+      "epoch": 3.718872636460935,
+      "grad_norm": 2.666916608810425,
+      "learning_rate": 1.8679923764637197e-05,
+      "loss": 0.2966,
+      "step": 13030
+    },
+    {
+      "epoch": 3.7217267213699605,
+      "grad_norm": 2.5603129863739014,
+      "learning_rate": 1.8602308126800495e-05,
+      "loss": 0.2978,
+      "step": 13040
+    },
+    {
+      "epoch": 3.7245808062789867,
+      "grad_norm": 3.1166634559631348,
+      "learning_rate": 1.852481720421722e-05,
+      "loss": 0.316,
+      "step": 13050
+    },
+    {
+      "epoch": 3.727434891188013,
+      "grad_norm": 3.2752957344055176,
+      "learning_rate": 1.8447451304690282e-05,
+      "loss": 0.2901,
+      "step": 13060
+    },
+    {
+      "epoch": 3.730288976097039,
+      "grad_norm": 2.604976177215576,
+      "learning_rate": 1.8370210735525966e-05,
+      "loss": 0.2808,
+      "step": 13070
+    },
+    {
+      "epoch": 3.733143061006065,
+      "grad_norm": 2.2477569580078125,
+      "learning_rate": 1.8293095803532705e-05,
+      "loss": 0.2847,
+      "step": 13080
+    },
+    {
+      "epoch": 3.735997145915091,
+      "grad_norm": 1.942840814590454,
+      "learning_rate": 1.8216106815019896e-05,
+      "loss": 0.2814,
+      "step": 13090
+    },
+    {
+      "epoch": 3.738851230824117,
+      "grad_norm": 2.8237407207489014,
+      "learning_rate": 1.813924407579669e-05,
+      "loss": 0.275,
+      "step": 13100
+    },
+    {
+      "epoch": 3.7417053157331432,
+      "grad_norm": 2.4978528022766113,
+      "learning_rate": 1.8062507891170748e-05,
+      "loss": 0.3022,
+      "step": 13110
+    },
+    {
+      "epoch": 3.744559400642169,
+      "grad_norm": 2.675534248352051,
+      "learning_rate": 1.7985898565947074e-05,
+      "loss": 0.3083,
+      "step": 13120
+    },
+    {
+      "epoch": 3.747413485551195,
+      "grad_norm": 2.376801013946533,
+      "learning_rate": 1.7909416404426698e-05,
+      "loss": 0.303,
+      "step": 13130
+    },
+    {
+      "epoch": 3.7502675704602213,
+      "grad_norm": 2.0604560375213623,
+      "learning_rate": 1.783306171040562e-05,
+      "loss": 0.312,
+      "step": 13140
+    },
+    {
+      "epoch": 3.753121655369247,
+      "grad_norm": 3.991927146911621,
+      "learning_rate": 1.775683478717347e-05,
+      "loss": 0.2929,
+      "step": 13150
+    },
+    {
+      "epoch": 3.755975740278273,
+      "grad_norm": 2.6539063453674316,
+      "learning_rate": 1.7680735937512394e-05,
+      "loss": 0.2791,
+      "step": 13160
+    },
+    {
+      "epoch": 3.7588298251872994,
+      "grad_norm": 2.813647747039795,
+      "learning_rate": 1.7604765463695816e-05,
+      "loss": 0.3272,
+      "step": 13170
+    },
+    {
+      "epoch": 3.7616839100963255,
+      "grad_norm": 3.2123379707336426,
+      "learning_rate": 1.7528923667487194e-05,
+      "loss": 0.2951,
+      "step": 13180
+    },
+    {
+      "epoch": 3.7645379950053512,
+      "grad_norm": 2.988056182861328,
+      "learning_rate": 1.7453210850138883e-05,
+      "loss": 0.3069,
+      "step": 13190
+    },
+    {
+      "epoch": 3.7673920799143774,
+      "grad_norm": 3.634990692138672,
+      "learning_rate": 1.7377627312390936e-05,
+      "loss": 0.3111,
+      "step": 13200
+    },
+    {
+      "epoch": 3.7702461648234036,
+      "grad_norm": 3.2585556507110596,
+      "learning_rate": 1.7302173354469907e-05,
+      "loss": 0.2987,
+      "step": 13210
+    },
+    {
+      "epoch": 3.7731002497324297,
+      "grad_norm": 3.427927017211914,
+      "learning_rate": 1.722684927608758e-05,
+      "loss": 0.2701,
+      "step": 13220
+    },
+    {
