{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9990717600856835,
  "eval_steps": 500,
  "global_step": 7002,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004284184219921457,
      "grad_norm": 7.21875,
      "learning_rate": 1.0688836104513063e-05,
      "loss": 2.91,
      "mean_token_accuracy": 0.5702525054415067,
      "step": 10
    },
    {
      "epoch": 0.008568368439842914,
      "grad_norm": 3.453125,
      "learning_rate": 2.2565320665083135e-05,
      "loss": 1.6569,
      "mean_token_accuracy": 0.6329187631607056,
      "step": 20
    },
    {
      "epoch": 0.01285255265976437,
      "grad_norm": 2.203125,
      "learning_rate": 3.44418052256532e-05,
      "loss": 1.4104,
      "mean_token_accuracy": 0.657142640153567,
      "step": 30
    },
    {
      "epoch": 0.017136736879685827,
      "grad_norm": 2.375,
      "learning_rate": 4.631828978622328e-05,
      "loss": 1.4205,
      "mean_token_accuracy": 0.657102554043134,
      "step": 40
    },
    {
      "epoch": 0.021420921099607283,
      "grad_norm": 2.0625,
      "learning_rate": 5.819477434679335e-05,
      "loss": 1.4479,
      "mean_token_accuracy": 0.6437717239061992,
      "step": 50
    },
    {
      "epoch": 0.02570510531952874,
      "grad_norm": 1.921875,
      "learning_rate": 7.007125890736342e-05,
      "loss": 1.4744,
      "mean_token_accuracy": 0.6392371306816736,
      "step": 60
    },
    {
      "epoch": 0.029989289539450195,
      "grad_norm": 1.8984375,
      "learning_rate": 8.194774346793349e-05,
      "loss": 1.4946,
      "mean_token_accuracy": 0.6394953320423762,
      "step": 70
    },
    {
      "epoch": 0.034273473759371655,
      "grad_norm": 1.9140625,
      "learning_rate": 9.382422802850355e-05,
      "loss": 1.5123,
      "mean_token_accuracy": 0.6348034431536992,
      "step": 80
    },
    {
      "epoch": 0.03855765797929311,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00010570071258907364,
      "loss": 1.5446,
      "mean_token_accuracy": 0.6331378062566121,
      "step": 90
    },
    {
      "epoch": 0.042841842199214566,
      "grad_norm": 1.671875,
      "learning_rate": 0.00011757719714964371,
      "loss": 1.6083,
      "mean_token_accuracy": 0.6170119682947794,
      "step": 100
    },
    {
      "epoch": 0.047126026419136026,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00012945368171021377,
      "loss": 1.6014,
      "mean_token_accuracy": 0.6203826546669007,
      "step": 110
    },
    {
      "epoch": 0.05141021063905748,
      "grad_norm": 6.375,
      "learning_rate": 0.00014133016627078385,
      "loss": 1.65,
      "mean_token_accuracy": 0.6120437135299047,
      "step": 120
    },
    {
      "epoch": 0.05569439485897894,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00015320665083135392,
      "loss": 1.7805,
      "mean_token_accuracy": 0.5932602127393086,
      "step": 130
    },
    {
      "epoch": 0.05997857907890039,
      "grad_norm": 1.484375,
      "learning_rate": 0.000165083135391924,
      "loss": 1.6786,
      "mean_token_accuracy": 0.605377835035324,
      "step": 140
    },
    {
      "epoch": 0.06426276329882184,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00017695961995249407,
      "loss": 1.6794,
      "mean_token_accuracy": 0.6040138930082322,
      "step": 150
    },
    {
      "epoch": 0.06854694751874331,
      "grad_norm": 1.5625,
      "learning_rate": 0.00018883610451306412,
      "loss": 1.6891,
      "mean_token_accuracy": 0.6048624157905579,
      "step": 160
    },
    {
      "epoch": 0.07283113173866476,
      "grad_norm": 1.5625,
      "learning_rate": 0.00020071258907363422,
      "loss": 1.7496,
      "mean_token_accuracy": 0.5943053344885508,
      "step": 170
    },
    {
      "epoch": 0.07711531595858621,
      "grad_norm": 1.6796875,
      "learning_rate": 0.00021258907363420426,
      "loss": 1.8166,
