diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6802 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 9676,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0010334849111202976,
+      "grad_norm": 3.4666478633880615,
+      "learning_rate": 0.0006200909466721785,
+      "loss": 7.9994,
+      "step": 10
+    },
+    {
+      "epoch": 0.002066969822240595,
+      "grad_norm": 1.0015512704849243,
+      "learning_rate": 0.001240181893344357,
+      "loss": 7.0943,
+      "step": 20
+    },
+    {
+      "epoch": 0.0031004547333608927,
+      "grad_norm": 0.8696095943450928,
+      "learning_rate": 0.0018602728400165356,
+      "loss": 6.3072,
+      "step": 30
+    },
+    {
+      "epoch": 0.00413393964448119,
+      "grad_norm": 0.32349398732185364,
+      "learning_rate": 0.002480363786688714,
+      "loss": 5.9694,
+      "step": 40
+    },
+    {
+      "epoch": 0.005167424555601488,
+      "grad_norm": 1.6740758419036865,
+      "learning_rate": 0.0031004547333608927,
+      "loss": 5.8536,
+      "step": 50
+    },
+    {
+      "epoch": 0.0062009094667217855,
+      "grad_norm": 1.6459004878997803,
+      "learning_rate": 0.003720545680033071,
+      "loss": 5.7039,
+      "step": 60
+    },
+    {
+      "epoch": 0.007234394377842084,
+      "grad_norm": 0.48763737082481384,
+      "learning_rate": 0.0043406366267052495,
+      "loss": 5.4789,
+      "step": 70
+    },
+    {
+      "epoch": 0.00826787928896238,
+      "grad_norm": 0.4977681338787079,
+      "learning_rate": 0.004960727573377428,
+      "loss": 5.269,
+      "step": 80
+    },
+    {
+      "epoch": 0.009301364200082679,
+      "grad_norm": 0.5395390391349792,
+      "learning_rate": 0.005580818520049607,
+      "loss": 5.0768,
+      "step": 90
+    },
+    {
+      "epoch": 0.010334849111202976,
+      "grad_norm": 0.4573260247707367,
+      "learning_rate": 0.006,
+      "loss": 4.9493,
+      "step": 100
+    },
+    {
+      "epoch": 0.011368334022323274,
+      "grad_norm": 0.5551472902297974,
+      "learning_rate": 0.006,
+      "loss": 4.7982,
+      "step": 110
+    },
+    {
+      "epoch": 0.012401818933443571,
+      "grad_norm": 0.4980376958847046,
+      "learning_rate": 0.006,
+      "loss": 4.6285,
+      "step": 120
+    },
+    {
+      "epoch": 0.01343530384456387,
+      "grad_norm": 0.4951136112213135,
+      "learning_rate": 0.006,
+      "loss": 4.5176,
+      "step": 130
+    },
+    {
+      "epoch": 0.014468788755684168,
+      "grad_norm": 0.5360985398292542,
+      "learning_rate": 0.006,
+      "loss": 4.3572,
+      "step": 140
+    },
+    {
+      "epoch": 0.015502273666804465,
+      "grad_norm": 0.4469200670719147,
+      "learning_rate": 0.006,
+      "loss": 4.214,
+      "step": 150
+    },
+    {
+      "epoch": 0.01653575857792476,
+      "grad_norm": 0.6157914400100708,
+      "learning_rate": 0.006,
+      "loss": 4.0854,
+      "step": 160
+    },
+    {
+      "epoch": 0.01756924348904506,
+      "grad_norm": 0.4493155777454376,
+      "learning_rate": 0.006,
+      "loss": 3.9583,
+      "step": 170
+    },
+    {
+      "epoch": 0.018602728400165358,
+      "grad_norm": 0.6523879766464233,
+      "learning_rate": 0.006,
+      "loss": 3.8434,
+      "step": 180
+    },
+    {
+      "epoch": 0.019636213311285657,
+      "grad_norm": 0.7103180885314941,
+      "learning_rate": 0.006,
+      "loss": 3.7414,
+      "step": 190
+    },
+    {
+      "epoch": 0.02066969822240595,
+      "grad_norm": 0.44575124979019165,
+      "learning_rate": 0.006,
+      "loss": 3.6829,
+      "step": 200
+    },
+    {
+      "epoch": 0.02170318313352625,
+      "grad_norm": 0.5272648334503174,
+      "learning_rate": 0.006,
+      "loss": 3.6056,
+      "step": 210
+    },
+    {
+      "epoch": 0.02273666804464655,
+      "grad_norm": 0.4514292776584625,
+      "learning_rate": 0.006,
+      "loss": 3.5279,
+      "step": 220
+    },
+    {
+      "epoch": 0.023770152955766847,
+      "grad_norm": 0.39017120003700256,
+      "learning_rate": 0.006,
+      "loss": 3.4585,
+      "step": 230
+    },
+    {
+      "epoch": 0.024803637866887142,
+      "grad_norm": 0.4593961536884308,
+      "learning_rate": 0.006,
+      "loss": 3.4018,
+      "step": 240
+    },
+    {
+      "epoch": 0.02583712277800744,
+      "grad_norm": 0.48860013484954834,
+      "learning_rate": 0.006,
+      "loss": 3.3523,
+      "step": 250
+    },
+    {
+      "epoch": 0.02687060768912774,
+      "grad_norm": 0.36506471037864685,
+      "learning_rate": 0.006,
+      "loss": 3.3124,
+      "step": 260
+    },
+    {
+      "epoch": 0.027904092600248037,
+      "grad_norm": 0.3506007194519043,
+      "learning_rate": 0.006,
+      "loss": 3.2678,
+      "step": 270
+    },
+    {
+      "epoch": 0.028937577511368336,
+      "grad_norm": 0.3703238070011139,
+      "learning_rate": 0.006,
+      "loss": 3.2231,
+      "step": 280
+    },
+    {
+      "epoch": 0.02997106242248863,
+      "grad_norm": 0.45603325963020325,
+      "learning_rate": 0.006,
+      "loss": 3.1847,
+      "step": 290
+    },
+    {
+      "epoch": 0.03100454733360893,
+      "grad_norm": 0.37536856532096863,
+      "learning_rate": 0.006,
+      "loss": 3.1585,
+      "step": 300
+    },
+    {
+      "epoch": 0.032038032244729224,
+      "grad_norm": 0.3810868561267853,
+      "learning_rate": 0.006,
+      "loss": 3.1063,
+      "step": 310
+    },
+    {
+      "epoch": 0.03307151715584952,
+      "grad_norm": 0.4568467438220978,
+      "learning_rate": 0.006,
+      "loss": 3.0649,
+      "step": 320
+    },
+    {
+      "epoch": 0.03410500206696982,
+      "grad_norm": 0.36088162660598755,
+      "learning_rate": 0.006,
+      "loss": 3.0495,
+      "step": 330
+    },
+    {
+      "epoch": 0.03513848697809012,
+      "grad_norm": 0.3663162887096405,
+      "learning_rate": 0.006,
+      "loss": 3.0188,
+      "step": 340
+    },
+    {
+      "epoch": 0.03617197188921042,
+      "grad_norm": 0.5975127220153809,
+      "learning_rate": 0.006,
+      "loss": 2.994,
+      "step": 350
+    },
+    {
+      "epoch": 0.037205456800330716,
+      "grad_norm": 0.3454865515232086,
+      "learning_rate": 0.006,
+      "loss": 2.9814,
+      "step": 360
+    },
+    {
+      "epoch": 0.038238941711451015,
+      "grad_norm": 0.4206918179988861,
+      "learning_rate": 0.006,
+      "loss": 2.9306,
+      "step": 370
+    },
+    {
+      "epoch": 0.03927242662257131,
+      "grad_norm": 0.3579295575618744,
+      "learning_rate": 0.006,
+      "loss": 2.9165,
+      "step": 380
+    },
+    {
+      "epoch": 0.04030591153369161,
+      "grad_norm": 0.3658403158187866,
+      "learning_rate": 0.006,
+      "loss": 2.9103,
+      "step": 390
+    },
+    {
+      "epoch": 0.0413393964448119,
+      "grad_norm": 0.3459131717681885,
+      "learning_rate": 0.006,
+      "loss": 2.8628,
+      "step": 400
+    },
+    {
+      "epoch": 0.0423728813559322,
+      "grad_norm": 0.4726301431655884,
+      "learning_rate": 0.006,
+      "loss": 2.8334,
+      "step": 410
+    },
+    {
+      "epoch": 0.0434063662670525,
+      "grad_norm": 0.3717091381549835,
+      "learning_rate": 0.006,
+      "loss": 2.8188,
+      "step": 420
+    },
+    {
+      "epoch": 0.0444398511781728,
+      "grad_norm": 0.3234303295612335,
+      "learning_rate": 0.006,
+      "loss": 2.8076,
+      "step": 430
+    },
+    {
+      "epoch": 0.0454733360892931,
+      "grad_norm": 0.4717100262641907,
+      "learning_rate": 0.006,
+      "loss": 2.7871,
+      "step": 440
+    },
+    {
+      "epoch": 0.046506821000413395,
+      "grad_norm": 0.35768577456474304,
+      "learning_rate": 0.006,
+      "loss": 2.7601,
+      "step": 450
+    },
+    {
+      "epoch": 0.047540305911533694,
+      "grad_norm": 0.3499695062637329,
+      "learning_rate": 0.006,
+      "loss": 2.73,
+      "step": 460
+    },
+    {
+      "epoch": 0.04857379082265399,
+      "grad_norm": 0.4336409866809845,
+      "learning_rate": 0.006,
+      "loss": 2.7235,
+      "step": 470
+    },
+    {
+      "epoch": 0.049607275733774284,
+      "grad_norm": 0.39916467666625977,
+      "learning_rate": 0.006,
+      "loss": 2.7007,
+      "step": 480
+    },
+    {
+      "epoch": 0.05064076064489458,
+      "grad_norm": 0.3249908983707428,
+      "learning_rate": 0.006,
+      "loss": 2.6705,
+      "step": 490
+    },
+    {
+      "epoch": 0.05167424555601488,
+      "grad_norm": 0.36379745602607727,
+      "learning_rate": 0.006,
+      "loss": 2.6393,
+      "step": 500
+    },
+    {
+      "epoch": 0.05270773046713518,
+      "grad_norm": 0.3417607843875885,
+      "learning_rate": 0.006,
+      "loss": 2.6767,
+      "step": 510
+    },
+    {
+      "epoch": 0.05374121537825548,
+      "grad_norm": 0.2848694920539856,
+      "learning_rate": 0.006,
+      "loss": 2.6244,
+      "step": 520
+    },
+    {
+      "epoch": 0.054774700289375776,
+      "grad_norm": 0.2725778818130493,
+      "learning_rate": 0.006,
+      "loss": 2.5704,
+      "step": 530
+    },
+    {
+      "epoch": 0.055808185200496074,
+      "grad_norm": 0.2887585461139679,
+      "learning_rate": 0.006,
+      "loss": 2.5508,
+      "step": 540
+    },
+    {
+      "epoch": 0.05684167011161637,
+      "grad_norm": 0.422413170337677,
+      "learning_rate": 0.006,
+      "loss": 2.5433,
+      "step": 550
+    },
+    {
+      "epoch": 0.05787515502273667,
+      "grad_norm": 0.5642419457435608,
+      "learning_rate": 0.006,
+      "loss": 2.5267,
+      "step": 560
+    },
+    {
+      "epoch": 0.05890863993385696,
+      "grad_norm": 0.33949220180511475,
+      "learning_rate": 0.006,
+      "loss": 2.4942,
+      "step": 570
+    },
+    {
+      "epoch": 0.05994212484497726,
+      "grad_norm": 0.36603158712387085,
+      "learning_rate": 0.006,
+      "loss": 2.4755,
+      "step": 580
+    },
+    {
+      "epoch": 0.06097560975609756,
+      "grad_norm": 0.3936167061328888,
+      "learning_rate": 0.006,
+      "loss": 2.4358,
+      "step": 590
+    },
+    {
+      "epoch": 0.06200909466721786,
+      "grad_norm": 0.46108072996139526,
+      "learning_rate": 0.006,
+      "loss": 2.4312,
+      "step": 600
+    },
+    {
+      "epoch": 0.06304257957833816,
+      "grad_norm": 0.3659123480319977,
+      "learning_rate": 0.006,
+      "loss": 2.4023,
+      "step": 610
+    },
+    {
+      "epoch": 0.06407606448945845,
+      "grad_norm": 0.4888301193714142,
+      "learning_rate": 0.006,
+      "loss": 2.3887,
+      "step": 620
+    },
+    {
+      "epoch": 0.06510954940057875,
+      "grad_norm": 0.33769381046295166,
+      "learning_rate": 0.006,
+      "loss": 2.3645,
+      "step": 630
+    },
+    {
+      "epoch": 0.06614303431169904,
+      "grad_norm": 0.4479112923145294,
+      "learning_rate": 0.006,
+      "loss": 2.3123,
+      "step": 640
+    },
+    {
+      "epoch": 0.06717651922281935,
+      "grad_norm": 0.3715905249118805,
+      "learning_rate": 0.006,
+      "loss": 2.3009,
+      "step": 650
+    },
+    {
+      "epoch": 0.06821000413393964,
+      "grad_norm": 0.43582767248153687,
+      "learning_rate": 0.006,
+      "loss": 2.3256,
+      "step": 660
+    },
+    {
+      "epoch": 0.06924348904505995,
+      "grad_norm": 0.295462042093277,
+      "learning_rate": 0.006,
+      "loss": 2.281,
+      "step": 670
+    },
+    {
+      "epoch": 0.07027697395618024,
+      "grad_norm": 0.2555060386657715,
+      "learning_rate": 0.006,
+      "loss": 2.2711,
+      "step": 680
+    },
+    {
+      "epoch": 0.07131045886730054,
+      "grad_norm": 0.2812485992908478,
+      "learning_rate": 0.006,
+      "loss": 2.2392,
+      "step": 690
+    },
+    {
+      "epoch": 0.07234394377842084,
+      "grad_norm": 0.2797698974609375,
+      "learning_rate": 0.006,
+      "loss": 2.2316,
+      "step": 700
+    },
+    {
+      "epoch": 0.07337742868954113,
+      "grad_norm": 0.4201503396034241,
+      "learning_rate": 0.006,
+      "loss": 2.1894,
+      "step": 710
+    },
+    {
+      "epoch": 0.07441091360066143,
+      "grad_norm": 0.48455363512039185,
+      "learning_rate": 0.006,
+      "loss": 2.1682,
+      "step": 720
+    },
+    {
+      "epoch": 0.07544439851178172,
+      "grad_norm": 0.4906412959098816,
+      "learning_rate": 0.006,
+      "loss": 2.1594,
+      "step": 730
+    },
+    {
+      "epoch": 0.07647788342290203,
+      "grad_norm": 0.3227013051509857,
+      "learning_rate": 0.006,
+      "loss": 2.148,
+      "step": 740
+    },
+    {
+      "epoch": 0.07751136833402232,
+      "grad_norm": 0.24952304363250732,
+      "learning_rate": 0.006,
+      "loss": 2.1323,
+      "step": 750
+    },
+    {
+      "epoch": 0.07854485324514263,
+      "grad_norm": 0.4688262641429901,
+      "learning_rate": 0.006,
+      "loss": 2.0781,
+      "step": 760
+    },
+    {
+      "epoch": 0.07957833815626292,
+      "grad_norm": 0.3051510453224182,
+      "learning_rate": 0.006,
+      "loss": 2.1005,
+      "step": 770
+    },
+    {
+      "epoch": 0.08061182306738322,
+      "grad_norm": 0.303608775138855,
+      "learning_rate": 0.006,
+      "loss": 2.0619,
+      "step": 780
+    },
+    {
+      "epoch": 0.08164530797850351,
+      "grad_norm": 0.41049811244010925,
+      "learning_rate": 0.006,
+      "loss": 2.069,
+      "step": 790
+    },
+    {
+      "epoch": 0.0826787928896238,
+      "grad_norm": 0.25236430764198303,
+      "learning_rate": 0.006,
+      "loss": 2.0536,
+      "step": 800
+    },
+    {
+      "epoch": 0.08371227780074411,
+      "grad_norm": 0.2919696569442749,
+      "learning_rate": 0.006,
+      "loss": 2.0221,
+      "step": 810
+    },
+    {
+      "epoch": 0.0847457627118644,
+      "grad_norm": 0.3965113162994385,
+      "learning_rate": 0.006,
+      "loss": 2.0008,
+      "step": 820
+    },
+    {
+      "epoch": 0.08577924762298471,
+      "grad_norm": 0.3784347176551819,
+      "learning_rate": 0.006,
+      "loss": 2.0041,
+      "step": 830
+    },
+    {
+      "epoch": 0.086812732534105,
+      "grad_norm": 0.32382646203041077,
+      "learning_rate": 0.006,
+      "loss": 1.9987,
+      "step": 840
+    },
+    {
+      "epoch": 0.0878462174452253,
+      "grad_norm": 0.28474798798561096,
+      "learning_rate": 0.006,
+      "loss": 1.9859,
+      "step": 850
+    },
+    {
+      "epoch": 0.0888797023563456,
+      "grad_norm": 0.35125505924224854,
+      "learning_rate": 0.006,
+      "loss": 1.9719,
+      "step": 860
+    },
+    {
+      "epoch": 0.08991318726746589,
+      "grad_norm": 0.2959253787994385,
+      "learning_rate": 0.006,
+      "loss": 1.9691,
+      "step": 870
+    },
+    {
+      "epoch": 0.0909466721785862,
+      "grad_norm": 0.2681068181991577,
+      "learning_rate": 0.006,
+      "loss": 1.946,
+      "step": 880
+    },
+    {
+      "epoch": 0.09198015708970649,
+      "grad_norm": 0.2882836163043976,
+      "learning_rate": 0.006,
+      "loss": 1.9227,
+      "step": 890
+    },
+    {
+      "epoch": 0.09301364200082679,
+      "grad_norm": 0.2792555093765259,
+      "learning_rate": 0.006,
+      "loss": 1.9067,
+      "step": 900
+    },
+    {
+      "epoch": 0.09404712691194708,
+      "grad_norm": 0.31801220774650574,
+      "learning_rate": 0.006,
+      "loss": 1.8997,
+      "step": 910
+    },
+    {
+      "epoch": 0.09508061182306739,
+      "grad_norm": 0.3104709982872009,
+      "learning_rate": 0.006,
+      "loss": 1.909,
+      "step": 920
+    },
+    {
+      "epoch": 0.09611409673418768,
+      "grad_norm": 0.4227132499217987,
+      "learning_rate": 0.006,
+      "loss": 1.8898,
+      "step": 930
+    },
+    {
+      "epoch": 0.09714758164530798,
+      "grad_norm": 0.2799486815929413,
+      "learning_rate": 0.006,
+      "loss": 1.8799,
+      "step": 940
+    },
+    {
+      "epoch": 0.09818106655642828,
+      "grad_norm": 0.25304150581359863,
+      "learning_rate": 0.006,
+      "loss": 1.8542,
+      "step": 950
+    },
+    {
+      "epoch": 0.09921455146754857,
+      "grad_norm": 0.40398165583610535,
+      "learning_rate": 0.006,
+      "loss": 1.8495,
+      "step": 960
+    },
+    {
+      "epoch": 0.10024803637866887,
+      "grad_norm": 0.24255605041980743,
+      "learning_rate": 0.006,
+      "loss": 1.8279,
+      "step": 970
+    },
+    {
+      "epoch": 0.10128152128978916,
+      "grad_norm": 0.251412957906723,
+      "learning_rate": 0.006,
+      "loss": 1.8448,
+      "step": 980
+    },
+    {
+      "epoch": 0.10231500620090947,
+      "grad_norm": 0.33690279722213745,
+      "learning_rate": 0.006,
+      "loss": 1.8345,
+      "step": 990
+    },
+    {
+      "epoch": 0.10334849111202976,
+      "grad_norm": 0.3986241817474365,
+      "learning_rate": 0.006,
+      "loss": 1.8272,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10438197602315007,
+      "grad_norm": 0.275290310382843,
+      "learning_rate": 0.006,
+      "loss": 1.8177,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10541546093427036,
+      "grad_norm": 0.24022069573402405,
+      "learning_rate": 0.006,
+      "loss": 1.8,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10644894584539066,
+      "grad_norm": 0.22507286071777344,
+      "learning_rate": 0.006,
+      "loss": 1.7825,
+      "step": 1030
+    },
+    {
+      "epoch": 0.10748243075651095,
+      "grad_norm": 0.2965947985649109,
+      "learning_rate": 0.006,
+      "loss": 1.778,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10851591566763125,
+      "grad_norm": 0.3291703164577484,
+      "learning_rate": 0.006,
+      "loss": 1.7864,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10954940057875155,
+      "grad_norm": 0.2484293133020401,
+      "learning_rate": 0.006,
+      "loss": 1.7649,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11058288548987184,
+      "grad_norm": 0.30289486050605774,
+      "learning_rate": 0.006,
+      "loss": 1.7756,
+      "step": 1070
+    },
+    {
+      "epoch": 0.11161637040099215,
+      "grad_norm": 0.2765989601612091,
+      "learning_rate": 0.006,
+      "loss": 1.7458,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11264985531211244,
+      "grad_norm": 0.24408799409866333,
+      "learning_rate": 0.006,
+      "loss": 1.7598,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11368334022323275,
+      "grad_norm": 0.2399667352437973,
+      "learning_rate": 0.006,
+      "loss": 1.754,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11471682513435304,
+      "grad_norm": 0.2846289873123169,
+      "learning_rate": 0.006,
+      "loss": 1.7406,
+      "step": 1110
+    },
+    {
+      "epoch": 0.11575031004547334,
+      "grad_norm": 0.26227328181266785,
+      "learning_rate": 0.006,
+      "loss": 1.7336,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11678379495659363,
+      "grad_norm": 0.2673957347869873,
+      "learning_rate": 0.006,
+      "loss": 1.7281,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11781727986771393,
+      "grad_norm": 0.3353629410266876,
+      "learning_rate": 0.006,
+      "loss": 1.7169,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11885076477883423,
+      "grad_norm": 0.21159133315086365,
+      "learning_rate": 0.006,
+      "loss": 1.7016,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11988424968995452,
+      "grad_norm": 0.24120958149433136,
+      "learning_rate": 0.006,
+      "loss": 1.7033,
+      "step": 1160
+    },
+    {
+      "epoch": 0.12091773460107483,
+      "grad_norm": 0.26428887248039246,
+      "learning_rate": 0.006,
+      "loss": 1.6867,
+      "step": 1170
+    },
+    {
+      "epoch": 0.12195121951219512,
+      "grad_norm": 0.27754953503608704,
+      "learning_rate": 0.006,
+      "loss": 1.6901,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12298470442331542,
+      "grad_norm": 0.36125853657722473,
+      "learning_rate": 0.006,
+      "loss": 1.6808,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12401818933443572,
+      "grad_norm": 0.23145093023777008,
+      "learning_rate": 0.006,
+      "loss": 1.6946,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12505167424555602,
+      "grad_norm": 0.22397373616695404,
+      "learning_rate": 0.006,
+      "loss": 1.6619,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1260851591566763,
+      "grad_norm": 0.2219894528388977,
+      "learning_rate": 0.006,
+      "loss": 1.6483,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1271186440677966,
+      "grad_norm": 0.2552240788936615,
+      "learning_rate": 0.006,
+      "loss": 1.6574,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1281521289789169,