+      "epoch": 3.7759543346414555,
+      "grad_norm": 1.788535714149475,
+      "learning_rate": 1.71516553764399e-05,
+      "loss": 0.3192,
+      "step": 13230
+    },
+    {
+      "epoch": 3.7788084195504816,
+      "grad_norm": 2.108631134033203,
+      "learning_rate": 1.7076591954205706e-05,
+      "loss": 0.3162,
+      "step": 13240
+    },
+    {
+      "epoch": 3.781662504459508,
+      "grad_norm": 2.7638750076293945,
+      "learning_rate": 1.7001659307545593e-05,
+      "loss": 0.3342,
+      "step": 13250
+    },
+    {
+      "epoch": 3.7845165893685335,
+      "grad_norm": 2.668900728225708,
+      "learning_rate": 1.6926857734100638e-05,
+      "loss": 0.3012,
+      "step": 13260
+    },
+    {
+      "epoch": 3.7873706742775597,
+      "grad_norm": 2.6184818744659424,
+      "learning_rate": 1.685218753099133e-05,
+      "loss": 0.3112,
+      "step": 13270
+    },
+    {
+      "epoch": 3.790224759186586,
+      "grad_norm": 2.3500993251800537,
+      "learning_rate": 1.677764899481636e-05,
+      "loss": 0.3093,
+      "step": 13280
+    },
+    {
+      "epoch": 3.7930788440956116,
+      "grad_norm": 2.0925824642181396,
+      "learning_rate": 1.670324242165134e-05,
+      "loss": 0.2909,
+      "step": 13290
+    },
+    {
+      "epoch": 3.7959329290046377,
+      "grad_norm": 2.932931900024414,
+      "learning_rate": 1.66289681070478e-05,
+      "loss": 0.2828,
+      "step": 13300
+    },
+    {
+      "epoch": 3.798787013913664,
+      "grad_norm": 2.048245906829834,
+      "learning_rate": 1.6554826346031855e-05,
+      "loss": 0.2886,
+      "step": 13310
+    },
+    {
+      "epoch": 3.80164109882269,
+      "grad_norm": 3.6790506839752197,
+      "learning_rate": 1.6480817433103145e-05,
+      "loss": 0.311,
+      "step": 13320
+    },
+    {
+      "epoch": 3.8044951837317162,
+      "grad_norm": 3.111971855163574,
+      "learning_rate": 1.6406941662233603e-05,
+      "loss": 0.2811,
+      "step": 13330
+    },
+    {
+      "epoch": 3.807349268640742,
+      "grad_norm": 2.412346839904785,
+      "learning_rate": 1.633319932686634e-05,
+      "loss": 0.294,
+      "step": 13340
+    },
+    {
+      "epoch": 3.810203353549768,
+      "grad_norm": 1.9881283044815063,
+      "learning_rate": 1.6259590719914375e-05,
+      "loss": 0.275,
+      "step": 13350
+    },
+    {
+      "epoch": 3.8130574384587943,
+      "grad_norm": 2.478259801864624,
+      "learning_rate": 1.6186116133759605e-05,
+      "loss": 0.289,
+      "step": 13360
+    },
+    {
+      "epoch": 3.81591152336782,
+      "grad_norm": 2.9626903533935547,
+      "learning_rate": 1.6112775860251555e-05,
+      "loss": 0.2881,
+      "step": 13370
+    },
+    {
+      "epoch": 3.818765608276846,
+      "grad_norm": 2.817441940307617,
+      "learning_rate": 1.6039570190706272e-05,
+      "loss": 0.3034,
+      "step": 13380
+    },
+    {
+      "epoch": 3.8216196931858724,
+      "grad_norm": 2.5437469482421875,
+      "learning_rate": 1.5966499415905084e-05,
+      "loss": 0.2885,
+      "step": 13390
+    },
+    {
+      "epoch": 3.824473778094898,
+      "grad_norm": 3.4150264263153076,
+      "learning_rate": 1.5893563826093573e-05,
+      "loss": 0.302,
+      "step": 13400
+    },
+    {
+      "epoch": 3.8273278630039242,
+      "grad_norm": 4.796387672424316,
+      "learning_rate": 1.5820763710980274e-05,
+      "loss": 0.3315,
+      "step": 13410
+    },
+    {
+      "epoch": 3.8301819479129504,
+      "grad_norm": 2.9776451587677,
+      "learning_rate": 1.574809935973566e-05,
+      "loss": 0.3092,
+      "step": 13420
+    },
+    {