      "mean_token_accuracy": 0.5864472662409147,
      "step": 180
    },
    {
      "epoch": 0.08139950017850768,
      "grad_norm": 1.5,
      "learning_rate": 0.00022446555819477434,
      "loss": 1.804,
      "mean_token_accuracy": 0.5849504172801971,
      "step": 190
    },
    {
      "epoch": 0.08568368439842913,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00023634204275534444,
      "loss": 1.8061,
      "mean_token_accuracy": 0.5850104331970215,
      "step": 200
    },
    {
      "epoch": 0.08996786861835059,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0002482185273159145,
      "loss": 1.8503,
      "mean_token_accuracy": 0.5780546754598618,
      "step": 210
    },
    {
      "epoch": 0.09425205283827205,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00026009501187648456,
      "loss": 1.7982,
      "mean_token_accuracy": 0.5852496673663458,
      "step": 220
    },
    {
      "epoch": 0.0985362370581935,
      "grad_norm": 1.25,
      "learning_rate": 0.00027197149643705463,
      "loss": 1.8589,
      "mean_token_accuracy": 0.5725395048658053,
      "step": 230
    },
    {
      "epoch": 0.10282042127811496,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002838479809976247,
      "loss": 1.8616,
      "mean_token_accuracy": 0.5743008524179458,
      "step": 240
    },
    {
      "epoch": 0.10710460549803641,
      "grad_norm": 5.71875,
      "learning_rate": 0.0002957244655581948,
      "loss": 1.8569,
      "mean_token_accuracy": 0.5772680620352427,
      "step": 250
    },
    {
      "epoch": 0.11138878971795788,
      "grad_norm": 1.234375,
      "learning_rate": 0.00030760095011876486,
      "loss": 1.883,
      "mean_token_accuracy": 0.5712258398532868,
      "step": 260
    },
    {
      "epoch": 0.11567297393787933,
      "grad_norm": 1.328125,
      "learning_rate": 0.00031947743467933493,
      "loss": 1.9089,
      "mean_token_accuracy": 0.5683296079436938,
      "step": 270
    },
    {
      "epoch": 0.11995715815780078,
      "grad_norm": 1.140625,
      "learning_rate": 0.000331353919239905,
      "loss": 1.8845,
      "mean_token_accuracy": 0.5682273174325625,
      "step": 280
    },
    {
      "epoch": 0.12424134237772225,
      "grad_norm": 1.046875,
      "learning_rate": 0.000343230403800475,
      "loss": 1.8897,
      "mean_token_accuracy": 0.5696620404720306,
      "step": 290
    },
    {
      "epoch": 0.12852552659764369,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00035510688836104515,
      "loss": 1.8809,
      "mean_token_accuracy": 0.571748511493206,
      "step": 300
    },
    {
      "epoch": 0.13280971081756515,
      "grad_norm": 2.640625,
      "learning_rate": 0.0003669833729216152,
      "loss": 1.9614,
      "mean_token_accuracy": 0.566138303776582,
      "step": 310
    },
    {
      "epoch": 0.13709389503748662,
      "grad_norm": 1.109375,
      "learning_rate": 0.00037885985748218525,
      "loss": 1.969,
      "mean_token_accuracy": 0.5570171405871709,
      "step": 320
    },
    {
      "epoch": 0.14137807925740806,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003907363420427554,
      "loss": 1.902,
      "mean_token_accuracy": 0.5694692318638166,
      "step": 330
    },
    {
      "epoch": 0.14566226347732952,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00040261282660332545,
      "loss": 1.9013,
      "mean_token_accuracy": 0.5666690587997436,
      "step": 340
    },
    {
      "epoch": 0.149946447697251,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00041448931116389547,
      "loss": 1.9115,
      "mean_token_accuracy": 0.5653235624233882,
      "step": 350
    },
    {
      "epoch": 0.15423063191717243,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00042636579572446554,
      "loss": 1.9273,
      "mean_token_accuracy": 0.5625237961610158,
      "step": 360
    },
    {