+      "grad_norm": 0.4382542073726654,
+      "learning_rate": 0.006,
+      "loss": 1.663,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12918561389003722,
+      "grad_norm": 0.2551495134830475,
+      "learning_rate": 0.006,
+      "loss": 1.6543,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1302190988011575,
+      "grad_norm": 0.21785561740398407,
+      "learning_rate": 0.006,
+      "loss": 1.6479,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1312525837122778,
+      "grad_norm": 0.25551608204841614,
+      "learning_rate": 0.006,
+      "loss": 1.6303,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1322860686233981,
+      "grad_norm": 0.30908089876174927,
+      "learning_rate": 0.006,
+      "loss": 1.6388,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1333195535345184,
+      "grad_norm": 0.39932873845100403,
+      "learning_rate": 0.006,
+      "loss": 1.6192,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1343530384456387,
+      "grad_norm": 0.2735249102115631,
+      "learning_rate": 0.006,
+      "loss": 1.6327,
+      "step": 1300
+    },
+    {
+      "epoch": 0.135386523356759,
+      "grad_norm": 0.23229070007801056,
+      "learning_rate": 0.006,
+      "loss": 1.6278,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13642000826787928,
+      "grad_norm": 0.29770010709762573,
+      "learning_rate": 0.006,
+      "loss": 1.6137,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13745349317899958,
+      "grad_norm": 0.2619315981864929,
+      "learning_rate": 0.006,
+      "loss": 1.6071,
+      "step": 1330
+    },
+    {
+      "epoch": 0.1384869780901199,
+      "grad_norm": 0.2108081877231598,
+      "learning_rate": 0.006,
+      "loss": 1.6189,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13952046300124019,
+      "grad_norm": 0.2384573072195053,
+      "learning_rate": 0.006,
+      "loss": 1.6035,
+      "step": 1350
+    },
+    {
+      "epoch": 0.14055394791236048,
+      "grad_norm": 0.2051675170660019,
+      "learning_rate": 0.006,
+      "loss": 1.5866,
+      "step": 1360
+    },
+    {
+      "epoch": 0.14158743282348077,
+      "grad_norm": 0.3068487346172333,
+      "learning_rate": 0.006,
+      "loss": 1.5727,
+      "step": 1370
+    },
+    {
+      "epoch": 0.1426209177346011,
+      "grad_norm": 0.24623431265354156,
+      "learning_rate": 0.006,
+      "loss": 1.5786,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14365440264572138,
+      "grad_norm": 0.3030165731906891,
+      "learning_rate": 0.006,
+      "loss": 1.5793,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14468788755684167,
+      "grad_norm": 0.236822709441185,
+      "learning_rate": 0.006,
+      "loss": 1.5737,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14572137246796196,
+      "grad_norm": 0.4476490318775177,
+      "learning_rate": 0.006,
+      "loss": 1.5743,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14675485737908225,
+      "grad_norm": 0.3274894952774048,
+      "learning_rate": 0.006,
+      "loss": 1.5579,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14778834229020257,
+      "grad_norm": 0.2623043358325958,
+      "learning_rate": 0.006,
+      "loss": 1.5666,
+      "step": 1430
+    },
+    {
+      "epoch": 0.14882182720132286,
+      "grad_norm": 0.23737861216068268,
+      "learning_rate": 0.006,
+      "loss": 1.553,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14985531211244316,
+      "grad_norm": 0.23686368763446808,
+      "learning_rate": 0.006,
+      "loss": 1.5598,
+      "step": 1450
+    },
+    {
+      "epoch": 0.15088879702356345,
+      "grad_norm": 0.2711767256259918,
+      "learning_rate": 0.006,
+      "loss": 1.5508,
+      "step": 1460
+    },
+    {
+      "epoch": 0.15192228193468377,
+      "grad_norm": 0.1919260323047638,
+      "learning_rate": 0.006,
+      "loss": 1.5364,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15295576684580406,
+      "grad_norm": 0.24973885715007782,
+      "learning_rate": 0.006,
+      "loss": 1.5444,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15398925175692435,
+      "grad_norm": 0.2096351981163025,
+      "learning_rate": 0.006,
+      "loss": 1.5422,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15502273666804464,
+      "grad_norm": 0.28911715745925903,
+      "learning_rate": 0.006,
+      "loss": 1.5562,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15605622157916493,
+      "grad_norm": 0.2608921229839325,
+      "learning_rate": 0.006,
+      "loss": 1.5153,
+      "step": 1510
+    },
+    {
+      "epoch": 0.15708970649028525,
+      "grad_norm": 0.2605489194393158,
+      "learning_rate": 0.006,
+      "loss": 1.5262,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15812319140140554,
+      "grad_norm": 0.2948496341705322,
+      "learning_rate": 0.006,
+      "loss": 1.5318,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15915667631252584,
+      "grad_norm": 0.33066225051879883,
+      "learning_rate": 0.006,
+      "loss": 1.5138,
+      "step": 1540
+    },
+    {
+      "epoch": 0.16019016122364613,
+      "grad_norm": 0.2134758085012436,
+      "learning_rate": 0.006,
+      "loss": 1.5199,
+      "step": 1550
+    },
+    {
+      "epoch": 0.16122364613476645,
+      "grad_norm": 0.22920167446136475,
+      "learning_rate": 0.006,
+      "loss": 1.5084,
+      "step": 1560
+    },
+    {
+      "epoch": 0.16225713104588674,
+      "grad_norm": 0.22044587135314941,
+      "learning_rate": 0.006,
+      "loss": 1.5254,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16329061595700703,
+      "grad_norm": 0.28658127784729004,
+      "learning_rate": 0.006,
+      "loss": 1.5149,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16432410086812732,
+      "grad_norm": 0.1932067573070526,
+      "learning_rate": 0.006,
+      "loss": 1.4943,
+      "step": 1590
+    },
+    {
+      "epoch": 0.1653575857792476,
+      "grad_norm": 0.19694995880126953,
+      "learning_rate": 0.006,
+      "loss": 1.4871,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16639107069036793,
+      "grad_norm": 0.2092551738023758,
+      "learning_rate": 0.006,
+      "loss": 1.481,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16742455560148822,
+      "grad_norm": 0.20461459457874298,
+      "learning_rate": 0.006,
+      "loss": 1.4958,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16845804051260851,
+      "grad_norm": 0.2160821110010147,
+      "learning_rate": 0.006,
+      "loss": 1.5013,
+      "step": 1630
+    },
+    {
+      "epoch": 0.1694915254237288,
+      "grad_norm": 0.23302793502807617,
+      "learning_rate": 0.006,
+      "loss": 1.4991,
+      "step": 1640
+    },
+    {
+      "epoch": 0.1705250103348491,
+      "grad_norm": 0.20844736695289612,
+      "learning_rate": 0.006,
+      "loss": 1.4765,
+      "step": 1650
+    },
+    {
+      "epoch": 0.17155849524596942,
+      "grad_norm": 0.18730531632900238,
+      "learning_rate": 0.006,
+      "loss": 1.4732,
+      "step": 1660
+    },
+    {
+      "epoch": 0.1725919801570897,
+      "grad_norm": 0.2798612415790558,
+      "learning_rate": 0.006,
+      "loss": 1.4809,
+      "step": 1670
+    },
+    {
+      "epoch": 0.17362546506821,
+      "grad_norm": 0.17376913130283356,
+      "learning_rate": 0.006,
+      "loss": 1.4847,
+      "step": 1680
+    },
+    {
+      "epoch": 0.1746589499793303,
+      "grad_norm": 0.1984642595052719,
+      "learning_rate": 0.006,
+      "loss": 1.4677,
+      "step": 1690
+    },
+    {
+      "epoch": 0.1756924348904506,
+      "grad_norm": 0.18918636441230774,
+      "learning_rate": 0.006,
+      "loss": 1.4623,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1767259198015709,
+      "grad_norm": 0.22821608185768127,
+      "learning_rate": 0.006,
+      "loss": 1.4836,
+      "step": 1710
+    },
+    {
+      "epoch": 0.1777594047126912,
+      "grad_norm": 0.2023961991071701,
+      "learning_rate": 0.006,
+      "loss": 1.4665,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17879288962381149,
+      "grad_norm": 0.21143494546413422,
+      "learning_rate": 0.006,
+      "loss": 1.4637,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17982637453493178,
+      "grad_norm": 0.18367299437522888,
+      "learning_rate": 0.006,
+      "loss": 1.4564,
+      "step": 1740
+    },
+    {
+      "epoch": 0.1808598594460521,
+      "grad_norm": 0.18500660359859467,
+      "learning_rate": 0.006,
+      "loss": 1.4485,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1818933443571724,
+      "grad_norm": 0.208240807056427,
+      "learning_rate": 0.006,
+      "loss": 1.449,
+      "step": 1760
+    },
+    {
+      "epoch": 0.18292682926829268,
+      "grad_norm": 0.28708386421203613,
+      "learning_rate": 0.006,
+      "loss": 1.4514,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18396031417941297,
+      "grad_norm": 0.23758843541145325,
+      "learning_rate": 0.006,
+      "loss": 1.4376,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1849937990905333,
+      "grad_norm": 0.23937605321407318,
+      "learning_rate": 0.006,
+      "loss": 1.4406,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18602728400165358,
+      "grad_norm": 0.44142553210258484,
+      "learning_rate": 0.006,
+      "loss": 1.4331,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18706076891277387,
+      "grad_norm": 0.1951988786458969,
+      "learning_rate": 0.006,
+      "loss": 1.4443,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18809425382389416,
+      "grad_norm": 0.22376669943332672,
+      "learning_rate": 0.006,
+      "loss": 1.4345,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18912773873501446,
+      "grad_norm": 0.24400733411312103,
+      "learning_rate": 0.006,
+      "loss": 1.4405,
+      "step": 1830
+    },
+    {
+      "epoch": 0.19016122364613477,
+      "grad_norm": 0.20870748162269592,
+      "learning_rate": 0.006,
+      "loss": 1.4291,
+      "step": 1840
+    },
+    {
+      "epoch": 0.19119470855725507,
+      "grad_norm": 0.2239631861448288,
+      "learning_rate": 0.006,
+      "loss": 1.4269,
+      "step": 1850
+    },
+    {
+      "epoch": 0.19222819346837536,
+      "grad_norm": 0.24455626308918,
+      "learning_rate": 0.006,
+      "loss": 1.4146,
+      "step": 1860
+    },
+    {
+      "epoch": 0.19326167837949565,
+      "grad_norm": 0.25358930230140686,
+      "learning_rate": 0.006,
+      "loss": 1.4431,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19429516329061597,
+      "grad_norm": 0.1898830235004425,
+      "learning_rate": 0.006,
+      "loss": 1.4222,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19532864820173626,
+      "grad_norm": 0.21167783439159393,
+      "learning_rate": 0.006,
+      "loss": 1.4251,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19636213311285655,
+      "grad_norm": 0.2603727877140045,
+      "learning_rate": 0.006,
+      "loss": 1.426,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19739561802397684,
+      "grad_norm": 0.193504199385643,
+      "learning_rate": 0.006,
+      "loss": 1.4041,
+      "step": 1910
+    },
+    {
+      "epoch": 0.19842910293509713,
+      "grad_norm": 0.22818297147750854,
+      "learning_rate": 0.006,
+      "loss": 1.4204,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19946258784621745,
+      "grad_norm": 0.20980095863342285,
+      "learning_rate": 0.006,
+      "loss": 1.405,
+      "step": 1930
+    },
+    {
+      "epoch": 0.20049607275733775,
+      "grad_norm": 0.17794106900691986,
+      "learning_rate": 0.006,
+      "loss": 1.3871,
+      "step": 1940
+    },
+    {
+      "epoch": 0.20152955766845804,
+      "grad_norm": 0.2572697103023529,
+      "learning_rate": 0.006,
+      "loss": 1.4001,
+      "step": 1950
+    },
+    {
+      "epoch": 0.20256304257957833,
+      "grad_norm": 0.2605222463607788,
+      "learning_rate": 0.006,
+      "loss": 1.4029,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20359652749069865,
+      "grad_norm": 0.27022120356559753,
+      "learning_rate": 0.006,
+      "loss": 1.4009,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20463001240181894,
+      "grad_norm": 0.20254716277122498,
+      "learning_rate": 0.006,
+      "loss": 1.3946,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20566349731293923,
+      "grad_norm": 0.19156356155872345,
+      "learning_rate": 0.006,
+      "loss": 1.3848,
+      "step": 1990
+    },
+    {
+      "epoch": 0.20669698222405952,
+      "grad_norm": 0.182766854763031,
+      "learning_rate": 0.006,
+      "loss": 1.3964,
+      "step": 2000
+    },
+    {
+      "epoch": 0.2077304671351798,
+      "grad_norm": 0.1684897243976593,
+      "learning_rate": 0.006,
+      "loss": 1.3872,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20876395204630013,
+      "grad_norm": 0.19054663181304932,
+      "learning_rate": 0.006,
+      "loss": 1.3817,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20979743695742042,
+      "grad_norm": 0.17960067093372345,
+      "learning_rate": 0.006,
+      "loss": 1.389,
+      "step": 2030
+    },
+    {
+      "epoch": 0.21083092186854072,
+      "grad_norm": 0.23440410196781158,
+      "learning_rate": 0.006,
+      "loss": 1.3707,
+      "step": 2040
+    },
+    {
+      "epoch": 0.211864406779661,
+      "grad_norm": 0.21306714415550232,
+      "learning_rate": 0.006,
+      "loss": 1.3789,
+      "step": 2050
+    },
+    {
+      "epoch": 0.21289789169078133,
+      "grad_norm": 0.24979044497013092,
+      "learning_rate": 0.006,
+      "loss": 1.3755,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21393137660190162,
+      "grad_norm": 0.2036939263343811,
+      "learning_rate": 0.006,
+      "loss": 1.3825,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2149648615130219,
+      "grad_norm": 0.21026568114757538,
+      "learning_rate": 0.006,
+      "loss": 1.3701,
+      "step": 2080
+    },
+    {
+      "epoch": 0.2159983464241422,
+      "grad_norm": 0.18213844299316406,
+      "learning_rate": 0.006,
+      "loss": 1.3835,
+      "step": 2090
+    },
+    {
+      "epoch": 0.2170318313352625,
+      "grad_norm": 0.23943543434143066,
+      "learning_rate": 0.006,
+      "loss": 1.3681,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2180653162463828,
+      "grad_norm": 0.27890509366989136,
+      "learning_rate": 0.006,
+      "loss": 1.3746,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2190988011575031,
+      "grad_norm": 0.1875576227903366,
+      "learning_rate": 0.006,
+      "loss": 1.3833,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2201322860686234,
+      "grad_norm": 0.21590693295001984,
+      "learning_rate": 0.006,
+      "loss": 1.3679,
+      "step": 2130
+    },
+    {
+      "epoch": 0.2211657709797437,
+      "grad_norm": 0.16515956819057465,
+      "learning_rate": 0.006,
+      "loss": 1.3628,
+      "step": 2140
+    },
+    {
+      "epoch": 0.222199255890864,
+      "grad_norm": 0.3273829221725464,
+      "learning_rate": 0.006,
+      "loss": 1.3605,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2232327408019843,
+      "grad_norm": 0.19835293292999268,
+      "learning_rate": 0.006,
+      "loss": 1.3599,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2242662257131046,
+      "grad_norm": 0.20204949378967285,
+      "learning_rate": 0.006,
+      "loss": 1.3548,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22529971062422488,
+      "grad_norm": 0.20704525709152222,
+      "learning_rate": 0.006,
+      "loss": 1.3625,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22633319553534517,
+      "grad_norm": 0.16702768206596375,
+      "learning_rate": 0.006,
+      "loss": 1.3475,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2273666804464655,
+      "grad_norm": 0.2254660427570343,
+      "learning_rate": 0.006,
+      "loss": 1.3508,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22840016535758578,
+      "grad_norm": 0.17011414468288422,
+      "learning_rate": 0.006,
+      "loss": 1.3416,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22943365026870607,
+      "grad_norm": 0.24636393785476685,
+      "learning_rate": 0.006,
+      "loss": 1.3487,
+      "step": 2220
+    },
+    {
+      "epoch": 0.23046713517982637,
+      "grad_norm": 0.2143375277519226,
+      "learning_rate": 0.006,
+      "loss": 1.345,
+      "step": 2230
+    },
+    {
+      "epoch": 0.23150062009094668,
+      "grad_norm": 0.1868714541196823,
+      "learning_rate": 0.006,
+      "loss": 1.3359,
+      "step": 2240
+    },
+    {
+      "epoch": 0.23253410500206698,
+      "grad_norm": 0.1683160364627838,
+      "learning_rate": 0.006,
+      "loss": 1.3399,
+      "step": 2250
+    },
+    {
+      "epoch": 0.23356758991318727,
+      "grad_norm": 0.199097141623497,
+      "learning_rate": 0.006,
+      "loss": 1.3568,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23460107482430756,
+      "grad_norm": 0.18377502262592316,
+      "learning_rate": 0.006,
+      "loss": 1.3488,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23563455973542785,
+      "grad_norm": 0.20199266076087952,
+      "learning_rate": 0.006,
+      "loss": 1.344,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23666804464654817,
+      "grad_norm": 0.271793931722641,
+      "learning_rate": 0.006,
+      "loss": 1.3169,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23770152955766846,
+      "grad_norm": 0.17339195311069489,
+      "learning_rate": 0.006,
+      "loss": 1.3328,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23873501446878875,
+      "grad_norm": 0.17549367249011993,
+      "learning_rate": 0.006,
+      "loss": 1.3354,
+      "step": 2310
+    },
+    {
+      "epoch": 0.23976849937990904,
+      "grad_norm": 0.1800873875617981,
+      "learning_rate": 0.006,
+      "loss": 1.322,
+      "step": 2320
+    },
+    {
+      "epoch": 0.24080198429102936,
+      "grad_norm": 0.22731319069862366,
+      "learning_rate": 0.006,
+      "loss": 1.3235,
+      "step": 2330
+    },
+    {
+      "epoch": 0.24183546920214966,
+      "grad_norm": 0.22910168766975403,
+      "learning_rate": 0.006,
+      "loss": 1.3243,
+      "step": 2340
+    },
+    {
+      "epoch": 0.24286895411326995,
+      "grad_norm": 0.1611885130405426,
+      "learning_rate": 0.006,
+      "loss": 1.3352,
+      "step": 2350
+    },
+    {
+      "epoch": 0.24390243902439024,
+      "grad_norm": 0.19426490366458893,
+      "learning_rate": 0.006,
+      "loss": 1.3091,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24493592393551053,
+      "grad_norm": 0.19100475311279297,
+      "learning_rate": 0.006,
+      "loss": 1.3198,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24596940884663085,
+      "grad_norm": 0.2597349286079407,
+      "learning_rate": 0.006,
+      "loss": 1.302,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24700289375775114,
+      "grad_norm": 0.16532281041145325,
+      "learning_rate": 0.006,
+      "loss": 1.3182,
+      "step": 2390
+    },
+    {
+      "epoch": 0.24803637866887143,
+      "grad_norm": 0.2065669149160385,
+      "learning_rate": 0.006,
+      "loss": 1.3182,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24906986357999172,
+      "grad_norm": 0.1802499145269394,
+      "learning_rate": 0.006,
+      "loss": 1.3206,
+      "step": 2410
+    },
+    {
+      "epoch": 0.25010334849111204,
+      "grad_norm": 0.2926013469696045,
+      "learning_rate": 0.006,
+      "loss": 1.3089,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2511368334022323,
+      "grad_norm": 0.28276434540748596,
+      "learning_rate": 0.006,
+      "loss": 1.307,
+      "step": 2430
+    },
+    {
+      "epoch": 0.2521703183133526,
+      "grad_norm": 0.18156251311302185,
+      "learning_rate": 0.006,
+      "loss": 1.3227,
+      "step": 2440
+    },
+    {
+      "epoch": 0.25320380322447295,
+      "grad_norm": 0.18208877742290497,
+      "learning_rate": 0.006,
+      "loss": 1.3065,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2542372881355932,
+      "grad_norm": 0.18455766141414642,
+      "learning_rate": 0.006,
+      "loss": 1.3019,
+      "step": 2460
+    },
+    {
+      "epoch": 0.25527077304671353,
+      "grad_norm": 0.23398499190807343,
+      "learning_rate": 0.006,
+      "loss": 1.3126,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2563042579578338,
+      "grad_norm": 0.1919931322336197,
+      "learning_rate": 0.006,