+      "epoch": 3.8330360328219766,
+      "grad_norm": 2.4496400356292725,
+      "learning_rate": 1.567557106099094e-05,
+      "loss": 0.2907,
+      "step": 13430
+    },
+    {
+      "epoch": 3.8358901177310027,
+      "grad_norm": 2.5180628299713135,
+      "learning_rate": 1.5603179102836836e-05,
+      "loss": 0.2923,
+      "step": 13440
+    },
+    {
+      "epoch": 3.8387442026400285,
+      "grad_norm": 3.9845352172851562,
+      "learning_rate": 1.5530923772822575e-05,
+      "loss": 0.3157,
+      "step": 13450
+    },
+    {
+      "epoch": 3.8415982875490546,
+      "grad_norm": 2.3469271659851074,
+      "learning_rate": 1.545880535795467e-05,
+      "loss": 0.2932,
+      "step": 13460
+    },
+    {
+      "epoch": 3.844452372458081,
+      "grad_norm": 2.2121784687042236,
+      "learning_rate": 1.5386824144695788e-05,
+      "loss": 0.2957,
+      "step": 13470
+    },
+    {
+      "epoch": 3.8473064573671065,
+      "grad_norm": 3.311739206314087,
+      "learning_rate": 1.5314980418963587e-05,
+      "loss": 0.3123,
+      "step": 13480
+    },
+    {
+      "epoch": 3.8501605422761327,
+      "grad_norm": 3.6167421340942383,
+      "learning_rate": 1.5243274466129642e-05,
+      "loss": 0.3211,
+      "step": 13490
+    },
+    {
+      "epoch": 3.853014627185159,
+      "grad_norm": 2.3548507690429688,
+      "learning_rate": 1.5171706571018285e-05,
+      "loss": 0.3233,
+      "step": 13500
+    },
+    {
+      "epoch": 3.8558687120941846,
+      "grad_norm": 3.3471810817718506,
+      "learning_rate": 1.5100277017905435e-05,
+      "loss": 0.3226,
+      "step": 13510
+    },
+    {
+      "epoch": 3.8587227970032107,
+      "grad_norm": 2.185408115386963,
+      "learning_rate": 1.5028986090517495e-05,
+      "loss": 0.2849,
+      "step": 13520
+    },
+    {
+      "epoch": 3.861576881912237,
+      "grad_norm": 2.6843299865722656,
+      "learning_rate": 1.4957834072030275e-05,
+      "loss": 0.3071,
+      "step": 13530
+    },
+    {
+      "epoch": 3.864430966821263,
+      "grad_norm": 1.5020560026168823,
+      "learning_rate": 1.4886821245067794e-05,
+      "loss": 0.2749,
+      "step": 13540
+    },
+    {
+      "epoch": 3.8672850517302892,
+      "grad_norm": 2.128812074661255,
+      "learning_rate": 1.4815947891701198e-05,
+      "loss": 0.3247,
+      "step": 13550
+    },
+    {
+      "epoch": 3.870139136639315,
+      "grad_norm": 2.378420829772949,
+      "learning_rate": 1.4745214293447623e-05,
+      "loss": 0.3175,
+      "step": 13560
+    },
+    {
+      "epoch": 3.872993221548341,
+      "grad_norm": 2.813582420349121,
+      "learning_rate": 1.4674620731269063e-05,
+      "loss": 0.304,
+      "step": 13570
+    },
+    {
+      "epoch": 3.8758473064573673,
+      "grad_norm": 2.3528764247894287,
+      "learning_rate": 1.4604167485571291e-05,
+      "loss": 0.2993,
+      "step": 13580
+    },
+    {
+      "epoch": 3.878701391366393,
+      "grad_norm": 2.2043848037719727,
+      "learning_rate": 1.4533854836202737e-05,
+      "loss": 0.3302,
+      "step": 13590
+    },
+    {
+      "epoch": 3.881555476275419,
+      "grad_norm": 2.5757205486297607,
+      "learning_rate": 1.4463683062453359e-05,
+      "loss": 0.3121,
+      "step": 13600
+    },
+    {
+      "epoch": 3.8844095611844454,
+      "grad_norm": 2.663599967956543,
+      "learning_rate": 1.4393652443053529e-05,
+      "loss": 0.2829,
+      "step": 13610
+    },
+    {
+      "epoch": 3.887263646093471,
+      "grad_norm": 2.6079862117767334,
+      "learning_rate": 1.4323763256172923e-05,
+      "loss": 0.2868,
+      "step": 13620
+    },