+      "loss": 1.2982,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2573377428689541,
+      "grad_norm": 0.24925625324249268,
+      "learning_rate": 0.006,
+      "loss": 1.32,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25837122778007443,
+      "grad_norm": 0.2440209984779358,
+      "learning_rate": 0.006,
+      "loss": 1.2904,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2594047126911947,
+      "grad_norm": 0.16576959192752838,
+      "learning_rate": 0.006,
+      "loss": 1.3065,
+      "step": 2510
+    },
+    {
+      "epoch": 0.260438197602315,
+      "grad_norm": 0.18232128024101257,
+      "learning_rate": 0.006,
+      "loss": 1.3048,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2614716825134353,
+      "grad_norm": 0.23745986819267273,
+      "learning_rate": 0.006,
+      "loss": 1.292,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2625051674245556,
+      "grad_norm": 0.22818538546562195,
+      "learning_rate": 0.006,
+      "loss": 1.2964,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2635386523356759,
+      "grad_norm": 0.1811428666114807,
+      "learning_rate": 0.006,
+      "loss": 1.2968,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2645721372467962,
+      "grad_norm": 0.2562517821788788,
+      "learning_rate": 0.006,
+      "loss": 1.2985,
+      "step": 2560
+    },
+    {
+      "epoch": 0.2656056221579165,
+      "grad_norm": 0.16540707647800446,
+      "learning_rate": 0.006,
+      "loss": 1.2962,
+      "step": 2570
+    },
+    {
+      "epoch": 0.2666391070690368,
+      "grad_norm": 0.17198024690151215,
+      "learning_rate": 0.006,
+      "loss": 1.2835,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2676725919801571,
+      "grad_norm": 0.15511544048786163,
+      "learning_rate": 0.006,
+      "loss": 1.2689,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2687060768912774,
+      "grad_norm": 0.1916809380054474,
+      "learning_rate": 0.006,
+      "loss": 1.2822,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26973956180239766,
+      "grad_norm": 0.17753975093364716,
+      "learning_rate": 0.006,
+      "loss": 1.2843,
+      "step": 2610
+    },
+    {
+      "epoch": 0.270773046713518,
+      "grad_norm": 0.18664464354515076,
+      "learning_rate": 0.006,
+      "loss": 1.2876,
+      "step": 2620
+    },
+    {
+      "epoch": 0.2718065316246383,
+      "grad_norm": 0.283917635679245,
+      "learning_rate": 0.006,
+      "loss": 1.2745,
+      "step": 2630
+    },
+    {
+      "epoch": 0.27284001653575857,
+      "grad_norm": 0.1494530886411667,
+      "learning_rate": 0.006,
+      "loss": 1.2848,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2738735014468789,
+      "grad_norm": 0.2093777358531952,
+      "learning_rate": 0.006,
+      "loss": 1.2786,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27490698635799915,
+      "grad_norm": 0.19870586693286896,
+      "learning_rate": 0.006,
+      "loss": 1.2895,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27594047126911947,
+      "grad_norm": 0.18807923793792725,
+      "learning_rate": 0.006,
+      "loss": 1.2784,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2769739561802398,
+      "grad_norm": 0.16479447484016418,
+      "learning_rate": 0.006,
+      "loss": 1.2815,
+      "step": 2680
+    },
+    {
+      "epoch": 0.27800744109136005,
+      "grad_norm": 0.19209875166416168,
+      "learning_rate": 0.006,
+      "loss": 1.2791,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27904092600248037,
+      "grad_norm": 0.20269426703453064,
+      "learning_rate": 0.006,
+      "loss": 1.2828,
+      "step": 2700
+    },
+    {
+      "epoch": 0.28007441091360064,
+      "grad_norm": 0.1880909502506256,
+      "learning_rate": 0.006,
+      "loss": 1.2786,
+      "step": 2710
+    },
+    {
+      "epoch": 0.28110789582472095,
+      "grad_norm": 0.16020233929157257,
+      "learning_rate": 0.006,
+      "loss": 1.2637,
+      "step": 2720
+    },
+    {
+      "epoch": 0.2821413807358413,
+      "grad_norm": 0.17531338334083557,
+      "learning_rate": 0.006,
+      "loss": 1.264,
+      "step": 2730
+    },
+    {
+      "epoch": 0.28317486564696154,
+      "grad_norm": 0.16038616001605988,
+      "learning_rate": 0.006,
+      "loss": 1.2692,
+      "step": 2740
+    },
+    {
+      "epoch": 0.28420835055808186,
+      "grad_norm": 0.19676966965198517,
+      "learning_rate": 0.006,
+      "loss": 1.271,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2852418354692022,
+      "grad_norm": 0.20371529459953308,
+      "learning_rate": 0.006,
+      "loss": 1.2637,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28627532038032244,
+      "grad_norm": 0.18698766827583313,
+      "learning_rate": 0.006,
+      "loss": 1.2776,
+      "step": 2770
+    },
+    {
+      "epoch": 0.28730880529144276,
+      "grad_norm": 0.20747795701026917,
+      "learning_rate": 0.006,
+      "loss": 1.2573,
+      "step": 2780
+    },
+    {
+      "epoch": 0.288342290202563,
+      "grad_norm": 0.2328963726758957,
+      "learning_rate": 0.006,
+      "loss": 1.2568,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28937577511368334,
+      "grad_norm": 0.1655196100473404,
+      "learning_rate": 0.006,
+      "loss": 1.2614,
+      "step": 2800
+    },
+    {
+      "epoch": 0.29040926002480366,
+      "grad_norm": 0.1615329086780548,
+      "learning_rate": 0.006,
+      "loss": 1.2712,
+      "step": 2810
+    },
+    {
+      "epoch": 0.2914427449359239,
+      "grad_norm": 0.15835103392601013,
+      "learning_rate": 0.006,
+      "loss": 1.2524,
+      "step": 2820
+    },
+    {
+      "epoch": 0.29247622984704424,
+      "grad_norm": 0.20250938832759857,
+      "learning_rate": 0.006,
+      "loss": 1.264,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2935097147581645,
+      "grad_norm": 0.17166636884212494,
+      "learning_rate": 0.006,
+      "loss": 1.2563,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2945431996692848,
+      "grad_norm": 0.2327214777469635,
+      "learning_rate": 0.006,
+      "loss": 1.2546,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29557668458040515,
+      "grad_norm": 0.19193734228610992,
+      "learning_rate": 0.006,
+      "loss": 1.2647,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2966101694915254,
+      "grad_norm": 0.21454569697380066,
+      "learning_rate": 0.006,
+      "loss": 1.2573,
+      "step": 2870
+    },
+    {
+      "epoch": 0.29764365440264573,
+      "grad_norm": 0.18797141313552856,
+      "learning_rate": 0.006,
+      "loss": 1.2661,
+      "step": 2880
+    },
+    {
+      "epoch": 0.298677139313766,
+      "grad_norm": 0.16613776981830597,
+      "learning_rate": 0.006,
+      "loss": 1.2593,
+      "step": 2890
+    },
+    {
+      "epoch": 0.2997106242248863,
+      "grad_norm": 0.15257228910923004,
+      "learning_rate": 0.006,
+      "loss": 1.2534,
+      "step": 2900
+    },
+    {
+      "epoch": 0.30074410913600663,
+      "grad_norm": 0.1718466728925705,
+      "learning_rate": 0.006,
+      "loss": 1.2526,
+      "step": 2910
+    },
+    {
+      "epoch": 0.3017775940471269,
+      "grad_norm": 0.17519408464431763,
+      "learning_rate": 0.006,
+      "loss": 1.236,
+      "step": 2920
+    },
+    {
+      "epoch": 0.3028110789582472,
+      "grad_norm": 0.1739313006401062,
+      "learning_rate": 0.006,
+      "loss": 1.2416,
+      "step": 2930
+    },
+    {
+      "epoch": 0.30384456386936753,
+      "grad_norm": 0.2000485360622406,
+      "learning_rate": 0.006,
+      "loss": 1.2525,
+      "step": 2940
+    },
+    {
+      "epoch": 0.3048780487804878,
+      "grad_norm": 0.21297138929367065,
+      "learning_rate": 0.006,
+      "loss": 1.2312,
+      "step": 2950
+    },
+    {
+      "epoch": 0.3059115336916081,
+      "grad_norm": 0.1601925939321518,
+      "learning_rate": 0.006,
+      "loss": 1.2575,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3069450186027284,
+      "grad_norm": 0.22686167061328888,
+      "learning_rate": 0.006,
+      "loss": 1.2421,
+      "step": 2970
+    },
+    {
+      "epoch": 0.3079785035138487,
+      "grad_norm": 0.2552824318408966,
+      "learning_rate": 0.006,
+      "loss": 1.255,
+      "step": 2980
+    },
+    {
+      "epoch": 0.309011988424969,
+      "grad_norm": 0.1520908921957016,
+      "learning_rate": 0.006,
+      "loss": 1.2335,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3100454733360893,
+      "grad_norm": 0.18067720532417297,
+      "learning_rate": 0.006,
+      "loss": 1.2278,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3110789582472096,
+      "grad_norm": 0.2269776612520218,
+      "learning_rate": 0.006,
+      "loss": 1.2401,
+      "step": 3010
+    },
+    {
+      "epoch": 0.31211244315832987,
+      "grad_norm": 0.17168264091014862,
+      "learning_rate": 0.006,
+      "loss": 1.245,
+      "step": 3020
+    },
+    {
+      "epoch": 0.3131459280694502,
+      "grad_norm": 0.20345386862754822,
+      "learning_rate": 0.006,
+      "loss": 1.2286,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3141794129805705,
+      "grad_norm": 0.17593909800052643,
+      "learning_rate": 0.006,
+      "loss": 1.2439,
+      "step": 3040
+    },
+    {
+      "epoch": 0.31521289789169077,
+      "grad_norm": 0.1948511153459549,
+      "learning_rate": 0.006,
+      "loss": 1.2513,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3162463828028111,
+      "grad_norm": 0.1758977472782135,
+      "learning_rate": 0.006,
+      "loss": 1.2428,
+      "step": 3060
+    },
+    {
+      "epoch": 0.31727986771393135,
+      "grad_norm": 0.14619344472885132,
+      "learning_rate": 0.006,
+      "loss": 1.2295,
+      "step": 3070
+    },
+    {
+      "epoch": 0.31831335262505167,
+      "grad_norm": 0.15746112167835236,
+      "learning_rate": 0.006,
+      "loss": 1.2133,
+      "step": 3080
+    },
+    {
+      "epoch": 0.319346837536172,
+      "grad_norm": 0.14768971502780914,
+      "learning_rate": 0.006,
+      "loss": 1.2234,
+      "step": 3090
+    },
+    {
+      "epoch": 0.32038032244729225,
+      "grad_norm": 0.15294450521469116,
+      "learning_rate": 0.006,
+      "loss": 1.2255,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3214138073584126,
+      "grad_norm": 0.18775102496147156,
+      "learning_rate": 0.006,
+      "loss": 1.2378,
+      "step": 3110
+    },
+    {
+      "epoch": 0.3224472922695329,
+      "grad_norm": 0.15432560443878174,
+      "learning_rate": 0.006,
+      "loss": 1.2359,
+      "step": 3120
+    },
+    {
+      "epoch": 0.32348077718065316,
+      "grad_norm": 0.17162322998046875,
+      "learning_rate": 0.006,
+      "loss": 1.2257,
+      "step": 3130
+    },
+    {
+      "epoch": 0.3245142620917735,
+      "grad_norm": 0.21141092479228973,
+      "learning_rate": 0.006,
+      "loss": 1.2301,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32554774700289374,
+      "grad_norm": 0.1810036152601242,
+      "learning_rate": 0.006,
+      "loss": 1.2204,
+      "step": 3150
+    },
+    {
+      "epoch": 0.32658123191401406,
+      "grad_norm": 0.20127083361148834,
+      "learning_rate": 0.006,
+      "loss": 1.2263,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3276147168251344,
+      "grad_norm": 0.1795072853565216,
+      "learning_rate": 0.006,
+      "loss": 1.2224,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32864820173625464,
+      "grad_norm": 0.26130935549736023,
+      "learning_rate": 0.006,
+      "loss": 1.2236,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32968168664737496,
+      "grad_norm": 0.17055651545524597,
+      "learning_rate": 0.006,
+      "loss": 1.2229,
+      "step": 3190
+    },
+    {
+      "epoch": 0.3307151715584952,
+      "grad_norm": 0.24257224798202515,
+      "learning_rate": 0.006,
+      "loss": 1.218,
+      "step": 3200
+    },
+    {
+      "epoch": 0.33174865646961554,
+      "grad_norm": 0.1574934422969818,
+      "learning_rate": 0.006,
+      "loss": 1.2248,
+      "step": 3210
+    },
+    {
+      "epoch": 0.33278214138073586,
+      "grad_norm": 0.18338268995285034,
+      "learning_rate": 0.006,
+      "loss": 1.211,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3338156262918561,
+      "grad_norm": 0.23949076235294342,
+      "learning_rate": 0.006,
+      "loss": 1.2107,
+      "step": 3230
+    },
+    {
+      "epoch": 0.33484911120297645,
+      "grad_norm": 0.23181311786174774,
+      "learning_rate": 0.006,
+      "loss": 1.2157,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3358825961140967,
+      "grad_norm": 0.1727987825870514,
+      "learning_rate": 0.006,
+      "loss": 1.2113,
+      "step": 3250
+    },
+    {
+      "epoch": 0.33691608102521703,
+      "grad_norm": 0.17050251364707947,
+      "learning_rate": 0.006,
+      "loss": 1.2196,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33794956593633735,
+      "grad_norm": 0.1836351752281189,
+      "learning_rate": 0.006,
+      "loss": 1.2164,
+      "step": 3270
+    },
+    {
+      "epoch": 0.3389830508474576,
+      "grad_norm": 0.16653425991535187,
+      "learning_rate": 0.006,
+      "loss": 1.2108,
+      "step": 3280
+    },
+    {
+      "epoch": 0.34001653575857793,
+      "grad_norm": 0.19799716770648956,
+      "learning_rate": 0.006,
+      "loss": 1.2072,
+      "step": 3290
+    },
+    {
+      "epoch": 0.3410500206696982,
+      "grad_norm": 0.15341606736183167,
+      "learning_rate": 0.006,
+      "loss": 1.2157,
+      "step": 3300
+    },
+    {
+      "epoch": 0.3420835055808185,
+      "grad_norm": 0.15945138037204742,
+      "learning_rate": 0.006,
+      "loss": 1.2065,
+      "step": 3310
+    },
+    {
+      "epoch": 0.34311699049193883,
+      "grad_norm": 0.17721377313137054,
+      "learning_rate": 0.006,
+      "loss": 1.2056,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3441504754030591,
+      "grad_norm": 0.1358070969581604,
+      "learning_rate": 0.006,
+      "loss": 1.2171,
+      "step": 3330
+    },
+    {
+      "epoch": 0.3451839603141794,
+      "grad_norm": 0.17187251150608063,
+      "learning_rate": 0.006,
+      "loss": 1.2046,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34621744522529974,
+      "grad_norm": 0.14732858538627625,
+      "learning_rate": 0.006,
+      "loss": 1.2025,
+      "step": 3350
+    },
+    {
+      "epoch": 0.34725093013642,
+      "grad_norm": 0.224739208817482,
+      "learning_rate": 0.006,
+      "loss": 1.2213,
+      "step": 3360
+    },
+    {
+      "epoch": 0.3482844150475403,
+      "grad_norm": 0.20575636625289917,
+      "learning_rate": 0.006,
+      "loss": 1.1922,
+      "step": 3370
+    },
+    {
+      "epoch": 0.3493178999586606,
+      "grad_norm": 0.15750053524971008,
+      "learning_rate": 0.006,
+      "loss": 1.209,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3503513848697809,
+      "grad_norm": 0.27751949429512024,
+      "learning_rate": 0.006,
+      "loss": 1.2079,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3513848697809012,
+      "grad_norm": 0.14315839111804962,
+      "learning_rate": 0.006,
+      "loss": 1.1934,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3524183546920215,
+      "grad_norm": 0.15051007270812988,
+      "learning_rate": 0.006,
+      "loss": 1.2067,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3534518396031418,
+      "grad_norm": 0.15940986573696136,
+      "learning_rate": 0.006,
+      "loss": 1.2097,
+      "step": 3420
+    },
+    {
+      "epoch": 0.35448532451426207,
+      "grad_norm": 0.1412818878889084,
+      "learning_rate": 0.006,
+      "loss": 1.1959,
+      "step": 3430
+    },
+    {
+      "epoch": 0.3555188094253824,
+      "grad_norm": 0.1425037980079651,
+      "learning_rate": 0.006,
+      "loss": 1.1901,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3565522943365027,
+      "grad_norm": 0.15428400039672852,
+      "learning_rate": 0.006,
+      "loss": 1.2037,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35758577924762297,
+      "grad_norm": 0.1565396934747696,
+      "learning_rate": 0.006,
+      "loss": 1.2066,
+      "step": 3460
+    },
+    {
+      "epoch": 0.3586192641587433,
+      "grad_norm": 0.1757258176803589,
+      "learning_rate": 0.006,
+      "loss": 1.1947,
+      "step": 3470
+    },
+    {
+      "epoch": 0.35965274906986355,
+      "grad_norm": 0.23932603001594543,
+      "learning_rate": 0.006,
+      "loss": 1.1993,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3606862339809839,
+      "grad_norm": 0.17783917486667633,
+      "learning_rate": 0.006,
+      "loss": 1.1881,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3617197188921042,
+      "grad_norm": 0.13933712244033813,
+      "learning_rate": 0.006,
+      "loss": 1.1969,
+      "step": 3500
+    },
+    {
+      "epoch": 0.36275320380322446,
+      "grad_norm": 0.13872799277305603,
+      "learning_rate": 0.006,
+      "loss": 1.1958,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3637866887143448,
+      "grad_norm": 0.18496429920196533,
+      "learning_rate": 0.006,
+      "loss": 1.1978,
+      "step": 3520
+    },
+    {
+      "epoch": 0.3648201736254651,
+      "grad_norm": 0.2182753086090088,
+      "learning_rate": 0.006,
+      "loss": 1.1957,
+      "step": 3530
+    },
+    {
+      "epoch": 0.36585365853658536,
+      "grad_norm": 0.15641410648822784,
+      "learning_rate": 0.006,
+      "loss": 1.1828,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3668871434477057,
+      "grad_norm": 0.15196652710437775,
+      "learning_rate": 0.006,
+      "loss": 1.1901,
+      "step": 3550
+    },
+    {
+      "epoch": 0.36792062835882594,
+      "grad_norm": 0.15420852601528168,
+      "learning_rate": 0.006,
+      "loss": 1.1966,
+      "step": 3560
+    },
+    {
+      "epoch": 0.36895411326994626,
+      "grad_norm": 0.1419685333967209,
+      "learning_rate": 0.006,
+      "loss": 1.1835,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3699875981810666,
+      "grad_norm": 0.20714856684207916,
+      "learning_rate": 0.006,
+      "loss": 1.1969,
+      "step": 3580
+    },
+    {
+      "epoch": 0.37102108309218684,
+      "grad_norm": 0.14852945506572723,
+      "learning_rate": 0.006,
+      "loss": 1.1979,
+      "step": 3590
+    },
+    {
+      "epoch": 0.37205456800330716,
+      "grad_norm": 0.1534539759159088,
+      "learning_rate": 0.006,
+      "loss": 1.1814,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3730880529144274,
+      "grad_norm": 0.1536993384361267,
+      "learning_rate": 0.006,
+      "loss": 1.1935,
+      "step": 3610
+    },
+    {
+      "epoch": 0.37412153782554775,
+      "grad_norm": 0.14678268134593964,
+      "learning_rate": 0.006,
+      "loss": 1.1879,
+      "step": 3620
+    },
+    {
+      "epoch": 0.37515502273666806,
+      "grad_norm": 0.15194503962993622,
+      "learning_rate": 0.006,
+      "loss": 1.1851,
+      "step": 3630
+    },
+    {
+      "epoch": 0.37618850764778833,
+      "grad_norm": 0.1853533834218979,
+      "learning_rate": 0.006,
+      "loss": 1.177,
+      "step": 3640
+    },
+    {
+      "epoch": 0.37722199255890865,
+      "grad_norm": 0.19894924759864807,
+      "learning_rate": 0.006,
+      "loss": 1.1792,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3782554774700289,
+      "grad_norm": 0.14289256930351257,
+      "learning_rate": 0.006,
+      "loss": 1.1916,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37928896238114923,
+      "grad_norm": 0.15364837646484375,
+      "learning_rate": 0.006,
+      "loss": 1.1773,
+      "step": 3670
+    },
+    {
+      "epoch": 0.38032244729226955,
+      "grad_norm": 0.26057279109954834,
+      "learning_rate": 0.006,
+      "loss": 1.1772,
+      "step": 3680
+    },
+    {
+      "epoch": 0.3813559322033898,
+      "grad_norm": 0.14714686572551727,
+      "learning_rate": 0.006,
+      "loss": 1.183,
+      "step": 3690
+    },
+    {
+      "epoch": 0.38238941711451013,
+      "grad_norm": 0.13475820422172546,
+      "learning_rate": 0.006,
+      "loss": 1.1734,
+      "step": 3700
+    },
+    {
+      "epoch": 0.38342290202563045,
+      "grad_norm": 0.169892817735672,
+      "learning_rate": 0.006,
+      "loss": 1.1833,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3844563869367507,
+      "grad_norm": 0.15819117426872253,
+      "learning_rate": 0.006,
+      "loss": 1.1693,
+      "step": 3720
+    },
+    {
+      "epoch": 0.38548987184787104,