+    {
+      "epoch": 3.8901177310024972,
+      "grad_norm": 2.833779811859131,
+      "learning_rate": 1.4254015779419471e-05,
+      "loss": 0.2869,
+      "step": 13630
+    },
+    {
+      "epoch": 3.8929718159115234,
+      "grad_norm": 4.291423797607422,
+      "learning_rate": 1.4184410289838213e-05,
+      "loss": 0.3272,
+      "step": 13640
+    },
+    {
+      "epoch": 3.895825900820549,
+      "grad_norm": 2.79622483253479,
+      "learning_rate": 1.4114947063910145e-05,
+      "loss": 0.315,
+      "step": 13650
+    },
+    {
+      "epoch": 3.8986799857295753,
+      "grad_norm": 2.05550479888916,
+      "learning_rate": 1.4045626377551224e-05,
+      "loss": 0.2795,
+      "step": 13660
+    },
+    {
+      "epoch": 3.9015340706386015,
+      "grad_norm": 2.668748378753662,
+      "learning_rate": 1.3976448506111222e-05,
+      "loss": 0.2925,
+      "step": 13670
+    },
+    {
+      "epoch": 3.9043881555476276,
+      "grad_norm": 2.7815065383911133,
+      "learning_rate": 1.390741372437262e-05,
+      "loss": 0.2608,
+      "step": 13680
+    },
+    {
+      "epoch": 3.907242240456654,
+      "grad_norm": 2.482853651046753,
+      "learning_rate": 1.3838522306549545e-05,
+      "loss": 0.3099,
+      "step": 13690
+    },
+    {
+      "epoch": 3.9100963253656795,
+      "grad_norm": 2.7168920040130615,
+      "learning_rate": 1.3769774526286627e-05,
+      "loss": 0.3012,
+      "step": 13700
+    },
+    {
+      "epoch": 3.9129504102747057,
+      "grad_norm": 3.665595769882202,
+      "learning_rate": 1.3701170656658008e-05,
+      "loss": 0.3395,
+      "step": 13710
+    },
+    {
+      "epoch": 3.915804495183732,
+      "grad_norm": 2.368117332458496,
+      "learning_rate": 1.363271097016614e-05,
+      "loss": 0.3121,
+      "step": 13720
+    },
+    {
+      "epoch": 3.9186585800927576,
+      "grad_norm": 2.8934361934661865,
+      "learning_rate": 1.356439573874082e-05,
+      "loss": 0.2885,
+      "step": 13730
+    },
+    {
+      "epoch": 3.9215126650017837,
+      "grad_norm": 2.124100685119629,
+      "learning_rate": 1.3496225233738003e-05,
+      "loss": 0.2802,
+      "step": 13740
+    },
+    {
+      "epoch": 3.92436674991081,
+      "grad_norm": 1.7779508829116821,
+      "learning_rate": 1.3428199725938812e-05,
+      "loss": 0.2835,
+      "step": 13750
+    },
+    {
+      "epoch": 3.9272208348198356,
+      "grad_norm": 1.9200053215026855,
+      "learning_rate": 1.3360319485548406e-05,
+      "loss": 0.3356,
+      "step": 13760
+    },
+    {
+      "epoch": 3.930074919728862,
+      "grad_norm": 2.548672914505005,
+      "learning_rate": 1.329258478219495e-05,
+      "loss": 0.3264,
+      "step": 13770
+    },
+    {
+      "epoch": 3.932929004637888,
+      "grad_norm": 2.681645154953003,
+      "learning_rate": 1.322499588492846e-05,
+      "loss": 0.3049,
+      "step": 13780
+    },
+    {
+      "epoch": 3.935783089546914,
+      "grad_norm": 3.1038978099823,
+      "learning_rate": 1.3157553062219851e-05,
+      "loss": 0.3223,
+      "step": 13790
+    },
+    {
+      "epoch": 3.9386371744559403,
+      "grad_norm": 2.6416921615600586,
+      "learning_rate": 1.3090256581959792e-05,
+      "loss": 0.3071,
+      "step": 13800
+    },
+    {
+      "epoch": 3.941491259364966,
+      "grad_norm": 2.4520456790924072,
+      "learning_rate": 1.3023106711457683e-05,
+      "loss": 0.307,
+      "step": 13810
+    },
+    {
+      "epoch": 3.944345344273992,
+      "grad_norm": 2.1851115226745605,
+      "learning_rate": 1.295610371744052e-05,
+      "loss": 0.2957,
+      "step": 13820
+    },
+    {
+      "epoch": 3.9471994291830184,
+      "grad_norm": 2.2908754348754883,
+      "learning_rate": 1.2889247866051956e-05,
+      "loss": 0.2919,
+      "step": 13830
+    },
+    {
+      "epoch": 3.950053514092044,
+      "grad_norm": 2.149648427963257,
+      "learning_rate": 1.282253942285111e-05,
+      "loss": 0.2872,
+      "step": 13840
+    },
+    {
+      "epoch": 3.9529075990010702,
+      "grad_norm": 3.1450397968292236,
+      "learning_rate": 1.2755978652811634e-05,
+      "loss": 0.303,
+      "step": 13850
+    },
+    {
+      "epoch": 3.9557616839100964,
+      "grad_norm": 3.7183732986450195,
+      "learning_rate": 1.2689565820320593e-05,
+      "loss": 0.3163,
+      "step": 13860
+    },
+    {
+      "epoch": 3.958615768819122,
+      "grad_norm": 4.74901819229126,
+      "learning_rate": 1.2623301189177388e-05,
+      "loss": 0.3026,
+      "step": 13870
+    },
+    {
+      "epoch": 3.9614698537281483,
+      "grad_norm": 3.503880023956299,
+      "learning_rate": 1.2557185022592794e-05,
+      "loss": 0.3185,
+      "step": 13880
+    },
+    {
+      "epoch": 3.9643239386371745,
+      "grad_norm": 2.3950393199920654,
+      "learning_rate": 1.2491217583187847e-05,
+      "loss": 0.2888,
+      "step": 13890
+    },
+    {
+      "epoch": 3.9671780235462006,
+      "grad_norm": 2.6387581825256348,
+      "learning_rate": 1.2425399132992843e-05,
+      "loss": 0.2724,
+      "step": 13900
+    },
+    {
+      "epoch": 3.970032108455227,
+      "grad_norm": 2.151468276977539,
+      "learning_rate": 1.235972993344623e-05,
+      "loss": 0.3204,
+      "step": 13910
+    },
+    {
+      "epoch": 3.9728861933642525,
+      "grad_norm": 3.798673391342163,
+      "learning_rate": 1.229421024539365e-05,
+      "loss": 0.316,
+      "step": 13920
+    },
+    {
+      "epoch": 3.9757402782732787,
+      "grad_norm": 2.186483383178711,
+      "learning_rate": 1.2228840329086882e-05,
+      "loss": 0.2883,
+      "step": 13930
+    },
+    {
+      "epoch": 3.978594363182305,
+      "grad_norm": 1.5906797647476196,
+      "learning_rate": 1.2163620444182744e-05,
+      "loss": 0.275,
+      "step": 13940
+    },
+    {
+      "epoch": 3.9814484480913306,
+      "grad_norm": 2.949838161468506,
+      "learning_rate": 1.2098550849742163e-05,
+      "loss": 0.2758,
+      "step": 13950
+    },
+    {
+      "epoch": 3.9843025330003567,
+      "grad_norm": 3.044550657272339,
+      "learning_rate": 1.2033631804229051e-05,
+      "loss": 0.2788,
+      "step": 13960
+    },
+    {
+      "epoch": 3.987156617909383,
+      "grad_norm": 3.4745771884918213,
+      "learning_rate": 1.1968863565509359e-05,
+      "loss": 0.3065,
+      "step": 13970
+    },
+    {
+      "epoch": 3.9900107028184086,
+      "grad_norm": 2.267524480819702,
+      "learning_rate": 1.1904246390849999e-05,
+      "loss": 0.2899,
+      "step": 13980
+    },
+    {
+      "epoch": 3.992864787727435,
+      "grad_norm": 2.396580457687378,
+      "learning_rate": 1.1839780536917854e-05,
+      "loss": 0.3214,
+      "step": 13990
+    },
+    {
+      "epoch": 3.995718872636461,
+      "grad_norm": 3.442814826965332,
+      "learning_rate": 1.1775466259778706e-05,
+      "loss": 0.3158,
+      "step": 14000
+    },
+    {
+      "epoch": 3.998572957545487,
+      "grad_norm": 2.5229299068450928,
+      "learning_rate": 1.1711303814896291e-05,
+      "loss": 0.3123,
+      "step": 14010
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 17515,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.597020138453336e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}