+      "grad_norm": 0.2074589729309082,
+      "learning_rate": 0.006,
+      "loss": 1.1773,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3865233567589913,
+      "grad_norm": 0.1474793255329132,
+      "learning_rate": 0.006,
+      "loss": 1.1746,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3875568416701116,
+      "grad_norm": 0.20919297635555267,
+      "learning_rate": 0.006,
+      "loss": 1.1644,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38859032658123194,
+      "grad_norm": 0.1802784502506256,
+      "learning_rate": 0.006,
+      "loss": 1.1785,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3896238114923522,
+      "grad_norm": 0.13777178525924683,
+      "learning_rate": 0.006,
+      "loss": 1.1762,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3906572964034725,
+      "grad_norm": 0.19703027606010437,
+      "learning_rate": 0.006,
+      "loss": 1.1715,
+      "step": 3780
+    },
+    {
+      "epoch": 0.3916907813145928,
+      "grad_norm": 0.13620315492153168,
+      "learning_rate": 0.006,
+      "loss": 1.1729,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3927242662257131,
+      "grad_norm": 0.19151124358177185,
+      "learning_rate": 0.006,
+      "loss": 1.165,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3937577511368334,
+      "grad_norm": 0.14211703836917877,
+      "learning_rate": 0.006,
+      "loss": 1.1705,
+      "step": 3810
+    },
+    {
+      "epoch": 0.3947912360479537,
+      "grad_norm": 0.13506080210208893,
+      "learning_rate": 0.006,
+      "loss": 1.1656,
+      "step": 3820
+    },
+    {
+      "epoch": 0.395824720959074,
+      "grad_norm": 0.15365293622016907,
+      "learning_rate": 0.006,
+      "loss": 1.1642,
+      "step": 3830
+    },
+    {
+      "epoch": 0.39685820587019427,
+      "grad_norm": 0.1530192494392395,
+      "learning_rate": 0.006,
+      "loss": 1.1621,
+      "step": 3840
+    },
+    {
+      "epoch": 0.3978916907813146,
+      "grad_norm": 0.1344013661146164,
+      "learning_rate": 0.006,
+      "loss": 1.1694,
+      "step": 3850
+    },
+    {
+      "epoch": 0.3989251756924349,
+      "grad_norm": 0.1469976007938385,
+      "learning_rate": 0.006,
+      "loss": 1.1706,
+      "step": 3860
+    },
+    {
+      "epoch": 0.39995866060355517,
+      "grad_norm": 0.14535875618457794,
+      "learning_rate": 0.006,
+      "loss": 1.1641,
+      "step": 3870
+    },
+    {
+      "epoch": 0.4009921455146755,
+      "grad_norm": 0.22596335411071777,
+      "learning_rate": 0.006,
+      "loss": 1.1835,
+      "step": 3880
+    },
+    {
+      "epoch": 0.4020256304257958,
+      "grad_norm": 0.214060440659523,
+      "learning_rate": 0.006,
+      "loss": 1.1688,
+      "step": 3890
+    },
+    {
+      "epoch": 0.4030591153369161,
+      "grad_norm": 0.14078359305858612,
+      "learning_rate": 0.006,
+      "loss": 1.158,
+      "step": 3900
+    },
+    {
+      "epoch": 0.4040926002480364,
+      "grad_norm": 0.15023833513259888,
+      "learning_rate": 0.006,
+      "loss": 1.1599,
+      "step": 3910
+    },
+    {
+      "epoch": 0.40512608515915666,
+      "grad_norm": 0.13650792837142944,
+      "learning_rate": 0.006,
+      "loss": 1.1512,
+      "step": 3920
+    },
+    {
+      "epoch": 0.406159570070277,
+      "grad_norm": 0.1389831155538559,
+      "learning_rate": 0.006,
+      "loss": 1.1624,
+      "step": 3930
+    },
+    {
+      "epoch": 0.4071930549813973,
+      "grad_norm": 0.1548847109079361,
+      "learning_rate": 0.006,
+      "loss": 1.1822,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40822653989251756,
+      "grad_norm": 0.14826175570487976,
+      "learning_rate": 0.006,
+      "loss": 1.1539,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4092600248036379,
+      "grad_norm": 0.27681705355644226,
+      "learning_rate": 0.006,
+      "loss": 1.1543,
+      "step": 3960
+    },
+    {
+      "epoch": 0.41029350971475814,
+      "grad_norm": 0.17979800701141357,
+      "learning_rate": 0.006,
+      "loss": 1.1583,
+      "step": 3970
+    },
+    {
+      "epoch": 0.41132699462587846,
+      "grad_norm": 0.14845605194568634,
+      "learning_rate": 0.006,
+      "loss": 1.1559,
+      "step": 3980
+    },
+    {
+      "epoch": 0.4123604795369988,
+      "grad_norm": 0.18015886843204498,
+      "learning_rate": 0.006,
+      "loss": 1.1663,
+      "step": 3990
+    },
+    {
+      "epoch": 0.41339396444811904,
+      "grad_norm": 0.2010691910982132,
+      "learning_rate": 0.006,
+      "loss": 1.1572,
+      "step": 4000
+    },
+    {
+      "epoch": 0.41442744935923936,
+      "grad_norm": 0.14041297137737274,
+      "learning_rate": 0.006,
+      "loss": 1.1621,
+      "step": 4010
+    },
+    {
+      "epoch": 0.4154609342703596,
+      "grad_norm": 0.13786040246486664,
+      "learning_rate": 0.006,
+      "loss": 1.157,
+      "step": 4020
+    },
+    {
+      "epoch": 0.41649441918147995,
+      "grad_norm": 0.17166124284267426,
+      "learning_rate": 0.006,
+      "loss": 1.178,
+      "step": 4030
+    },
+    {
+      "epoch": 0.41752790409260027,
+      "grad_norm": 0.13718323409557343,
+      "learning_rate": 0.006,
+      "loss": 1.1514,
+      "step": 4040
+    },
+    {
+      "epoch": 0.41856138900372053,
+      "grad_norm": 0.15878678858280182,
+      "learning_rate": 0.006,
+      "loss": 1.1718,
+      "step": 4050
+    },
+    {
+      "epoch": 0.41959487391484085,
+      "grad_norm": 0.14852339029312134,
+      "learning_rate": 0.006,
+      "loss": 1.1559,
+      "step": 4060
+    },
+    {
+      "epoch": 0.42062835882596117,
+      "grad_norm": 0.17088137567043304,
+      "learning_rate": 0.006,
+      "loss": 1.1527,
+      "step": 4070
+    },
+    {
+      "epoch": 0.42166184373708143,
+      "grad_norm": 0.15074481070041656,
+      "learning_rate": 0.006,
+      "loss": 1.1519,
+      "step": 4080
+    },
+    {
+      "epoch": 0.42269532864820175,
+      "grad_norm": 0.13103771209716797,
+      "learning_rate": 0.006,
+      "loss": 1.1475,
+      "step": 4090
+    },
+    {
+      "epoch": 0.423728813559322,
+      "grad_norm": 0.1428011655807495,
+      "learning_rate": 0.006,
+      "loss": 1.1528,
+      "step": 4100
+    },
+    {
+      "epoch": 0.42476229847044233,
+      "grad_norm": 0.14280404150485992,
+      "learning_rate": 0.006,
+      "loss": 1.1542,
+      "step": 4110
+    },
+    {
+      "epoch": 0.42579578338156265,
+      "grad_norm": 0.14426454901695251,
+      "learning_rate": 0.006,
+      "loss": 1.1632,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4268292682926829,
+      "grad_norm": 0.13187375664710999,
+      "learning_rate": 0.006,
+      "loss": 1.1564,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42786275320380324,
+      "grad_norm": 0.1555539071559906,
+      "learning_rate": 0.006,
+      "loss": 1.1497,
+      "step": 4140
+    },
+    {
+      "epoch": 0.4288962381149235,
+      "grad_norm": 0.14076194167137146,
+      "learning_rate": 0.006,
+      "loss": 1.1431,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4299297230260438,
+      "grad_norm": 0.14553508162498474,
+      "learning_rate": 0.006,
+      "loss": 1.1444,
+      "step": 4160
+    },
+    {
+      "epoch": 0.43096320793716414,
+      "grad_norm": 0.14131340384483337,
+      "learning_rate": 0.006,
+      "loss": 1.1504,
+      "step": 4170
+    },
+    {
+      "epoch": 0.4319966928482844,
+      "grad_norm": 0.1349923461675644,
+      "learning_rate": 0.006,
+      "loss": 1.1499,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4330301777594047,
+      "grad_norm": 0.1801183521747589,
+      "learning_rate": 0.006,
+      "loss": 1.1458,
+      "step": 4190
+    },
+    {
+      "epoch": 0.434063662670525,
+      "grad_norm": 0.16793182492256165,
+      "learning_rate": 0.006,
+      "loss": 1.1527,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4350971475816453,
+      "grad_norm": 0.19141018390655518,
+      "learning_rate": 0.006,
+      "loss": 1.1451,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4361306324927656,
+      "grad_norm": 0.1311262547969818,
+      "learning_rate": 0.006,
+      "loss": 1.1433,
+      "step": 4220
+    },
+    {
+      "epoch": 0.4371641174038859,
+      "grad_norm": 0.1293700635433197,
+      "learning_rate": 0.006,
+      "loss": 1.1251,
+      "step": 4230
+    },
+    {
+      "epoch": 0.4381976023150062,
+      "grad_norm": 0.13121871650218964,
+      "learning_rate": 0.006,
+      "loss": 1.1392,
+      "step": 4240
+    },
+    {
+      "epoch": 0.43923108722612647,
+      "grad_norm": 0.13999488949775696,
+      "learning_rate": 0.006,
+      "loss": 1.135,
+      "step": 4250
+    },
+    {
+      "epoch": 0.4402645721372468,
+      "grad_norm": 0.14969255030155182,
+      "learning_rate": 0.006,
+      "loss": 1.1399,
+      "step": 4260
+    },
+    {
+      "epoch": 0.4412980570483671,
+      "grad_norm": 0.14516979455947876,
+      "learning_rate": 0.006,
+      "loss": 1.1385,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4423315419594874,
+      "grad_norm": 0.13377705216407776,
+      "learning_rate": 0.006,
+      "loss": 1.1409,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4433650268706077,
+      "grad_norm": 0.1788671314716339,
+      "learning_rate": 0.006,
+      "loss": 1.1368,
+      "step": 4290
+    },
+    {
+      "epoch": 0.444398511781728,
+      "grad_norm": 0.1459660530090332,
+      "learning_rate": 0.006,
+      "loss": 1.1399,
+      "step": 4300
+    },
+    {
+      "epoch": 0.4454319966928483,
+      "grad_norm": 0.14903658628463745,
+      "learning_rate": 0.006,
+      "loss": 1.1347,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4464654816039686,
+      "grad_norm": 0.16043029725551605,
+      "learning_rate": 0.006,
+      "loss": 1.1395,
+      "step": 4320
+    },
+    {
+      "epoch": 0.44749896651508886,
+      "grad_norm": 0.1458473652601242,
+      "learning_rate": 0.006,
+      "loss": 1.1442,
+      "step": 4330
+    },
+    {
+      "epoch": 0.4485324514262092,
+      "grad_norm": 0.1839747577905655,
+      "learning_rate": 0.006,
+      "loss": 1.1278,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4495659363373295,
+      "grad_norm": 0.14207588136196136,
+      "learning_rate": 0.006,
+      "loss": 1.1355,
+      "step": 4350
+    },
+    {
+      "epoch": 0.45059942124844976,
+      "grad_norm": 0.1426732838153839,
+      "learning_rate": 0.006,
+      "loss": 1.1345,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4516329061595701,
+      "grad_norm": 0.19898781180381775,
+      "learning_rate": 0.006,
+      "loss": 1.128,
+      "step": 4370
+    },
+    {
+      "epoch": 0.45266639107069034,
+      "grad_norm": 0.1501811444759369,
+      "learning_rate": 0.006,
+      "loss": 1.1294,
+      "step": 4380
+    },
+    {
+      "epoch": 0.45369987598181066,
+      "grad_norm": 0.13546693325042725,
+      "learning_rate": 0.006,
+      "loss": 1.1459,
+      "step": 4390
+    },
+    {
+      "epoch": 0.454733360892931,
+      "grad_norm": 0.1429995894432068,
+      "learning_rate": 0.006,
+      "loss": 1.1401,
+      "step": 4400
+    },
+    {
+      "epoch": 0.45576684580405125,
+      "grad_norm": 0.16272728145122528,
+      "learning_rate": 0.006,
+      "loss": 1.1388,
+      "step": 4410
+    },
+    {
+      "epoch": 0.45680033071517157,
+      "grad_norm": 0.13995374739170074,
+      "learning_rate": 0.006,
+      "loss": 1.1348,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45783381562629183,
+      "grad_norm": 0.13635413348674774,
+      "learning_rate": 0.006,
+      "loss": 1.13,
+      "step": 4430
+    },
+    {
+      "epoch": 0.45886730053741215,
+      "grad_norm": 0.12863661348819733,
+      "learning_rate": 0.006,
+      "loss": 1.1289,
+      "step": 4440
+    },
+    {
+      "epoch": 0.45990078544853247,
+      "grad_norm": 0.14227396249771118,
+      "learning_rate": 0.006,
+      "loss": 1.1304,
+      "step": 4450
+    },
+    {
+      "epoch": 0.46093427035965273,
+      "grad_norm": 0.21000246703624725,
+      "learning_rate": 0.006,
+      "loss": 1.132,
+      "step": 4460
+    },
+    {
+      "epoch": 0.46196775527077305,
+      "grad_norm": 0.15686342120170593,
+      "learning_rate": 0.006,
+      "loss": 1.1456,
+      "step": 4470
+    },
+    {
+      "epoch": 0.46300124018189337,
+      "grad_norm": 0.13623683154582977,
+      "learning_rate": 0.006,
+      "loss": 1.1229,
+      "step": 4480
+    },
+    {
+      "epoch": 0.46403472509301363,
+      "grad_norm": 0.13092273473739624,
+      "learning_rate": 0.006,
+      "loss": 1.1199,
+      "step": 4490
+    },
+    {
+      "epoch": 0.46506821000413395,
+      "grad_norm": 0.13690875470638275,
+      "learning_rate": 0.006,
+      "loss": 1.1095,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4661016949152542,
+      "grad_norm": 0.12770096957683563,
+      "learning_rate": 0.006,
+      "loss": 1.1251,
+      "step": 4510
+    },
+    {
+      "epoch": 0.46713517982637454,
+      "grad_norm": 0.12008727341890335,
+      "learning_rate": 0.006,
+      "loss": 1.1151,
+      "step": 4520
+    },
+    {
+      "epoch": 0.46816866473749486,
+      "grad_norm": 0.13290061056613922,
+      "learning_rate": 0.006,
+      "loss": 1.1278,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4692021496486151,
+      "grad_norm": 0.16319867968559265,
+      "learning_rate": 0.006,
+      "loss": 1.1261,
+      "step": 4540
+    },
+    {
+      "epoch": 0.47023563455973544,
+      "grad_norm": 0.13294318318367004,
+      "learning_rate": 0.006,
+      "loss": 1.1314,
+      "step": 4550
+    },
+    {
+      "epoch": 0.4712691194708557,
+      "grad_norm": 0.15630632638931274,
+      "learning_rate": 0.006,
+      "loss": 1.1423,
+      "step": 4560
+    },
+    {
+      "epoch": 0.472302604381976,
+      "grad_norm": 0.16916899383068085,
+      "learning_rate": 0.006,
+      "loss": 1.1385,
+      "step": 4570
+    },
+    {
+      "epoch": 0.47333608929309634,
+      "grad_norm": 0.2240544855594635,
+      "learning_rate": 0.006,
+      "loss": 1.1307,
+      "step": 4580
+    },
+    {
+      "epoch": 0.4743695742042166,
+      "grad_norm": 0.12742292881011963,
+      "learning_rate": 0.006,
+      "loss": 1.1278,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4754030591153369,
+      "grad_norm": 0.13861480355262756,
+      "learning_rate": 0.006,
+      "loss": 1.1248,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4764365440264572,
+      "grad_norm": 0.16158758103847504,
+      "learning_rate": 0.006,
+      "loss": 1.1437,
+      "step": 4610
+    },
+    {
+      "epoch": 0.4774700289375775,
+      "grad_norm": 0.13849808275699615,
+      "learning_rate": 0.006,
+      "loss": 1.1345,
+      "step": 4620
+    },
+    {
+      "epoch": 0.4785035138486978,
+      "grad_norm": 0.1381804645061493,
+      "learning_rate": 0.006,
+      "loss": 1.124,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4795369987598181,
+      "grad_norm": 0.1629112809896469,
+      "learning_rate": 0.006,
+      "loss": 1.1198,
+      "step": 4640
+    },
+    {
+      "epoch": 0.4805704836709384,
+      "grad_norm": 0.1835983693599701,
+      "learning_rate": 0.006,
+      "loss": 1.1198,
+      "step": 4650
+    },
+    {
+      "epoch": 0.48160396858205873,
+      "grad_norm": 0.13495475053787231,
+      "learning_rate": 0.006,
+      "loss": 1.1266,
+      "step": 4660
+    },
+    {
+      "epoch": 0.482637453493179,
+      "grad_norm": 0.15339604020118713,
+      "learning_rate": 0.006,
+      "loss": 1.1111,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4836709384042993,
+      "grad_norm": 0.14230936765670776,
+      "learning_rate": 0.006,
+      "loss": 1.1002,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4847044233154196,
+      "grad_norm": 0.13988140225410461,
+      "learning_rate": 0.006,
+      "loss": 1.1132,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4857379082265399,
+      "grad_norm": 0.17499862611293793,
+      "learning_rate": 0.006,
+      "loss": 1.1223,
+      "step": 4700
+    },
+    {
+      "epoch": 0.4867713931376602,
+      "grad_norm": 0.15850752592086792,
+      "learning_rate": 0.006,
+      "loss": 1.1199,
+      "step": 4710
+    },
+    {
+      "epoch": 0.4878048780487805,
+      "grad_norm": 0.12476572394371033,
+      "learning_rate": 0.006,
+      "loss": 1.1252,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4888383629599008,
+      "grad_norm": 0.14426189661026,
+      "learning_rate": 0.006,
+      "loss": 1.1132,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48987184787102106,
+      "grad_norm": 0.14036618173122406,
+      "learning_rate": 0.006,
+      "loss": 1.1226,
+      "step": 4740
+    },
+    {
+      "epoch": 0.4909053327821414,
+      "grad_norm": 0.1434011608362198,
+      "learning_rate": 0.006,
+      "loss": 1.1123,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4919388176932617,
+      "grad_norm": 0.19470463693141937,
+      "learning_rate": 0.006,
+      "loss": 1.1321,
+      "step": 4760
+    },
+    {
+      "epoch": 0.49297230260438196,
+      "grad_norm": 0.12864422798156738,
+      "learning_rate": 0.006,
+      "loss": 1.1176,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4940057875155023,
+      "grad_norm": 0.16909734904766083,
+      "learning_rate": 0.006,
+      "loss": 1.1164,
+      "step": 4780
+    },
+    {
+      "epoch": 0.49503927242662255,
+      "grad_norm": 0.14817242324352264,
+      "learning_rate": 0.006,
+      "loss": 1.1086,
+      "step": 4790
+    },
+    {
+      "epoch": 0.49607275733774286,
+      "grad_norm": 0.16644009947776794,
+      "learning_rate": 0.006,
+      "loss": 1.1142,
+      "step": 4800
+    },
+    {
+      "epoch": 0.4971062422488632,
+      "grad_norm": 0.16777795553207397,
+      "learning_rate": 0.006,
+      "loss": 1.1077,
+      "step": 4810
+    },
+    {
+      "epoch": 0.49813972715998345,
+      "grad_norm": 0.1416415572166443,
+      "learning_rate": 0.006,
+      "loss": 1.122,
+      "step": 4820
+    },
+    {
+      "epoch": 0.49917321207110377,
+      "grad_norm": 0.1440482884645462,
+      "learning_rate": 0.006,
+      "loss": 1.1129,
+      "step": 4830
+    },
+    {
+      "epoch": 0.5002066969822241,
+      "grad_norm": 0.1519850343465805,
+      "learning_rate": 0.006,
+      "loss": 1.1144,
+      "step": 4840
+    },
+    {
+      "epoch": 0.5012401818933444,
+      "grad_norm": 0.24164853990077972,
+      "learning_rate": 0.006,
+      "loss": 1.1235,
+      "step": 4850
+    },
+    {
+      "epoch": 0.5022736668044646,
+      "grad_norm": 0.16202867031097412,
+      "learning_rate": 0.006,
+      "loss": 1.1054,
+      "step": 4860
+    },
+    {
+      "epoch": 0.5033071517155849,
+      "grad_norm": 0.13450877368450165,
+      "learning_rate": 0.006,
+      "loss": 1.1112,
+      "step": 4870
+    },
+    {
+      "epoch": 0.5043406366267053,
+      "grad_norm": 0.1484173834323883,
+      "learning_rate": 0.006,
+      "loss": 1.103,
+      "step": 4880
+    },
+    {
+      "epoch": 0.5053741215378256,
+      "grad_norm": 0.14877624809741974,
+      "learning_rate": 0.006,
+      "loss": 1.1353,
+      "step": 4890
+    },
+    {
+      "epoch": 0.5064076064489459,
+      "grad_norm": 0.14426067471504211,
+      "learning_rate": 0.006,
+      "loss": 1.1172,
+      "step": 4900
+    },
+    {
+      "epoch": 0.5074410913600661,
+      "grad_norm": 0.14011584222316742,
+      "learning_rate": 0.006,
+      "loss": 1.1171,
+      "step": 4910
+    },
+    {
+      "epoch": 0.5084745762711864,
+      "grad_norm": 0.20109255611896515,
+      "learning_rate": 0.006,
+      "loss": 1.1097,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5095080611823067,
+      "grad_norm": 0.1302032619714737,
+      "learning_rate": 0.006,
+      "loss": 1.1092,
+      "step": 4930
+    },
+    {
+      "epoch": 0.5105415460934271,
+      "grad_norm": 0.13182514905929565,
+      "learning_rate": 0.006,
+      "loss": 1.1043,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5115750310045474,
+      "grad_norm": 0.13346487283706665,
+      "learning_rate": 0.006,
+      "loss": 1.1133,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5126085159156676,
+      "grad_norm": 0.13681770861148834,
+      "learning_rate": 0.006,
+      "loss": 1.1153,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5136420008267879,
+      "grad_norm": 0.14637672901153564,
+      "learning_rate": 0.006,
+      "loss": 1.0984,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5146754857379082,
+      "grad_norm": 0.14188942313194275,
+      "learning_rate": 0.006,
+      "loss": 1.1033,
+      "step": 4980
+    },
+    {
+      "epoch": 0.5157089706490285,
+      "grad_norm": 0.1467074155807495,
+      "learning_rate": 0.006,
+      "loss": 1.095,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5167424555601489,
+      "grad_norm": 0.15010304749011993,
+      "learning_rate": 0.006,
+      "loss": 1.1029,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5177759404712691,
+      "grad_norm": 0.13308066129684448,
+      "learning_rate": 0.006,
+      "loss": 1.1004,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5188094253823894,
+      "grad_norm": 0.13878673315048218,
+      "learning_rate": 0.006,
+      "loss": 1.1055,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5198429102935097,
+      "grad_norm": 0.14125706255435944,
+      "learning_rate": 0.006,
+      "loss": 1.0927,
+      "step": 5030
+    },
+    {
+      "epoch": 0.52087639520463,
+      "grad_norm": 0.1262088418006897,
+      "learning_rate": 0.006,
+      "loss": 1.1112,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5219098801157503,
+      "grad_norm": 0.20926526188850403,
+      "learning_rate": 0.006,
+      "loss": 1.1169,
+      "step": 5050
+    },
+    {
+      "epoch": 0.5229433650268706,
+      "grad_norm": 0.2341676652431488,
+      "learning_rate": 0.006,
+      "loss": 1.1198,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5239768499379909,
+      "grad_norm": 0.15894795954227448,
+      "learning_rate": 0.006,
+      "loss": 1.1017,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5250103348491112,
+      "grad_norm": 0.14738234877586365,
+      "learning_rate": 0.006,
+      "loss": 1.1021,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5260438197602315,
+      "grad_norm": 0.14371953904628754,
+      "learning_rate": 0.006,
+      "loss": 1.0978,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5270773046713518,
+      "grad_norm": 0.13194578886032104,
+      "learning_rate": 0.006,
+      "loss": 1.1073,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5281107895824722,
+      "grad_norm": 0.1411438137292862,
+      "learning_rate": 0.006,
+      "loss": 1.0931,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5291442744935924,
+      "grad_norm": 0.1347801685333252,
+      "learning_rate": 0.006,
+      "loss": 1.0896,
+      "step": 5120
+    },
+    {
+      "epoch": 0.5301777594047127,
+      "grad_norm": 0.12243802100419998,
+      "learning_rate": 0.006,
+      "loss": 1.0912,
+      "step": 5130
+    },
+    {
+      "epoch": 0.531211244315833,
+      "grad_norm": 0.15618611872196198,
+      "learning_rate": 0.006,
+      "loss": 1.0925,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5322447292269533,
+      "grad_norm": 0.133057102560997,
+      "learning_rate": 0.006,
+      "loss": 1.0937,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5332782141380736,
+      "grad_norm": 0.14488716423511505,
+      "learning_rate": 0.006,
+      "loss": 1.1057,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5343116990491938,
+      "grad_norm": 0.15518608689308167,
+      "learning_rate": 0.006,
+      "loss": 1.1073,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5353451839603142,
+      "grad_norm": 0.14580237865447998,
+      "learning_rate": 0.006,
+      "loss": 1.1064,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5363786688714345,
+      "grad_norm": 0.12775352597236633,
+      "learning_rate": 0.006,
+      "loss": 1.1043,
+      "step": 5190
+    },
+    {
+      "epoch": 0.5374121537825548,
+      "grad_norm": 0.15589575469493866,
+      "learning_rate": 0.006,
+      "loss": 1.0995,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5384456386936751,
+      "grad_norm": 0.1348804533481598,
+      "learning_rate": 0.006,
+      "loss": 1.0959,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5394791236047953,
+      "grad_norm": 0.1302793323993683,
+      "learning_rate": 0.006,
+      "loss": 1.0833,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5405126085159156,
+      "grad_norm": 0.1520976573228836,
+      "learning_rate": 0.006,
+      "loss": 1.0991,
+      "step": 5230
+    },
+    {
+      "epoch": 0.541546093427036,
+      "grad_norm": 0.13820461928844452,
+      "learning_rate": 0.006,
+      "loss": 1.1039,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5425795783381563,
+      "grad_norm": 0.13032610714435577,
+      "learning_rate": 0.006,
+      "loss": 1.0947,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5436130632492766,
+      "grad_norm": 0.1633882224559784,
+      "learning_rate": 0.006,
+      "loss": 1.1009,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5446465481603968,
+      "grad_norm": 0.2063656449317932,
+      "learning_rate": 0.006,
+      "loss": 1.0947,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5456800330715171,
+      "grad_norm": 0.13242821395397186,
+      "learning_rate": 0.006,
+      "loss": 1.0904,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5467135179826375,
+      "grad_norm": 0.15254896879196167,
+      "learning_rate": 0.006,
+      "loss": 1.0955,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5477470028937578,
+      "grad_norm": 0.12627582252025604,
+      "learning_rate": 0.006,
+      "loss": 1.1101,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5487804878048781,
+      "grad_norm": 0.132407084107399,
+      "learning_rate": 0.006,
+      "loss": 1.0947,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5498139727159983,
+      "grad_norm": 0.1298685371875763,
+      "learning_rate": 0.006,
+      "loss": 1.0853,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5508474576271186,
+      "grad_norm": 0.25235456228256226,
+      "learning_rate": 0.006,
+      "loss": 1.1048,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5518809425382389,
+      "grad_norm": 0.15979771316051483,
+      "learning_rate": 0.006,
+      "loss": 1.0945,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5529144274493593,
+      "grad_norm": 0.12209047377109528,
+      "learning_rate": 0.006,
+      "loss": 1.0918,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5539479123604796,
+      "grad_norm": 0.14376087486743927,
+      "learning_rate": 0.006,
+      "loss": 1.1055,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5549813972715998,
+      "grad_norm": 0.12132777273654938,
+      "learning_rate": 0.006,
+      "loss": 1.1009,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5560148821827201,
+      "grad_norm": 0.12317829579114914,
+      "learning_rate": 0.006,
+      "loss": 1.0873,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5570483670938404,
+      "grad_norm": 0.15228794515132904,
+      "learning_rate": 0.006,
+      "loss": 1.0883,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5580818520049607,
+      "grad_norm": 0.11975687742233276,
+      "learning_rate": 0.006,
+      "loss": 1.1075,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5591153369160811,
+      "grad_norm": 0.15219765901565552,
+      "learning_rate": 0.006,
+      "loss": 1.087,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5601488218272013,
+      "grad_norm": 0.12764790654182434,
+      "learning_rate": 0.006,
+      "loss": 1.0928,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5611823067383216,
+      "grad_norm": 0.18155109882354736,
+      "learning_rate": 0.006,
+      "loss": 1.0924,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5622157916494419,
+      "grad_norm": 0.16299748420715332,
+      "learning_rate": 0.006,
+      "loss": 1.0923,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5632492765605622,
+      "grad_norm": 0.1478397250175476,
+      "learning_rate": 0.006,
+      "loss": 1.0838,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5642827614716825,
+      "grad_norm": 0.14866343140602112,
+      "learning_rate": 0.006,
+      "loss": 1.0861,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5653162463828028,
+      "grad_norm": 0.15744392573833466,
+      "learning_rate": 0.006,
+      "loss": 1.0981,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5663497312939231,
+      "grad_norm": 0.1575409471988678,
+      "learning_rate": 0.006,
+      "loss": 1.0972,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5673832162050434,
+      "grad_norm": 0.2215876430273056,
+      "learning_rate": 0.006,
+      "loss": 1.0926,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5684167011161637,
+      "grad_norm": 0.2060500532388687,
+      "learning_rate": 0.006,
+      "loss": 1.0885,
+      "step": 5500
+    },
+    {
+      "epoch": 0.569450186027284,
+      "grad_norm": 0.19035674631595612,
+      "learning_rate": 0.006,
+      "loss": 1.0852,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5704836709384044,
+      "grad_norm": 0.15188254415988922,
+      "learning_rate": 0.006,
+      "loss": 1.0883,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5715171558495246,
+      "grad_norm": 0.1428184062242508,
+      "learning_rate": 0.006,
+      "loss": 1.0777,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5725506407606449,
+      "grad_norm": 0.12388137727975845,
+      "learning_rate": 0.006,
+      "loss": 1.0884,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5735841256717652,
+      "grad_norm": 0.1246667355298996,
+      "learning_rate": 0.006,
+      "loss": 1.0807,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5746176105828855,
+      "grad_norm": 0.15170446038246155,
+      "learning_rate": 0.006,
+      "loss": 1.0908,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5756510954940058,
+      "grad_norm": 0.12245498597621918,
+      "learning_rate": 0.006,
+      "loss": 1.0915,
+      "step": 5570
+    },
+    {
+      "epoch": 0.576684580405126,
+      "grad_norm": 0.12406784296035767,
+      "learning_rate": 0.006,
+      "loss": 1.0821,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5777180653162464,
+      "grad_norm": 0.173844575881958,
+      "learning_rate": 0.006,
+      "loss": 1.0879,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5787515502273667,
+      "grad_norm": 0.1362677961587906,
+      "learning_rate": 0.006,
+      "loss": 1.0929,
+      "step": 5600
+    },
+    {
+      "epoch": 0.579785035138487,
+      "grad_norm": 0.13046661019325256,
+      "learning_rate": 0.006,
+      "loss": 1.0891,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5808185200496073,
+      "grad_norm": 0.13165681064128876,
+      "learning_rate": 0.006,
+      "loss": 1.0842,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5818520049607275,
+      "grad_norm": 0.1174607053399086,
+      "learning_rate": 0.006,
+      "loss": 1.0827,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5828854898718479,
+      "grad_norm": 0.12023981660604477,
+      "learning_rate": 0.006,
+      "loss": 1.0725,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5839189747829682,
+      "grad_norm": 0.16525237262248993,
+      "learning_rate": 0.006,
+      "loss": 1.0812,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5849524596940885,
+      "grad_norm": 0.14079253375530243,
+      "learning_rate": 0.006,
+      "loss": 1.0859,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5859859446052088,
+      "grad_norm": 0.1366300880908966,
+      "learning_rate": 0.006,
+      "loss": 1.0847,
+      "step": 5670
+    },
+    {
+      "epoch": 0.587019429516329,
+      "grad_norm": 0.12389519810676575,
+      "learning_rate": 0.006,
+      "loss": 1.0897,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5880529144274493,
+      "grad_norm": 0.1193692535161972,
+      "learning_rate": 0.006,
+      "loss": 1.0727,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5890863993385697,
+      "grad_norm": 0.14679639041423798,
+      "learning_rate": 0.006,
+      "loss": 1.0849,
+      "step": 5700
+    },
+    {
+      "epoch": 0.59011988424969,
+      "grad_norm": 0.13050967454910278,
+      "learning_rate": 0.006,
+      "loss": 1.0753,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5911533691608103,
+      "grad_norm": 0.12082899361848831,
+      "learning_rate": 0.006,
+      "loss": 1.078,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5921868540719305,
+      "grad_norm": 0.1468031257390976,
+      "learning_rate": 0.006,
+      "loss": 1.0866,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5932203389830508,
+      "grad_norm": 0.12028497457504272,
+      "learning_rate": 0.006,
+      "loss": 1.0726,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5942538238941711,
+      "grad_norm": 0.14379620552062988,
+      "learning_rate": 0.006,
+      "loss": 1.088,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5952873088052915,
+      "grad_norm": 0.13077853620052338,
+      "learning_rate": 0.006,
+      "loss": 1.0743,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5963207937164118,
+      "grad_norm": 0.14271710813045502,
+      "learning_rate": 0.006,
+      "loss": 1.078,
+      "step": 5770
+    },
+    {
+      "epoch": 0.597354278627532,
+      "grad_norm": 0.1357925534248352,
+      "learning_rate": 0.006,
+      "loss": 1.0777,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5983877635386523,
+      "grad_norm": 0.12296663224697113,
+      "learning_rate": 0.006,
+      "loss": 1.0561,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5994212484497726,
+      "grad_norm": 0.1592247039079666,
+      "learning_rate": 0.006,
+      "loss": 1.0731,
+      "step": 5800
+    },
+    {
+      "epoch": 0.600454733360893,
+      "grad_norm": 0.24744771420955658,
+      "learning_rate": 0.006,
+      "loss": 1.0703,
+      "step": 5810
+    },
+    {
+      "epoch": 0.6014882182720133,
+      "grad_norm": 0.18649180233478546,
+      "learning_rate": 0.006,
+      "loss": 1.0663,
+      "step": 5820
+    },
+    {
+      "epoch": 0.6025217031831335,
+      "grad_norm": 0.15970060229301453,
+      "learning_rate": 0.006,
+      "loss": 1.0814,
+      "step": 5830
+    },
+    {
+      "epoch": 0.6035551880942538,
+      "grad_norm": 0.15505217015743256,
+      "learning_rate": 0.006,
+      "loss": 1.0819,
+      "step": 5840
+    },
+    {
+      "epoch": 0.6045886730053741,
+      "grad_norm": 0.1281612068414688,
+      "learning_rate": 0.006,
+      "loss": 1.073,
+      "step": 5850
+    },
+    {
+      "epoch": 0.6056221579164944,
+      "grad_norm": 0.13322196900844574,
+      "learning_rate": 0.006,
+      "loss": 1.0851,
+      "step": 5860
+    },
+    {
+      "epoch": 0.6066556428276147,
+      "grad_norm": 0.1385001540184021,
+      "learning_rate": 0.006,
+      "loss": 1.0714,
+      "step": 5870
+    },
+    {
+      "epoch": 0.6076891277387351,
+      "grad_norm": 0.12118349224328995,
+      "learning_rate": 0.006,
+      "loss": 1.0744,
+      "step": 5880
+    },
+    {
+      "epoch": 0.6087226126498553,
+      "grad_norm": 0.12691213190555573,
+      "learning_rate": 0.006,
+      "loss": 1.0664,
+      "step": 5890
+    },
+    {
+      "epoch": 0.6097560975609756,
+      "grad_norm": 0.14859186112880707,
+      "learning_rate": 0.006,
+      "loss": 1.0821,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6107895824720959,
+      "grad_norm": 0.14851497113704681,
+      "learning_rate": 0.006,
+      "loss": 1.08,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6118230673832162,
+      "grad_norm": 0.1358513981103897,
+      "learning_rate": 0.006,
+      "loss": 1.0747,
+      "step": 5920
+    },
+    {
+      "epoch": 0.6128565522943366,
+      "grad_norm": 0.14147254824638367,
+      "learning_rate": 0.006,
+      "loss": 1.0805,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6138900372054568,
+      "grad_norm": 0.13193708658218384,
+      "learning_rate": 0.006,
+      "loss": 1.079,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6149235221165771,
+      "grad_norm": 0.1255359649658203,
+      "learning_rate": 0.006,
+      "loss": 1.0785,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6159570070276974,
+      "grad_norm": 0.19004510343074799,
+      "learning_rate": 0.006,
+      "loss": 1.0728,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6169904919388177,
+      "grad_norm": 0.2386738657951355,
+      "learning_rate": 0.006,
+      "loss": 1.0836,
+      "step": 5970
+    },
+    {
+      "epoch": 0.618023976849938,
+      "grad_norm": 0.14890602231025696,
+      "learning_rate": 0.006,
+      "loss": 1.0791,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6190574617610582,
+      "grad_norm": 0.15145020186901093,
+      "learning_rate": 0.006,
+      "loss": 1.0717,
+      "step": 5990
+    },
+    {
+      "epoch": 0.6200909466721786,
+      "grad_norm": 0.12245503813028336,
+      "learning_rate": 0.006,
+      "loss": 1.064,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6211244315832989,
+      "grad_norm": 0.13569168746471405,
+      "learning_rate": 0.006,
+      "loss": 1.0717,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6221579164944192,
+      "grad_norm": 0.14238642156124115,
+      "learning_rate": 0.006,
+      "loss": 1.072,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6231914014055395,
+      "grad_norm": 0.13332600891590118,
+      "learning_rate": 0.006,
+      "loss": 1.0821,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6242248863166597,
+      "grad_norm": 0.12775711715221405,
+      "learning_rate": 0.006,
+      "loss": 1.0665,
+      "step": 6040
+    },
+    {
+      "epoch": 0.62525837122778,
+      "grad_norm": 0.12378199398517609,
+      "learning_rate": 0.006,
+      "loss": 1.0594,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6262918561389004,
+      "grad_norm": 0.127638041973114,
+      "learning_rate": 0.006,
+      "loss": 1.064,
+      "step": 6060
+    },
+    {
+      "epoch": 0.6273253410500207,
+      "grad_norm": 0.1450446993112564,
+      "learning_rate": 0.006,
+      "loss": 1.0718,
+      "step": 6070
+    },
+    {
+      "epoch": 0.628358825961141,
+      "grad_norm": 0.14722521603107452,
+      "learning_rate": 0.006,
+      "loss": 1.0786,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6293923108722612,
+      "grad_norm": 0.11357062309980392,
+      "learning_rate": 0.006,
+      "loss": 1.0713,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6304257957833815,
+      "grad_norm": 0.1290910542011261,
+      "learning_rate": 0.006,
+      "loss": 1.0692,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6314592806945019,
+      "grad_norm": 0.12275319546461105,
+      "learning_rate": 0.006,
+      "loss": 1.0599,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6324927656056222,
+      "grad_norm": 0.14953821897506714,
+      "learning_rate": 0.006,
+      "loss": 1.0735,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6335262505167425,
+      "grad_norm": 0.15243026614189148,
+      "learning_rate": 0.006,
+      "loss": 1.0641,
+      "step": 6130
+    },
+    {
+      "epoch": 0.6345597354278627,
+      "grad_norm": 0.17082957923412323,
+      "learning_rate": 0.006,
+      "loss": 1.0669,
+      "step": 6140
+    },
+    {
+      "epoch": 0.635593220338983,
+      "grad_norm": 0.14030581712722778,
+      "learning_rate": 0.006,
+      "loss": 1.0659,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6366267052501033,
+      "grad_norm": 0.1736230105161667,
+      "learning_rate": 0.006,
+      "loss": 1.0698,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6376601901612237,
+      "grad_norm": 0.12367543578147888,
+      "learning_rate": 0.006,
+      "loss": 1.0657,
+      "step": 6170
+    },
+    {
+      "epoch": 0.638693675072344,
+      "grad_norm": 0.14306339621543884,
+      "learning_rate": 0.006,
+      "loss": 1.0731,
+      "step": 6180
+    },
+    {
+      "epoch": 0.6397271599834642,
+      "grad_norm": 0.1417074203491211,
+      "learning_rate": 0.006,
+      "loss": 1.0704,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6407606448945845,
+      "grad_norm": 0.12904278934001923,
+      "learning_rate": 0.006,
+      "loss": 1.0619,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6417941298057048,
+      "grad_norm": 0.13394799828529358,
+      "learning_rate": 0.006,
+      "loss": 1.054,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6428276147168251,
+      "grad_norm": 0.13006740808486938,
+      "learning_rate": 0.006,
+      "loss": 1.058,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6438610996279455,
+      "grad_norm": 0.1411658525466919,
+      "learning_rate": 0.006,
+      "loss": 1.0688,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6448945845390658,
+      "grad_norm": 0.12172463536262512,
+      "learning_rate": 0.006,
+      "loss": 1.0599,
+      "step": 6240
+    },
+    {
+      "epoch": 0.645928069450186,
+      "grad_norm": 0.1132531389594078,
+      "learning_rate": 0.006,
+      "loss": 1.0493,
+      "step": 6250
+    },
+    {
+      "epoch": 0.6469615543613063,
+      "grad_norm": 0.15614502131938934,
+      "learning_rate": 0.006,
+      "loss": 1.0637,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6479950392724266,
+      "grad_norm": 0.14655426144599915,
+      "learning_rate": 0.006,
+      "loss": 1.0537,
+      "step": 6270
+    },
+    {
+      "epoch": 0.649028524183547,
+      "grad_norm": 0.1248323991894722,
+      "learning_rate": 0.006,
+      "loss": 1.0735,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6500620090946673,
+      "grad_norm": 0.13113942742347717,
+      "learning_rate": 0.006,
+      "loss": 1.0561,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6510954940057875,
+      "grad_norm": 0.11488605290651321,
+      "learning_rate": 0.006,
+      "loss": 1.0626,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6521289789169078,
+      "grad_norm": 0.13470590114593506,
+      "learning_rate": 0.006,
+      "loss": 1.0629,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6531624638280281,
+      "grad_norm": 0.13424867391586304,
+      "learning_rate": 0.006,
+      "loss": 1.0634,
+      "step": 6320
+    },
+    {
+      "epoch": 0.6541959487391484,
+      "grad_norm": 0.13190287351608276,
+      "learning_rate": 0.006,
+      "loss": 1.0697,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6552294336502688,
+      "grad_norm": 0.12076553702354431,
+      "learning_rate": 0.006,
+      "loss": 1.0559,
+      "step": 6340
+    },
+    {
+      "epoch": 0.656262918561389,
+      "grad_norm": 0.12332470715045929,
+      "learning_rate": 0.006,
+      "loss": 1.0511,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6572964034725093,
+      "grad_norm": 0.13250276446342468,
+      "learning_rate": 0.006,
+      "loss": 1.0772,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6583298883836296,
+      "grad_norm": 0.12431011348962784,
+      "learning_rate": 0.006,
+      "loss": 1.0574,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6593633732947499,
+      "grad_norm": 0.11310591548681259,
+      "learning_rate": 0.006,
+      "loss": 1.0548,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6603968582058702,
+      "grad_norm": 0.12900187075138092,
+      "learning_rate": 0.006,
+      "loss": 1.0624,
+      "step": 6390
+    },
+    {
+      "epoch": 0.6614303431169904,
+      "grad_norm": 0.21367403864860535,
+      "learning_rate": 0.006,
+      "loss": 1.0681,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6624638280281108,
+      "grad_norm": 0.17459499835968018,
+      "learning_rate": 0.006,
+      "loss": 1.0547,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6634973129392311,
+      "grad_norm": 0.12943004071712494,
+      "learning_rate": 0.006,
+      "loss": 1.0659,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6645307978503514,
+      "grad_norm": 0.12108403444290161,
+      "learning_rate": 0.006,
+      "loss": 1.0721,
+      "step": 6430
+    },
+    {
+      "epoch": 0.6655642827614717,
+      "grad_norm": 0.14485575258731842,
+      "learning_rate": 0.006,
+      "loss": 1.0502,
+      "step": 6440
+    },
+    {
+      "epoch": 0.6665977676725919,
+      "grad_norm": 0.12145304679870605,
+      "learning_rate": 0.006,
+      "loss": 1.0489,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6676312525837123,
+      "grad_norm": 0.12879317998886108,
+      "learning_rate": 0.006,
+      "loss": 1.0603,
+      "step": 6460
+    },
+    {
+      "epoch": 0.6686647374948326,
+      "grad_norm": 0.13428542017936707,
+      "learning_rate": 0.006,
+      "loss": 1.0538,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6696982224059529,
+      "grad_norm": 0.13670314848423004,
+      "learning_rate": 0.006,
+      "loss": 1.053,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6707317073170732,
+      "grad_norm": 0.12726275622844696,
+      "learning_rate": 0.006,
+      "loss": 1.0555,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6717651922281934,
+      "grad_norm": 0.1293756067752838,
+      "learning_rate": 0.006,
+      "loss": 1.0579,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6727986771393137,
+      "grad_norm": 0.15011049807071686,
+      "learning_rate": 0.006,
+      "loss": 1.0528,
+      "step": 6510
+    },
+    {
+      "epoch": 0.6738321620504341,
+      "grad_norm": 0.1232418343424797,
+      "learning_rate": 0.006,
+      "loss": 1.0474,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6748656469615544,
+      "grad_norm": 0.13466301560401917,
+      "learning_rate": 0.006,
+      "loss": 1.0561,
+      "step": 6530
+    },
+    {
+      "epoch": 0.6758991318726747,
+      "grad_norm": 0.18297189474105835,
+      "learning_rate": 0.006,
+      "loss": 1.0569,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6769326167837949,
+      "grad_norm": 0.13041065633296967,
+      "learning_rate": 0.006,
+      "loss": 1.0614,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6779661016949152,
+      "grad_norm": 0.15428952872753143,
+      "learning_rate": 0.006,
+      "loss": 1.0526,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6789995866060355,
+      "grad_norm": 0.1226574033498764,
+      "learning_rate": 0.006,
+      "loss": 1.0476,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6800330715171559,
+      "grad_norm": 0.1592113971710205,
+      "learning_rate": 0.006,
+      "loss": 1.057,
+      "step": 6580
+    },
+    {
+      "epoch": 0.6810665564282762,
+      "grad_norm": 0.131056547164917,
+      "learning_rate": 0.006,
+      "loss": 1.0643,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6821000413393964,
+      "grad_norm": 0.17470133304595947,
+      "learning_rate": 0.006,
+      "loss": 1.0545,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6831335262505167,
+      "grad_norm": 0.14145781099796295,
+      "learning_rate": 0.006,
+      "loss": 1.0459,
+      "step": 6610
+    },
+    {
+      "epoch": 0.684167011161637,
+      "grad_norm": 0.11432395130395889,
+      "learning_rate": 0.006,
+      "loss": 1.0454,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6852004960727573,
+      "grad_norm": 0.12126456946134567,
+      "learning_rate": 0.006,
+      "loss": 1.0467,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6862339809838777,
+      "grad_norm": 0.12609295547008514,
+      "learning_rate": 0.006,
+      "loss": 1.0734,
+      "step": 6640
+    },
+    {
+      "epoch": 0.687267465894998,
+      "grad_norm": 0.15360839664936066,
+      "learning_rate": 0.006,
+      "loss": 1.0424,
+      "step": 6650
+    },
+    {
+      "epoch": 0.6883009508061182,
+      "grad_norm": 0.13050951063632965,
+      "learning_rate": 0.006,
+      "loss": 1.0532,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6893344357172385,
+      "grad_norm": 0.11345363408327103,
+      "learning_rate": 0.006,
+      "loss": 1.0616,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6903679206283588,
+      "grad_norm": 0.12678933143615723,
+      "learning_rate": 0.006,
+      "loss": 1.0472,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6914014055394792,
+      "grad_norm": 0.12220112234354019,
+      "learning_rate": 0.006,
+      "loss": 1.048,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6924348904505995,
+      "grad_norm": 0.12634249031543732,
+      "learning_rate": 0.006,
+      "loss": 1.0489,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6934683753617197,
+      "grad_norm": 0.14149808883666992,
+      "learning_rate": 0.006,
+      "loss": 1.0548,
+      "step": 6710
+    },
+    {
+      "epoch": 0.69450186027284,
+      "grad_norm": 0.12240725010633469,
+      "learning_rate": 0.006,
+      "loss": 1.0466,
+      "step": 6720
+    },
+    {
+      "epoch": 0.6955353451839603,
+      "grad_norm": 0.1353718340396881,
+      "learning_rate": 0.006,
+      "loss": 1.0423,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6965688300950806,
+      "grad_norm": 0.18347500264644623,
+      "learning_rate": 0.006,
+      "loss": 1.0485,
+      "step": 6740
+    },
+    {
+      "epoch": 0.697602315006201,
+      "grad_norm": 0.12348928302526474,
+      "learning_rate": 0.006,
+      "loss": 1.0573,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6986357999173212,
+      "grad_norm": 0.1428414285182953,
+      "learning_rate": 0.006,
+      "loss": 1.0412,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6996692848284415,
+      "grad_norm": 0.1410830169916153,
+      "learning_rate": 0.006,
+      "loss": 1.0569,
+      "step": 6770
+    },
+    {
+      "epoch": 0.7007027697395618,
+      "grad_norm": 0.14577074348926544,
+      "learning_rate": 0.005935620400122873,
+      "loss": 1.0546,
+      "step": 6780
+    },
+    {
+      "epoch": 0.7017362546506821,
+      "grad_norm": 0.14377078413963318,
+      "learning_rate": 0.005842197295969981,
+      "loss": 1.045,
+      "step": 6790
+    },
+    {
+      "epoch": 0.7027697395618024,
+      "grad_norm": 0.12980251014232635,
+      "learning_rate": 0.005750244615429315,
+      "loss": 1.0595,
+      "step": 6800
+    },
+    {
+      "epoch": 0.7038032244729227,
+      "grad_norm": 0.13310031592845917,
+      "learning_rate": 0.005659739214915371,
+      "loss": 1.0492,
+      "step": 6810
+    },
+    {
+      "epoch": 0.704836709384043,
+      "grad_norm": 0.12173794209957123,
+      "learning_rate": 0.005570658315108789,
+      "loss": 1.0409,
+      "step": 6820
+    },
+    {
+      "epoch": 0.7058701942951633,
+      "grad_norm": 0.17054332792758942,
+      "learning_rate": 0.005482979495223033,
+      "loss": 1.0505,
+      "step": 6830
+    },
+    {
+      "epoch": 0.7069036792062836,
+      "grad_norm": 0.18403039872646332,
+      "learning_rate": 0.005396680687361299,
+      "loss": 1.0462,
+      "step": 6840
+    },
+    {
+      "epoch": 0.7079371641174039,
+      "grad_norm": 0.13148552179336548,
+      "learning_rate": 0.005311740170962234,
+      "loss": 1.049,
+      "step": 6850
+    },
+    {
+      "epoch": 0.7089706490285241,
+      "grad_norm": 0.13557471334934235,
+      "learning_rate": 0.005228136567333095,
+      "loss": 1.0348,
+      "step": 6860
+    },
+    {
+      "epoch": 0.7100041339396445,
+      "grad_norm": 0.11231327056884766,
+      "learning_rate": 0.005145848834268932,
+      "loss": 1.0309,
+      "step": 6870
+    },
+    {
+      "epoch": 0.7110376188507648,
+      "grad_norm": 0.14308036863803864,
+      "learning_rate": 0.005064856260756483,
+      "loss": 1.024,
+      "step": 6880
+    },
+    {
+      "epoch": 0.7120711037618851,
+      "grad_norm": 0.11384549736976624,
+      "learning_rate": 0.0049851384617614035,
+      "loss": 1.0255,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7131045886730054,
+      "grad_norm": 0.1373758465051651,
+      "learning_rate": 0.004906675373097566,
+      "loss": 1.0248,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7141380735841256,
+      "grad_norm": 0.13972637057304382,
+      "learning_rate": 0.00482944724637709,
+      "loss": 1.0197,
+      "step": 6910
+    },
+    {
+      "epoch": 0.7151715584952459,
+      "grad_norm": 0.11777855455875397,
+      "learning_rate": 0.004753434644039878,
+      "loss": 1.0197,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7162050434063663,
+      "grad_norm": 0.10751146823167801,
+      "learning_rate": 0.004678618434461362,
+      "loss": 1.0272,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7172385283174866,
+      "grad_norm": 0.11851785331964493,
+      "learning_rate": 0.004604979787137271,
+      "loss": 1.0109,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7182720132286069,
+      "grad_norm": 0.1341877430677414,
+      "learning_rate": 0.004532500167944173,
+      "loss": 1.0106,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7193054981397271,
+      "grad_norm": 0.11008955538272858,
+      "learning_rate": 0.004461161334474619,
+      "loss": 1.0152,
+      "step": 6960
+    },
+    {
+      "epoch": 0.7203389830508474,
+      "grad_norm": 0.16443827748298645,
+      "learning_rate": 0.0043909453314457106,
+      "loss": 1.0118,
+      "step": 6970
+    },
+    {
+      "epoch": 0.7213724679619677,
+      "grad_norm": 0.11731505393981934,
+      "learning_rate": 0.004321834486179928,
+      "loss": 1.0081,
+      "step": 6980
+    },
+    {
+      "epoch": 0.7224059528730881,
+      "grad_norm": 0.11518187075853348,
+      "learning_rate": 0.004253811404157096,
+      "loss": 1.004,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7234394377842084,
+      "grad_norm": 0.10781227797269821,
+      "learning_rate": 0.004186858964636351,
+      "loss": 0.9879,
+      "step": 7000
+    },
+    {
+      "epoch": 0.7244729226953287,
+      "grad_norm": 0.12845131754875183,
+      "learning_rate": 0.004120960316347017,
+      "loss": 1.0148,
+      "step": 7010
+    },
+    {
+      "epoch": 0.7255064076064489,
+      "grad_norm": 0.11902609467506409,
+      "learning_rate": 0.004056098873247311,
+      "loss": 1.0012,
+      "step": 7020
+    },
+    {
+      "epoch": 0.7265398925175692,
+      "grad_norm": 0.11711709946393967,
+      "learning_rate": 0.003992258310349796,
+      "loss": 1.0134,
+      "step": 7030
+    },
+    {
+      "epoch": 0.7275733774286895,
+      "grad_norm": 0.12396284192800522,
+      "learning_rate": 0.003929422559612546,
+      "loss": 1.0023,
+      "step": 7040
+    },
+    {
+      "epoch": 0.7286068623398099,
+      "grad_norm": 0.10830879211425781,
+      "learning_rate": 0.0038675758058949712,
+      "loss": 1.0126,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7296403472509302,
+      "grad_norm": 0.11904972046613693,
+      "learning_rate": 0.003806702482977311,
+      "loss": 0.9993,
+      "step": 7060
+    },
+    {
+      "epoch": 0.7306738321620504,
+      "grad_norm": 0.14758719503879547,
+      "learning_rate": 0.0037467872696427614,
+      "loss": 0.9919,
+      "step": 7070
+    },
+    {
+      "epoch": 0.7317073170731707,
+      "grad_norm": 0.10925702750682831,
+      "learning_rate": 0.0036878150858212825,
+      "loss": 0.9991,
+      "step": 7080
+    },
+    {
+      "epoch": 0.732740801984291,
+      "grad_norm": 0.1237645223736763,
+      "learning_rate": 0.0036297710887940875,
+      "loss": 1.004,
+      "step": 7090
+    },
+    {
+      "epoch": 0.7337742868954114,
+      "grad_norm": 0.12318721413612366,
+      "learning_rate": 0.003572640669457879,
+      "loss": 0.9833,
+      "step": 7100
+    },
+    {
+      "epoch": 0.7348077718065317,
+      "grad_norm": 0.10309872031211853,
+      "learning_rate": 0.003516409448647884,
+      "loss": 0.9919,
+      "step": 7110
+    },
+    {
+      "epoch": 0.7358412567176519,
+      "grad_norm": 0.10170971602201462,
+      "learning_rate": 0.003461063273518752,
+      "loss": 0.9908,
+      "step": 7120
+    },
+    {
+      "epoch": 0.7368747416287722,
+      "grad_norm": 0.13282230496406555,
+      "learning_rate": 0.0034065882139824315,
+      "loss": 0.9849,
+      "step": 7130
+    },
+    {
+      "epoch": 0.7379082265398925,
+      "grad_norm": 0.10541505366563797,
+      "learning_rate": 0.0033529705592020968,
+      "loss": 0.9893,
+      "step": 7140
+    },
+    {
+      "epoch": 0.7389417114510128,
+      "grad_norm": 0.10669828951358795,
+      "learning_rate": 0.0033001968141412714,
+      "loss": 0.9841,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7399751963621332,
+      "grad_norm": 0.11185015738010406,
+      "learning_rate": 0.0032482536961672545,
+      "loss": 0.9888,
+      "step": 7160
+    },
+    {
+      "epoch": 0.7410086812732534,
+      "grad_norm": 0.11848310381174088,
+      "learning_rate": 0.003197128131708016,
+      "loss": 0.9753,
+      "step": 7170
+    },
+    {
+      "epoch": 0.7420421661843737,
+      "grad_norm": 0.12162330746650696,
+      "learning_rate": 0.0031468072529617073,
+      "loss": 0.9896,
+      "step": 7180
+    },
+    {
+      "epoch": 0.743075651095494,
+      "grad_norm": 0.10726076364517212,
+      "learning_rate": 0.0030972783946579593,
+      "loss": 0.967,
+      "step": 7190
+    },
+    {
+      "epoch": 0.7441091360066143,
+      "grad_norm": 0.11569386720657349,
+      "learning_rate": 0.003048529090870162,
+      "loss": 0.9691,
+      "step": 7200
+    },
+    {
+      "epoch": 0.7451426209177346,
+      "grad_norm": 0.11116993427276611,
+      "learning_rate": 0.0030005470718779117,
+      "loss": 0.9783,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7461761058288549,
+      "grad_norm": 0.10824961215257645,
+      "learning_rate": 0.002953320261078841,
+      "loss": 0.9589,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7472095907399752,
+      "grad_norm": 0.1097089946269989,
+      "learning_rate": 0.00290683677194906,
+      "loss": 0.9707,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7482430756510955,
+      "grad_norm": 0.13413402438163757,
+      "learning_rate": 0.002861084905051434,
+      "loss": 0.9583,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7492765605622158,
+      "grad_norm": 0.10966306924819946,
+      "learning_rate": 0.0028160531450909495,
+      "loss": 0.9656,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7503100454733361,
+      "grad_norm": 0.10541026294231415,
+      "learning_rate": 0.0027717301580164277,
+      "loss": 0.9568,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7513435303844563,
+      "grad_norm": 0.10642149299383163,
+      "learning_rate": 0.002728104788167857,
+      "loss": 0.973,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7523770152955767,
+      "grad_norm": 0.10307744145393372,
+      "learning_rate": 0.0026851660554686207,
+      "loss": 0.973,
+      "step": 7280
+    },
+    {
+      "epoch": 0.753410500206697,
+      "grad_norm": 0.10917676985263824,
+      "learning_rate": 0.002642903152661921,
+      "loss": 0.9739,
+      "step": 7290
+    },
+    {
+      "epoch": 0.7544439851178173,
+      "grad_norm": 0.12075357139110565,
+      "learning_rate": 0.002601305442590699,
+      "loss": 0.9668,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7554774700289376,
+      "grad_norm": 0.10885104537010193,
+      "learning_rate": 0.0025603624555203654,
+      "loss": 0.9732,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7565109549400578,
+      "grad_norm": 0.11787907034158707,
+      "learning_rate": 0.0025200638865036733,
+      "loss": 0.9597,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7575444398511781,
+      "grad_norm": 0.11566518247127533,
+      "learning_rate": 0.0024803995927870632,
+      "loss": 0.9595,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7585779247622985,
+      "grad_norm": 0.09844059497117996,
+      "learning_rate": 0.002441359591257831,
+      "loss": 0.9561,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7596114096734188,
+      "grad_norm": 0.1304013878107071,
+      "learning_rate": 0.002402934055931478,
+      "loss": 0.9454,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7606448945845391,
+      "grad_norm": 0.1041470617055893,
+      "learning_rate": 0.002365113315478606,
+      "loss": 0.9453,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7616783794956593,
+      "grad_norm": 0.10921091586351395,
+      "learning_rate": 0.00232788785079074,
+      "loss": 0.9644,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7627118644067796,
+      "grad_norm": 0.12734107673168182,
+      "learning_rate": 0.0022912482925844615,
+      "loss": 0.9491,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7637453493179,
+      "grad_norm": 0.11103381216526031,
+      "learning_rate": 0.0022551854190432517,
+      "loss": 0.9379,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7647788342290203,
+      "grad_norm": 0.10525085031986237,
+      "learning_rate": 0.0022196901534964524,
+      "loss": 0.956,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7658123191401406,
+      "grad_norm": 0.10325931012630463,
+      "learning_rate": 0.0021847535621347543,
+      "loss": 0.9478,
+      "step": 7410
+    },
+    {
+      "epoch": 0.7668458040512609,
+      "grad_norm": 0.10702349990606308,
+      "learning_rate": 0.0021503668517616494,
+      "loss": 0.9408,
+      "step": 7420
+    },
+    {
+      "epoch": 0.7678792889623811,
+      "grad_norm": 0.10343906283378601,
+      "learning_rate": 0.002116521367580266,
+      "loss": 0.9505,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7689127738735014,
+      "grad_norm": 0.11631717532873154,
+      "learning_rate": 0.0020832085910150425,
+      "loss": 0.9483,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7699462587846218,
+      "grad_norm": 0.114792600274086,
+      "learning_rate": 0.0020504201375676876,
+      "loss": 0.9425,
+      "step": 7450
+    },
+    {
+      "epoch": 0.7709797436957421,
+      "grad_norm": 0.10078386217355728,
+      "learning_rate": 0.0020181477547068817,
+      "loss": 0.955,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7720132286068624,
+      "grad_norm": 0.10332197695970535,
+      "learning_rate": 0.0019863833197911983,
+      "loss": 0.9488,
+      "step": 7470
+    },
+    {
+      "epoch": 0.7730467135179826,
+      "grad_norm": 0.1078406572341919,
+      "learning_rate": 0.001955118838024712,
+      "loss": 0.9536,
+      "step": 7480
+    },
+    {
+      "epoch": 0.7740801984291029,
+      "grad_norm": 0.1059098169207573,
+      "learning_rate": 0.0019243464404447924,
+      "loss": 0.9396,
+      "step": 7490
+    },
+    {
+      "epoch": 0.7751136833402232,
+      "grad_norm": 0.10873686522245407,
+      "learning_rate": 0.0018940583819415564,
+      "loss": 0.9412,
+      "step": 7500
+    },
+    {
+      "epoch": 0.7761471682513436,
+      "grad_norm": 0.10201577097177505,
+      "learning_rate": 0.0018642470393085062,
+      "loss": 0.9406,
+      "step": 7510
+    },
+    {
+      "epoch": 0.7771806531624639,
+      "grad_norm": 0.10363386571407318,
+      "learning_rate": 0.0018349049093238407,
+      "loss": 0.9349,
+      "step": 7520
+    },
+    {
+      "epoch": 0.7782141380735841,
+      "grad_norm": 0.10290543735027313,
+      "learning_rate": 0.001806024606861967,
+      "loss": 0.9433,
+      "step": 7530
+    },
+    {
+      "epoch": 0.7792476229847044,
+      "grad_norm": 0.09915654361248016,
+      "learning_rate": 0.0017775988630347405,
+      "loss": 0.9333,
+      "step": 7540
+    },
+    {
+      "epoch": 0.7802811078958247,
+      "grad_norm": 0.10667795687913895,
+      "learning_rate": 0.0017496205233619542,
+      "loss": 0.9282,
+      "step": 7550
+    },
+    {
+      "epoch": 0.781314592806945,
+      "grad_norm": 0.0981353297829628,
+      "learning_rate": 0.001722082545970627,
+      "loss": 0.9443,
+      "step": 7560
+    },
+    {
+      "epoch": 0.7823480777180654,
+      "grad_norm": 0.09341371059417725,
+      "learning_rate": 0.0016949779998226353,
+      "loss": 0.9255,
+      "step": 7570
+    },
+    {
+      "epoch": 0.7833815626291856,
+      "grad_norm": 0.09437743574380875,
+      "learning_rate": 0.0016683000629702357,
+      "loss": 0.9374,
+      "step": 7580
+    },
+    {
+      "epoch": 0.7844150475403059,
+      "grad_norm": 0.10137518495321274,
+      "learning_rate": 0.001642042020839051,
+      "loss": 0.9434,
+      "step": 7590
+    },
+    {
+      "epoch": 0.7854485324514262,
+      "grad_norm": 0.10156513750553131,
+      "learning_rate": 0.001616197264538076,
+      "loss": 0.9282,
+      "step": 7600
+    },
+    {
+      "epoch": 0.7864820173625465,
+      "grad_norm": 0.10588687658309937,
+      "learning_rate": 0.0015907592891962843,
+      "loss": 0.9329,
+      "step": 7610
+    },
+    {
+      "epoch": 0.7875155022736668,
+      "grad_norm": 0.10155037045478821,
+      "learning_rate": 0.0015657216923254177,
+      "loss": 0.9268,
+      "step": 7620
+    },
+    {
+      "epoch": 0.788548987184787,
+      "grad_norm": 0.10379229485988617,
+      "learning_rate": 0.0015410781722085428,
+      "loss": 0.9437,
+      "step": 7630
+    },
+    {
+      "epoch": 0.7895824720959074,
+      "grad_norm": 0.10199972987174988,
+      "learning_rate": 0.0015168225263139693,
+      "loss": 0.9425,
+      "step": 7640
+    },
+    {
+      "epoch": 0.7906159570070277,
+      "grad_norm": 0.10469642281532288,
+      "learning_rate": 0.0014929486497341341,
+      "loss": 0.924,
+      "step": 7650
+    },
+    {
+      "epoch": 0.791649441918148,
+      "grad_norm": 0.09616528451442719,
+      "learning_rate": 0.0014694505336490583,
+      "loss": 0.931,
+      "step": 7660
+    },
+    {
+      "epoch": 0.7926829268292683,
+      "grad_norm": 0.10645470768213272,
+      "learning_rate": 0.001446322263813983,
+      "loss": 0.9266,
+      "step": 7670
+    },
+    {
+      "epoch": 0.7937164117403885,
+      "grad_norm": 0.10266025364398956,
+      "learning_rate": 0.0014235580190708143,
+      "loss": 0.9251,
+      "step": 7680
+    },
+    {
+      "epoch": 0.7947498966515089,
+      "grad_norm": 0.10014299303293228,
+      "learning_rate": 0.0014011520698829947,
+      "loss": 0.9268,
+      "step": 7690
+    },
+    {
+      "epoch": 0.7957833815626292,
+      "grad_norm": 0.10017547011375427,
+      "learning_rate": 0.0013790987768934346,
+      "loss": 0.9345,
+      "step": 7700
+    },
+    {
+      "epoch": 0.7968168664737495,
+      "grad_norm": 0.09650063514709473,
+      "learning_rate": 0.0013573925895051418,
+      "loss": 0.919,
+      "step": 7710
+    },
+    {
+      "epoch": 0.7978503513848698,
+      "grad_norm": 0.11645365506410599,
+      "learning_rate": 0.00133602804448419,
+      "loss": 0.9256,
+      "step": 7720
+    },
+    {
+      "epoch": 0.79888383629599,
+      "grad_norm": 0.09640100598335266,
+      "learning_rate": 0.0013149997645846788,
+      "loss": 0.9252,
+      "step": 7730
+    },
+    {
+      "epoch": 0.7999173212071103,
+      "grad_norm": 0.09893805533647537,
+      "learning_rate": 0.0012943024571953315,
+      "loss": 0.9209,
+      "step": 7740
+    },
+    {
+      "epoch": 0.8009508061182307,
+      "grad_norm": 0.10975757986307144,
+      "learning_rate": 0.001273930913007398,
+      "loss": 0.9259,
+      "step": 7750
+    },
+    {
+      "epoch": 0.801984291029351,
+      "grad_norm": 0.09872964024543762,
+      "learning_rate": 0.0012538800047035218,
+      "loss": 0.9139,
+      "step": 7760
+    },
+    {
+      "epoch": 0.8030177759404713,
+      "grad_norm": 0.10294859856367111,
+      "learning_rate": 0.0012341446856672469,
+      "loss": 0.9219,
+      "step": 7770
+    },
+    {
+      "epoch": 0.8040512608515916,
+      "grad_norm": 0.09699690341949463,
+      "learning_rate": 0.0012147199887128318,
+      "loss": 0.9171,
+      "step": 7780
+    },
+    {
+      "epoch": 0.8050847457627118,
+      "grad_norm": 0.10195200890302658,
+      "learning_rate": 0.0011956010248350593,
+      "loss": 0.9181,
+      "step": 7790
+    },
+    {
+      "epoch": 0.8061182306738321,
+      "grad_norm": 0.10153238475322723,
+      "learning_rate": 0.0011767829819787205,
+      "loss": 0.9122,
+      "step": 7800
+    },
+    {
+      "epoch": 0.8071517155849525,
+      "grad_norm": 0.09719067811965942,
+      "learning_rate": 0.0011582611238274692,
+      "loss": 0.9261,
+      "step": 7810
+    },
+    {
+      "epoch": 0.8081852004960728,
+      "grad_norm": 0.09960606694221497,
+      "learning_rate": 0.0011400307886117367,
+      "loss": 0.9159,
+      "step": 7820
+    },
+    {
+      "epoch": 0.8092186854071931,
+      "grad_norm": 0.10018911957740784,
+      "learning_rate": 0.001122087387935411,
+      "loss": 0.9043,
+      "step": 7830
+    },
+    {
+      "epoch": 0.8102521703183133,
+      "grad_norm": 0.09987188875675201,
+      "learning_rate": 0.0011044264056209812,
+      "loss": 0.923,
+      "step": 7840
+    },
+    {
+      "epoch": 0.8112856552294336,
+      "grad_norm": 0.10447081923484802,
+      "learning_rate": 0.0010870433965728626,
+      "loss": 0.912,
+      "step": 7850
+    },
+    {
+      "epoch": 0.812319140140554,
+      "grad_norm": 0.09891889244318008,
+      "learning_rate": 0.0010699339856586065,
+      "loss": 0.9231,
+      "step": 7860
+    },
+    {
+      "epoch": 0.8133526250516743,
+      "grad_norm": 0.11543388664722443,
+      "learning_rate": 0.0010530938666077263,
+      "loss": 0.9189,
+      "step": 7870
+    },
+    {
+      "epoch": 0.8143861099627946,
+      "grad_norm": 0.10272625833749771,
+      "learning_rate": 0.0010365188009278474,
+      "loss": 0.9218,
+      "step": 7880
+    },
+    {
+      "epoch": 0.8154195948739148,
+      "grad_norm": 0.11380596458911896,
+      "learning_rate": 0.0010202046168379232,
+      "loss": 0.9089,
+      "step": 7890
+    },
+    {
+      "epoch": 0.8164530797850351,
+      "grad_norm": 0.10323087871074677,
+      "learning_rate": 0.0010041472082182382,
+      "loss": 0.9053,
+      "step": 7900
+    },
+    {
+      "epoch": 0.8174865646961554,
+      "grad_norm": 0.10377652198076248,
+      "learning_rate": 0.0009883425335769375,
+      "loss": 0.9167,
+      "step": 7910
+    },
+    {
+      "epoch": 0.8185200496072758,
+      "grad_norm": 0.09777701646089554,
+      "learning_rate": 0.0009727866150328237,
+      "loss": 0.9134,
+      "step": 7920
+    },
+    {
+      "epoch": 0.8195535345183961,
+      "grad_norm": 0.10352311283349991,
+      "learning_rate": 0.0009574755373141626,
+      "loss": 0.9203,
+      "step": 7930
+    },
+    {
+      "epoch": 0.8205870194295163,
+      "grad_norm": 0.1033698320388794,
+      "learning_rate": 0.0009424054467732486,
+      "loss": 0.916,
+      "step": 7940
+    },
+    {
+      "epoch": 0.8216205043406366,
+      "grad_norm": 0.10148289799690247,
+      "learning_rate": 0.0009275725504164789,
+      "loss": 0.9055,
+      "step": 7950
+    },
+    {
+      "epoch": 0.8226539892517569,
+      "grad_norm": 0.1043614000082016,
+      "learning_rate": 0.0009129731149496945,
+      "loss": 0.9103,
+      "step": 7960
+    },
+    {
+      "epoch": 0.8236874741628772,
+      "grad_norm": 0.10638295859098434,
+      "learning_rate": 0.0008986034658385467,
+      "loss": 0.9095,
+      "step": 7970
+    },
+    {
+      "epoch": 0.8247209590739976,
+      "grad_norm": 0.09755829721689224,
+      "learning_rate": 0.0008844599863836532,
+      "loss": 0.911,
+      "step": 7980
+    },
+    {
+      "epoch": 0.8257544439851178,
+      "grad_norm": 0.09516286849975586,
+      "learning_rate": 0.0008705391168103099,
+      "loss": 0.8951,
+      "step": 7990
+    },
+    {
+      "epoch": 0.8267879288962381,
+      "grad_norm": 0.09857280552387238,
+      "learning_rate": 0.0008568373533725312,
+      "loss": 0.9129,
+      "step": 8000
+    },
+    {
+      "epoch": 0.8278214138073584,
+      "grad_norm": 0.09658730775117874,
+      "learning_rate": 0.0008433512474711912,
+      "loss": 0.9123,
+      "step": 8010
+    },
+    {
+      "epoch": 0.8288548987184787,
+      "grad_norm": 0.09653860330581665,
+      "learning_rate": 0.0008300774047860455,
+      "loss": 0.8981,
+      "step": 8020
+    },
+    {
+      "epoch": 0.829888383629599,
+      "grad_norm": 0.10416875779628754,
+      "learning_rate": 0.0008170124844214137,
+      "loss": 0.9173,
+      "step": 8030
+    },
+    {
+      "epoch": 0.8309218685407193,
+      "grad_norm": 0.09936217218637466,
+      "learning_rate": 0.0008041531980653093,
+      "loss": 0.9129,
+      "step": 8040
+    },
+    {
+      "epoch": 0.8319553534518396,
+      "grad_norm": 0.10089723765850067,
+      "learning_rate": 0.0007914963091618035,
+      "loss": 0.9138,
+      "step": 8050
+    },
+    {
+      "epoch": 0.8329888383629599,
+      "grad_norm": 0.09607098996639252,
+      "learning_rate": 0.0007790386320964167,
+      "loss": 0.8986,
+      "step": 8060
+    },
+    {
+      "epoch": 0.8340223232740802,
+      "grad_norm": 0.09406198561191559,
+      "learning_rate": 0.0007667770313943293,
+      "loss": 0.9062,
+      "step": 8070
+    },
+    {
+      "epoch": 0.8350558081852005,
+      "grad_norm": 0.1027892455458641,
+      "learning_rate": 0.0007547084209312148,
+      "loss": 0.9082,
+      "step": 8080
+    },
+    {
+      "epoch": 0.8360892930963207,
+      "grad_norm": 0.09503350406885147,
+      "learning_rate": 0.0007428297631564922,
+      "loss": 0.915,
+      "step": 8090
+    },
+    {
+      "epoch": 0.8371227780074411,
+      "grad_norm": 0.11514163017272949,
+      "learning_rate": 0.000731138068328804,
+      "loss": 0.8963,
+      "step": 8100
+    },
+    {
+      "epoch": 0.8381562629185614,
+      "grad_norm": 0.09855664521455765,
+      "learning_rate": 0.0007196303937635279,
+      "loss": 0.8902,
+      "step": 8110
+    },
+    {
+      "epoch": 0.8391897478296817,
+      "grad_norm": 0.0985339879989624,
+      "learning_rate": 0.000708303843092133,
+      "loss": 0.9087,
+      "step": 8120
+    },
+    {
+      "epoch": 0.840223232740802,
+      "grad_norm": 0.10324376076459885,
+      "learning_rate": 0.0006971555655331906,
+      "loss": 0.9129,
+      "step": 8130
+    },
+    {
+      "epoch": 0.8412567176519223,
+      "grad_norm": 0.10233034193515778,
+      "learning_rate": 0.0006861827551748618,
+      "loss": 0.8974,
+      "step": 8140
+    },
+    {
+      "epoch": 0.8422902025630425,
+      "grad_norm": 0.09102931618690491,
+      "learning_rate": 0.0006753826502686767,
+      "loss": 0.8945,
+      "step": 8150
+    },
+    {
+      "epoch": 0.8433236874741629,
+      "grad_norm": 0.09884581714868546,
+      "learning_rate": 0.0006647525325344291,
+      "loss": 0.9076,
+      "step": 8160
+    },
+    {
+      "epoch": 0.8443571723852832,
+      "grad_norm": 0.10044873505830765,
+      "learning_rate": 0.0006542897264760131,
+      "loss": 0.897,
+      "step": 8170
+    },
+    {
+      "epoch": 0.8453906572964035,
+      "grad_norm": 0.10729029029607773,
+      "learning_rate": 0.000643991598708026,
+      "loss": 0.9061,
+      "step": 8180
+    },
+    {
+      "epoch": 0.8464241422075238,
+      "grad_norm": 0.09865026921033859,
+      "learning_rate": 0.0006338555572929715,
+      "loss": 0.9198,
+      "step": 8190
+    },
+    {
+      "epoch": 0.847457627118644,
+      "grad_norm": 0.0979701355099678,
+      "learning_rate": 0.0006238790510888947,
+      "loss": 0.9027,
+      "step": 8200
+    },
+    {
+      "epoch": 0.8484911120297643,
+      "grad_norm": 0.10607071965932846,
+      "learning_rate": 0.0006140595691072842,
+      "loss": 0.9005,
+      "step": 8210
+    },
+    {
+      "epoch": 0.8495245969408847,
+      "grad_norm": 0.09750446677207947,
+      "learning_rate": 0.0006043946398810819,
+      "loss": 0.8955,
+      "step": 8220
+    },
+    {
+      "epoch": 0.850558081852005,
+      "grad_norm": 0.10503144562244415,
+      "learning_rate": 0.0005948818308426382,
+      "loss": 0.8987,
+      "step": 8230
+    },
+    {
+      "epoch": 0.8515915667631253,
+      "grad_norm": 0.09300017356872559,
+      "learning_rate": 0.0005855187477114587,
+      "loss": 0.8943,
+      "step": 8240
+    },
+    {
+      "epoch": 0.8526250516742455,
+      "grad_norm": 0.09706971049308777,
+      "learning_rate": 0.0005763030338915879,
+      "loss": 0.8942,
+      "step": 8250
+    },
+    {
+      "epoch": 0.8536585365853658,
+      "grad_norm": 0.11037931591272354,
+      "learning_rate": 0.000567232369878477,
+      "loss": 0.8958,
+      "step": 8260
+    },
+    {
+      "epoch": 0.8546920214964862,
+      "grad_norm": 0.09634577482938766,
+      "learning_rate": 0.0005583044726751868,
+      "loss": 0.8976,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8557255064076065,
+      "grad_norm": 0.10242386907339096,
+      "learning_rate": 0.0005495170952177806,
+      "loss": 0.8902,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8567589913187268,
+      "grad_norm": 0.10059966146945953,
+      "learning_rate": 0.00054086802580976,
+      "loss": 0.8896,
+      "step": 8290
+    },
+    {
+      "epoch": 0.857792476229847,
+      "grad_norm": 0.09923160821199417,
+      "learning_rate": 0.0005323550875654025,
+      "loss": 0.9073,
+      "step": 8300
+    },
+    {
+      "epoch": 0.8588259611409673,
+      "grad_norm": 0.09592054039239883,
+      "learning_rate": 0.000523976137861861,
+      "loss": 0.896,
+      "step": 8310
+    },
+    {
+      "epoch": 0.8598594460520876,
+      "grad_norm": 0.09789422899484634,
+      "learning_rate": 0.0005157290677998864,
+      "loss": 0.8878,
+      "step": 8320
+    },
+    {
+      "epoch": 0.860892930963208,
+      "grad_norm": 0.09961704909801483,
+      "learning_rate": 0.0005076118016730388,
+      "loss": 0.9077,
+      "step": 8330
+    },
+    {
+      "epoch": 0.8619264158743283,
+      "grad_norm": 0.10076375305652618,
+      "learning_rate": 0.0004996222964452523,
+      "loss": 0.8979,
+      "step": 8340
+    },
+    {
+      "epoch": 0.8629599007854485,
+      "grad_norm": 0.0950293093919754,
+      "learning_rate": 0.000491758541236623,
+      "loss": 0.8966,
+      "step": 8350
+    },
+    {
+      "epoch": 0.8639933856965688,
+      "grad_norm": 0.10502886027097702,
+      "learning_rate": 0.0004840185568172905,
+      "loss": 0.9094,
+      "step": 8360
+    },
+    {
+      "epoch": 0.8650268706076891,
+      "grad_norm": 0.09783428907394409,
+      "learning_rate": 0.0004764003951092846,
+      "loss": 0.904,
+      "step": 8370
+    },
+    {
+      "epoch": 0.8660603555188094,
+      "grad_norm": 0.10145466774702072,
+      "learning_rate": 0.00046890213869621406,
+      "loss": 0.8912,
+      "step": 8380
+    },
+    {
+      "epoch": 0.8670938404299298,
+      "grad_norm": 0.09995938092470169,
+      "learning_rate": 0.000461521900340671,
+      "loss": 0.91,
+      "step": 8390
+    },
+    {
+      "epoch": 0.86812732534105,
+      "grad_norm": 0.09882347285747528,
+      "learning_rate": 0.0004542578225092324,
+      "loss": 0.8935,
+      "step": 8400
+    },
+    {
+      "epoch": 0.8691608102521703,
+      "grad_norm": 0.1028551384806633,
+      "learning_rate": 0.00044710807690493667,
+      "loss": 0.9005,
+      "step": 8410
+    },
+    {
+      "epoch": 0.8701942951632906,
+      "grad_norm": 0.09993860870599747,
+      "learning_rate": 0.00044007086400711984,
+      "loss": 0.8883,
+      "step": 8420
+    },
+    {
+      "epoch": 0.8712277800744109,
+      "grad_norm": 0.09587843716144562,
+      "learning_rate": 0.0004331444126184934,
+      "loss": 0.896,
+      "step": 8430
+    },
+    {
+      "epoch": 0.8722612649855312,
+      "grad_norm": 0.10601107031106949,
+      "learning_rate": 0.0004263269794193516,
+      "loss": 0.9027,
+      "step": 8440
+    },
+    {
+      "epoch": 0.8732947498966515,
+      "grad_norm": 0.09365765005350113,
+      "learning_rate": 0.00041961684852879507,
+      "loss": 0.8928,
+      "step": 8450
+    },
+    {
+      "epoch": 0.8743282348077718,
+      "grad_norm": 0.10556492209434509,
+      "learning_rate": 0.0004130123310728602,
+      "loss": 0.8872,
+      "step": 8460
+    },
+    {
+      "epoch": 0.8753617197188921,
+      "grad_norm": 0.09618480503559113,
+      "learning_rate": 0.0004065117647594465,
+      "loss": 0.8917,
+      "step": 8470
+    },
+    {
+      "epoch": 0.8763952046300124,
+      "grad_norm": 0.100033700466156,
+      "learning_rate": 0.0004001135134599342,
+      "loss": 0.8949,
+      "step": 8480
+    },
+    {
+      "epoch": 0.8774286895411327,
+      "grad_norm": 0.09465377777814865,
+      "learning_rate": 0.00039381596679738596,
+      "loss": 0.8931,
+      "step": 8490
+    },
+    {
+      "epoch": 0.8784621744522529,
+      "grad_norm": 0.09475496411323547,
+      "learning_rate": 0.00038761753974123146,
+      "loss": 0.8856,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8794956593633733,
+      "grad_norm": 0.09431330114603043,
+      "learning_rate": 0.0003815166722083307,
+      "loss": 0.8921,
+      "step": 8510
+    },
+    {
+      "epoch": 0.8805291442744936,
+      "grad_norm": 0.11336002498865128,
+      "learning_rate": 0.00037551182867031637,
+      "loss": 0.8944,
+      "step": 8520
+    },
+    {
+      "epoch": 0.8815626291856139,
+      "grad_norm": 0.09444057941436768,
+      "learning_rate": 0.00036960149776711637,
+      "loss": 0.8907,
+      "step": 8530
+    },
+    {
+      "epoch": 0.8825961140967342,
+      "grad_norm": 0.09592239558696747,
+      "learning_rate": 0.0003637841919265597,
+      "loss": 0.8993,
+      "step": 8540
+    },
+    {
+      "epoch": 0.8836295990078545,
+      "grad_norm": 0.09695941209793091,
+      "learning_rate": 0.0003580584469899685,
+      "loss": 0.8948,
+      "step": 8550
+    },
+    {
+      "epoch": 0.8846630839189747,
+      "grad_norm": 0.10035042464733124,
+      "learning_rate": 0.0003524228218436444,
+      "loss": 0.8897,
+      "step": 8560
+    },
+    {
+      "epoch": 0.8856965688300951,
+      "grad_norm": 0.11594710499048233,
+      "learning_rate": 0.0003468758980561539,
+      "loss": 0.896,
+      "step": 8570
+    },
+    {
+      "epoch": 0.8867300537412154,
+      "grad_norm": 0.10346604883670807,
+      "learning_rate": 0.00034141627952132355,
+      "loss": 0.8885,
+      "step": 8580
+    },
+    {
+      "epoch": 0.8877635386523357,
+      "grad_norm": 0.09101969003677368,
+      "learning_rate": 0.0003360425921068532,
+      "loss": 0.8867,
+      "step": 8590
+    },
+    {
+      "epoch": 0.888797023563456,
+      "grad_norm": 0.09395504742860794,
+      "learning_rate": 0.00033075348330846094,
+      "loss": 0.8871,
+      "step": 8600
+    },
+    {
+      "epoch": 0.8898305084745762,
+      "grad_norm": 0.09610182791948318,
+      "learning_rate": 0.00032554762190947073,
+      "loss": 0.8913,
+      "step": 8610
+    },
+    {
+      "epoch": 0.8908639933856966,
+      "grad_norm": 0.11420192569494247,
+      "learning_rate": 0.0003204236976457585,
+      "loss": 0.8897,
+      "step": 8620
+    },
+    {
+      "epoch": 0.8918974782968169,
+      "grad_norm": 0.10842256993055344,
+      "learning_rate": 0.00031538042087597126,
+      "loss": 0.8887,
+      "step": 8630
+    },
+    {
+      "epoch": 0.8929309632079372,
+      "grad_norm": 0.10306430608034134,
+      "learning_rate": 0.00031041652225693745,
+      "loss": 0.8882,
+      "step": 8640
+    },
+    {
+      "epoch": 0.8939644481190575,
+      "grad_norm": 0.09563726931810379,
+      "learning_rate": 0.00030553075242418533,
+      "loss": 0.8884,
+      "step": 8650
+    },
+    {
+      "epoch": 0.8949979330301777,
+      "grad_norm": 0.09344979375600815,
+      "learning_rate": 0.0003007218816774904,
+      "loss": 0.8853,
+      "step": 8660
+    },
+    {
+      "epoch": 0.896031417941298,
+      "grad_norm": 0.09084004163742065,
+      "learning_rate": 0.0002959886996713722,
+      "loss": 0.8844,
+      "step": 8670
+    },
+    {
+      "epoch": 0.8970649028524184,
+      "grad_norm": 0.0981561467051506,
+      "learning_rate": 0.0002913300151104616,
+      "loss": 0.8987,
+      "step": 8680
+    },
+    {
+      "epoch": 0.8980983877635387,
+      "grad_norm": 0.09380181133747101,
+      "learning_rate": 0.00028674465544966445,
+      "loss": 0.8881,
+      "step": 8690
+    },
+    {
+      "epoch": 0.899131872674659,
+      "grad_norm": 0.10047593712806702,
+      "learning_rate": 0.00028223146659904236,
+      "loss": 0.8949,
+      "step": 8700
+    },
+    {
+      "epoch": 0.9001653575857792,
+      "grad_norm": 0.10739541053771973,
+      "learning_rate": 0.0002777893126333406,
+      "loss": 0.8969,
+      "step": 8710
+    },
+    {
+      "epoch": 0.9011988424968995,
+      "grad_norm": 0.09939514100551605,
+      "learning_rate": 0.0002734170755060864,
+      "loss": 0.8808,
+      "step": 8720
+    },
+    {
+      "epoch": 0.9022323274080198,
+      "grad_norm": 0.09798012673854828,
+      "learning_rate": 0.0002691136547681875,
+      "loss": 0.8898,
+      "step": 8730
+    },
+    {
+      "epoch": 0.9032658123191402,
+      "grad_norm": 0.09667760878801346,
+      "learning_rate": 0.00026487796729096044,
+      "loss": 0.8996,
+      "step": 8740
+    },
+    {
+      "epoch": 0.9042992972302605,
+      "grad_norm": 0.09401620179414749,
+      "learning_rate": 0.0002607089469935173,
+      "loss": 0.8797,
+      "step": 8750
+    },
+    {
+      "epoch": 0.9053327821413807,
+      "grad_norm": 0.09637662023305893,
+      "learning_rate": 0.0002566055445744437,
+      "loss": 0.885,
+      "step": 8760
+    },
+    {
+      "epoch": 0.906366267052501,
+      "grad_norm": 0.0979810431599617,
+      "learning_rate": 0.00025256672724770034,
+      "loss": 0.8765,
+      "step": 8770
+    },
+    {
+      "epoch": 0.9073997519636213,
+      "grad_norm": 0.09577486664056778,
+      "learning_rate": 0.0002485914784826802,
+      "loss": 0.878,
+      "step": 8780
+    },
+    {
+      "epoch": 0.9084332368747416,
+      "grad_norm": 0.09690549224615097,
+      "learning_rate": 0.0002446787977483582,
+      "loss": 0.8951,
+      "step": 8790
+    },
+    {
+      "epoch": 0.909466721785862,
+      "grad_norm": 0.09132607281208038,
+      "learning_rate": 0.00024082770026146767,
+      "loss": 0.8922,
+      "step": 8800
+    },
+    {
+      "epoch": 0.9105002066969822,
+      "grad_norm": 0.10059286653995514,
+      "learning_rate": 0.00023703721673863946,
+      "loss": 0.8961,
+      "step": 8810
+    },
+    {
+      "epoch": 0.9115336916081025,
+      "grad_norm": 0.09476503729820251,
+      "learning_rate": 0.00023330639315244326,
+      "loss": 0.8891,
+      "step": 8820
+    },
+    {
+      "epoch": 0.9125671765192228,
+      "grad_norm": 0.097599096596241,
+      "learning_rate": 0.00022963429049126816,
+      "loss": 0.8877,
+      "step": 8830
+    },
+    {
+      "epoch": 0.9136006614303431,
+      "grad_norm": 0.09217467904090881,
+      "learning_rate": 0.00022601998452298264,
+      "loss": 0.893,
+      "step": 8840
+    },
+    {
+      "epoch": 0.9146341463414634,
+      "grad_norm": 0.11072806268930435,
+      "learning_rate": 0.00022246256556231452,
+      "loss": 0.8943,
+      "step": 8850
+    },
+    {
+      "epoch": 0.9156676312525837,
+      "grad_norm": 0.10245074331760406,
+      "learning_rate": 0.00021896113824189194,
+      "loss": 0.9008,
+      "step": 8860
+    },
+    {
+      "epoch": 0.916701116163704,
+      "grad_norm": 0.0995858684182167,
+      "learning_rate": 0.00021551482128688842,
+      "loss": 0.8831,
+      "step": 8870
+    },
+    {
+      "epoch": 0.9177346010748243,
+      "grad_norm": 0.09560049325227737,
+      "learning_rate": 0.0002121227472932145,
+      "loss": 0.8978,
+      "step": 8880
+    },
+    {
+      "epoch": 0.9187680859859446,
+      "grad_norm": 0.09800171852111816,
+      "learning_rate": 0.00020878406250920078,
+      "loss": 0.8993,
+      "step": 8890
+    },
+    {
+      "epoch": 0.9198015708970649,
+      "grad_norm": 0.09587955474853516,
+      "learning_rate": 0.00020549792662071688,
+      "loss": 0.8947,
+      "step": 8900
+    },
+    {
+      "epoch": 0.9208350558081853,
+      "grad_norm": 0.09761986881494522,
+      "learning_rate": 0.00020226351253967265,
+      "loss": 0.8968,
+      "step": 8910
+    },
+    {
+      "epoch": 0.9218685407193055,
+      "grad_norm": 0.10232323408126831,
+      "learning_rate": 0.00019908000619584844,
+      "loss": 0.8743,
+      "step": 8920
+    },
+    {
+      "epoch": 0.9229020256304258,
+      "grad_norm": 0.09598308801651001,
+      "learning_rate": 0.00019594660633200127,
+      "loss": 0.8886,
+      "step": 8930
+    },
+    {
+      "epoch": 0.9239355105415461,
+      "grad_norm": 0.09552961587905884,
+      "learning_rate": 0.0001928625243021967,
+      "loss": 0.898,
+      "step": 8940
+    },
+    {
+      "epoch": 0.9249689954526664,
+      "grad_norm": 0.09763569384813309,
+      "learning_rate": 0.000189826983873314,
+      "loss": 0.8872,
+      "step": 8950
+    },
+    {
+      "epoch": 0.9260024803637867,
+      "grad_norm": 0.10132598876953125,
+      "learning_rate": 0.00018683922102967604,
+      "loss": 0.8844,
+      "step": 8960
+    },
+    {
+      "epoch": 0.927035965274907,
+      "grad_norm": 0.10523135215044022,
+      "learning_rate": 0.00018389848378075436,
+      "loss": 0.8886,
+      "step": 8970
+    },
+    {
+      "epoch": 0.9280694501860273,
+      "grad_norm": 0.10074016451835632,
+      "learning_rate": 0.0001810040319719,
+      "loss": 0.8868,
+      "step": 8980
+    },
+    {
+      "epoch": 0.9291029350971476,
+      "grad_norm": 0.09313800185918808,
+      "learning_rate": 0.00017815513709805425,
+      "loss": 0.8944,
+      "step": 8990
+    },
+    {
+      "epoch": 0.9301364200082679,
+      "grad_norm": 0.09687028080224991,
+      "learning_rate": 0.00017535108212039094,
+      "loss": 0.8764,
+      "step": 9000
+    },
+    {
+      "epoch": 0.9311699049193882,
+      "grad_norm": 0.09959317743778229,
+      "learning_rate": 0.00017259116128584488,
+      "loss": 0.8879,
+      "step": 9010
+    },
+    {
+      "epoch": 0.9322033898305084,
+      "grad_norm": 0.09753499180078506,
+      "learning_rate": 0.0001698746799494807,
+      "loss": 0.8878,
+      "step": 9020
+    },
+    {
+      "epoch": 0.9332368747416288,
+      "grad_norm": 0.10160617530345917,
+      "learning_rate": 0.00016720095439965764,
+      "loss": 0.8878,
+      "step": 9030
+    },
+    {
+      "epoch": 0.9342703596527491,
+      "grad_norm": 0.09427917003631592,
+      "learning_rate": 0.00016456931168594587,
+      "loss": 0.8917,
+      "step": 9040
+    },
+    {
+      "epoch": 0.9353038445638694,
+      "grad_norm": 0.09126557409763336,
+      "learning_rate": 0.0001619790894497517,
+      "loss": 0.8756,
+      "step": 9050
+    },
+    {
+      "epoch": 0.9363373294749897,
+      "grad_norm": 0.10263457894325256,
+      "learning_rate": 0.0001594296357576083,
+      "loss": 0.8968,
+      "step": 9060
+    },
+    {
+      "epoch": 0.9373708143861099,
+      "grad_norm": 0.09004776179790497,
+      "learning_rate": 0.00015692030893709055,
+      "loss": 0.8747,
+      "step": 9070
+    },
+    {
+      "epoch": 0.9384042992972302,
+      "grad_norm": 0.09706980735063553,
+      "learning_rate": 0.0001544504774153122,
+      "loss": 0.8782,
+      "step": 9080
+    },
+    {
+      "epoch": 0.9394377842083506,
+      "grad_norm": 0.11443266272544861,
+      "learning_rate": 0.00015201951955996554,
+      "loss": 0.8954,
+      "step": 9090
+    },
+    {
+      "epoch": 0.9404712691194709,
+      "grad_norm": 0.09983173757791519,
+      "learning_rate": 0.00014962682352286227,
+      "loss": 0.8839,
+      "step": 9100
+    },
+    {
+      "epoch": 0.9415047540305912,
+      "grad_norm": 0.09485773742198944,
+      "learning_rate": 0.00014727178708593758,
+      "loss": 0.8873,
+      "step": 9110
+    },
+    {
+      "epoch": 0.9425382389417114,
+      "grad_norm": 0.09533528983592987,
+      "learning_rate": 0.00014495381750967773,
+      "loss": 0.8803,
+      "step": 9120
+    },
+    {
+      "epoch": 0.9435717238528317,
+      "grad_norm": 0.09723920375108719,
+      "learning_rate": 0.00014267233138393328,
+      "loss": 0.894,
+      "step": 9130
+    },
+    {
+      "epoch": 0.944605208763952,
+      "grad_norm": 0.10340782254934311,
+      "learning_rate": 0.00014042675448108058,
+      "loss": 0.8876,
+      "step": 9140
+    },
+    {
+      "epoch": 0.9456386936750724,
+      "grad_norm": 0.10088169574737549,
+      "learning_rate": 0.00013821652161149427,
+      "loss": 0.8797,
+      "step": 9150
+    },
+    {
+      "epoch": 0.9466721785861927,
+      "grad_norm": 0.09791452437639236,
+      "learning_rate": 0.0001360410764812946,
+      "loss": 0.8897,
+      "step": 9160
+    },
+    {
+      "epoch": 0.9477056634973129,
+      "grad_norm": 0.0989762470126152,
+      "learning_rate": 0.00013389987155233373,
+      "loss": 0.885,
+      "step": 9170
+    },
+    {
+      "epoch": 0.9487391484084332,
+      "grad_norm": 0.09686373174190521,
+      "learning_rate": 0.0001317923679043859,
+      "loss": 0.88,
+      "step": 9180
+    },
+    {
+      "epoch": 0.9497726333195535,
+      "grad_norm": 0.094807930290699,
+      "learning_rate": 0.00012971803509950622,
+      "loss": 0.8877,
+      "step": 9190
+    },
+    {
+      "epoch": 0.9508061182306738,
+      "grad_norm": 0.0951213538646698,
+      "learning_rate": 0.000127676351048525,
+      "loss": 0.8858,
+      "step": 9200
+    },
+    {
+      "epoch": 0.9518396031417942,
+      "grad_norm": 0.09922289848327637,
+      "learning_rate": 0.00012566680187964274,
+      "loss": 0.8826,
+      "step": 9210
+    },
+    {
+      "epoch": 0.9528730880529144,
+      "grad_norm": 0.09398606419563293,
+      "learning_rate": 0.0001236888818090938,
+      "loss": 0.8817,
+      "step": 9220
+    },
+    {
+      "epoch": 0.9539065729640347,
+      "grad_norm": 0.1025305688381195,
+      "learning_rate": 0.0001217420930138456,
+      "loss": 0.8809,
+      "step": 9230
+    },
+    {
+      "epoch": 0.954940057875155,
+      "grad_norm": 0.0998004898428917,
+      "learning_rate": 0.00011982594550630144,
+      "loss": 0.8878,
+      "step": 9240
+    },
+    {
+      "epoch": 0.9559735427862753,
+      "grad_norm": 0.09735240042209625,
+      "learning_rate": 0.00011793995701097545,
+      "loss": 0.8713,
+      "step": 9250
+    },
+    {
+      "epoch": 0.9570070276973957,
+      "grad_norm": 0.09462594240903854,
+      "learning_rate": 0.00011608365284310853,
+      "loss": 0.8932,
+      "step": 9260
+    },
+    {
+      "epoch": 0.958040512608516,
+      "grad_norm": 0.0939006358385086,
+      "learning_rate": 0.00011425656578919491,
+      "loss": 0.8865,
+      "step": 9270
+    },
+    {
+      "epoch": 0.9590739975196362,
+      "grad_norm": 0.10327833890914917,
+      "learning_rate": 0.00011245823598938918,
+      "loss": 0.885,
+      "step": 9280
+    },
+    {
+      "epoch": 0.9601074824307565,
+      "grad_norm": 0.10130874067544937,
+      "learning_rate": 0.00011068821082176394,
+      "loss": 0.8737,
+      "step": 9290
+    },
+    {
+      "epoch": 0.9611409673418768,
+      "grad_norm": 0.0958571583032608,
+      "learning_rate": 0.00010894604478838941,
+      "loss": 0.8836,
+      "step": 9300
+    },
+    {
+      "epoch": 0.9621744522529971,
+      "grad_norm": 0.10654882341623306,
+      "learning_rate": 0.00010723129940320598,
+      "loss": 0.8808,
+      "step": 9310
+    },
+    {
+      "epoch": 0.9632079371641175,
+      "grad_norm": 0.10343246906995773,
+      "learning_rate": 0.00010554354308166149,
+      "loss": 0.8838,
+      "step": 9320
+    },
+    {
+      "epoch": 0.9642414220752377,
+      "grad_norm": 0.09667105227708817,
+      "learning_rate": 0.00010388235103208581,
+      "loss": 0.8944,
+      "step": 9330
+    },
+    {
+      "epoch": 0.965274906986358,
+      "grad_norm": 0.09733343869447708,
+      "learning_rate": 0.00010224730514877476,
+      "loss": 0.8807,
+      "step": 9340
+    },
+    {
+      "epoch": 0.9663083918974783,
+      "grad_norm": 0.09809593111276627,
+      "learning_rate": 0.0001006379939067572,
+      "loss": 0.8824,
+      "step": 9350
+    },
+    {
+      "epoch": 0.9673418768085986,
+      "grad_norm": 0.09950044006109238,
+      "learning_rate": 9.905401225821806e-05,
+      "loss": 0.8895,
+      "step": 9360
+    },
+    {
+      "epoch": 0.9683753617197189,
+      "grad_norm": 0.09585009515285492,
+      "learning_rate": 9.749496153055196e-05,
+      "loss": 0.871,
+      "step": 9370
+    },
+    {
+      "epoch": 0.9694088466308391,
+      "grad_norm": 0.09470534324645996,
+      "learning_rate": 9.596044932602107e-05,
+      "loss": 0.8805,
+      "step": 9380
+    },
+    {
+      "epoch": 0.9704423315419595,
+      "grad_norm": 0.09828654676675797,
+      "learning_rate": 9.445008942299262e-05,
+      "loss": 0.8796,
+      "step": 9390
+    },
+    {
+      "epoch": 0.9714758164530798,
+      "grad_norm": 0.09540653228759766,
+      "learning_rate": 9.296350167873062e-05,
+      "loss": 0.8769,
+      "step": 9400
+    },
+    {
+      "epoch": 0.9725093013642001,
+      "grad_norm": 0.09175027906894684,
+      "learning_rate": 9.150031193371763e-05,
+      "loss": 0.8823,
+      "step": 9410
+    },
+    {
+      "epoch": 0.9735427862753204,
+      "grad_norm": 0.09416258335113525,
+      "learning_rate": 9.00601519174826e-05,
+      "loss": 0.8826,
+      "step": 9420
+    },
+    {
+      "epoch": 0.9745762711864406,
+      "grad_norm": 0.102226123213768,
+      "learning_rate": 8.864265915591077e-05,
+      "loss": 0.8813,
+      "step": 9430
+    },
+    {
+      "epoch": 0.975609756097561,
+      "grad_norm": 0.1030995100736618,
+      "learning_rate": 8.724747688001238e-05,
+      "loss": 0.89,
+      "step": 9440
+    },
+    {
+      "epoch": 0.9766432410086813,
+      "grad_norm": 0.09412838518619537,
+      "learning_rate": 8.587425393612774e-05,
+      "loss": 0.8869,
+      "step": 9450
+    },
+    {
+      "epoch": 0.9776767259198016,
+      "grad_norm": 0.09783417731523514,
+      "learning_rate": 8.45226446975449e-05,
+      "loss": 0.8745,
+      "step": 9460
+    },
+    {
+      "epoch": 0.9787102108309219,
+      "grad_norm": 0.09457999467849731,
+      "learning_rate": 8.31923089775091e-05,
+      "loss": 0.8854,
+      "step": 9470
+    },
+    {
+      "epoch": 0.9797436957420421,
+      "grad_norm": 0.0977085754275322,
+      "learning_rate": 8.188291194360097e-05,
+      "loss": 0.8743,
+      "step": 9480
+    },
+    {
+      "epoch": 0.9807771806531624,
+      "grad_norm": 0.09691467881202698,
+      "learning_rate": 8.059412403346259e-05,
+      "loss": 0.8798,
+      "step": 9490
+    },
+    {
+      "epoch": 0.9818106655642828,
+      "grad_norm": 0.09877024590969086,
+      "learning_rate": 7.932562087184977e-05,
+      "loss": 0.8785,
+      "step": 9500
+    },
+    {
+      "epoch": 0.9828441504754031,
+      "grad_norm": 0.09752603620290756,
+      "learning_rate": 7.807708318899015e-05,
+      "loss": 0.8787,
+      "step": 9510
+    },
+    {
+      "epoch": 0.9838776353865234,
+      "grad_norm": 0.10295706242322922,
+      "learning_rate": 7.684819674022599e-05,
+      "loss": 0.8782,
+      "step": 9520
+    },
+    {
+      "epoch": 0.9849111202976436,
+      "grad_norm": 0.09535824507474899,
+      "learning_rate": 7.563865222692196e-05,
+      "loss": 0.8861,
+      "step": 9530
+    },
+    {
+      "epoch": 0.9859446052087639,
+      "grad_norm": 0.09367913007736206,
+      "learning_rate": 7.444814521861768e-05,
+      "loss": 0.8791,
+      "step": 9540
+    },
+    {
+      "epoch": 0.9869780901198842,
+      "grad_norm": 0.10222353786230087,
+      "learning_rate": 7.327637607640558e-05,
+      "loss": 0.884,
+      "step": 9550
+    },
+    {
+      "epoch": 0.9880115750310046,
+      "grad_norm": 0.1052263155579567,
+      "learning_rate": 7.212304987751476e-05,
+      "loss": 0.8766,
+      "step": 9560
+    },
+    {
+      "epoch": 0.9890450599421249,
+      "grad_norm": 0.09064412862062454,
+      "learning_rate": 7.098787634108176e-05,
+      "loss": 0.8846,
+      "step": 9570
+    },
+    {
+      "epoch": 0.9900785448532451,
+      "grad_norm": 0.09217038005590439,
+      "learning_rate": 6.98705697550898e-05,
+      "loss": 0.878,
+      "step": 9580
+    },
+    {
+      "epoch": 0.9911120297643654,
+      "grad_norm": 0.08925709873437881,
+      "learning_rate": 6.87708489044578e-05,
+      "loss": 0.8659,
+      "step": 9590
+    },
+    {
+      "epoch": 0.9921455146754857,
+      "grad_norm": 0.09898435324430466,
+      "learning_rate": 6.768843700026141e-05,
+      "loss": 0.8891,
+      "step": 9600
+    },
+    {
+      "epoch": 0.993178999586606,
+      "grad_norm": 0.09614672511816025,
+      "learning_rate": 6.662306161006783e-05,
+      "loss": 0.8828,
+      "step": 9610
+    },
+    {
+      "epoch": 0.9942124844977264,
+      "grad_norm": 0.09297246485948563,
+      "learning_rate": 6.557445458936731e-05,
+      "loss": 0.8905,
+      "step": 9620
+    },
+    {
+      "epoch": 0.9952459694088466,
+      "grad_norm": 0.10573872923851013,
+      "learning_rate": 6.45423520140839e-05,
+      "loss": 0.8876,
+      "step": 9630
+    },
+    {
+      "epoch": 0.9962794543199669,
+      "grad_norm": 0.09352509677410126,
+      "learning_rate": 6.352649411414819e-05,
+      "loss": 0.8856,
+      "step": 9640
+    },
+    {
+      "epoch": 0.9973129392310872,
+      "grad_norm": 0.0931333377957344,
+      "learning_rate": 6.252662520811583e-05,
+      "loss": 0.8848,
+      "step": 9650
+    },
+    {
+      "epoch": 0.9983464241422075,
+      "grad_norm": 0.10176722705364227,
+      "learning_rate": 6.154249363881505e-05,
+      "loss": 0.8792,
+      "step": 9660
+    },
+    {
+      "epoch": 0.9993799090533279,
+      "grad_norm": 0.09999672323465347,
+      "learning_rate": 6.057385171000694e-05,
+      "loss": 0.8789,
+      "step": 9670
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9676,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.24620217896713e+16,
+  "train_batch_size": 1024,
+  "trial_name": null,
+  "trial_params": null
+}