      "epoch": 0.1585148161370939,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00043824228028503567,
      "loss": 1.9705,
      "mean_token_accuracy": 0.5571655298272769,
      "step": 370
    },
    {
      "epoch": 0.16279900035701536,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004501187648456057,
      "loss": 1.9421,
      "mean_token_accuracy": 0.5626261870066325,
      "step": 380
    },
    {
      "epoch": 0.1670831845769368,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00046199524940617576,
      "loss": 1.9847,
      "mean_token_accuracy": 0.5572134067614873,
      "step": 390
    },
    {
      "epoch": 0.17136736879685827,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004738717339667459,
      "loss": 1.9701,
      "mean_token_accuracy": 0.5576820741097133,
      "step": 400
    },
    {
      "epoch": 0.17565155301677973,
      "grad_norm": 0.875,
      "learning_rate": 0.0004857482185273159,
      "loss": 1.9652,
      "mean_token_accuracy": 0.5562737271189689,
      "step": 410
    },
    {
      "epoch": 0.17993573723670117,
      "grad_norm": 0.984375,
      "learning_rate": 0.000497624703087886,
      "loss": 1.9838,
      "mean_token_accuracy": 0.553861757616202,
      "step": 420
    },
    {
      "epoch": 0.18421992145662264,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004999981769212751,
      "loss": 1.9842,
      "mean_token_accuracy": 0.5521314447124799,
      "step": 430
    },
    {
      "epoch": 0.1885041056765441,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0004999907707095249,
      "loss": 1.9573,
      "mean_token_accuracy": 0.5590350007017454,
      "step": 440
    },
    {
      "epoch": 0.19278828989646554,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0004999776675909755,
      "loss": 1.9179,
      "mean_token_accuracy": 0.5650409559408823,
      "step": 450
    },
    {
      "epoch": 0.197072474116387,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000499958867864227,
      "loss": 1.9485,
      "mean_token_accuracy": 0.5609820206960042,
      "step": 460
    },
    {
      "epoch": 0.20135665833630847,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0004999343719576963,
      "loss": 1.9483,
      "mean_token_accuracy": 0.5612422108650208,
      "step": 470
    },
    {
      "epoch": 0.2056408425562299,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0004999041804296074,
      "loss": 1.9683,
      "mean_token_accuracy": 0.561284634967645,
      "step": 480
    },
    {
      "epoch": 0.20992502677615138,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0004998682939679794,
      "loss": 1.9333,
      "mean_token_accuracy": 0.56248503079017,
      "step": 490
    },
    {
      "epoch": 0.21420921099607282,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0004998267133906095,
      "loss": 1.8902,
      "mean_token_accuracy": 0.568443168203036,
      "step": 500
    },
    {
      "epoch": 0.21849339521599428,
      "grad_norm": 0.8125,
      "learning_rate": 0.0004997794396450555,
      "loss": 1.9334,
      "mean_token_accuracy": 0.5650926142930984,
      "step": 510
    },
    {
      "epoch": 0.22277757943591575,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0004997264738086136,
      "loss": 1.9271,
      "mean_token_accuracy": 0.5642434308926264,
      "step": 520
    },
    {
      "epoch": 0.2270617636558372,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0004996678170882941,
      "loss": 1.9352,
      "mean_token_accuracy": 0.5623646408319474,
      "step": 530
    },
    {
      "epoch": 0.23134594787575866,
      "grad_norm": 0.8125,
      "learning_rate": 0.000499603470820794,
      "loss": 1.9214,
      "mean_token_accuracy": 0.5678030242522557,
      "step": 540
    },
    {
      "epoch": 0.23563013209568012,
      "grad_norm": 0.65625,
      "learning_rate": 0.0004995334364724658,
      "loss": 1.9247,
      "mean_token_accuracy": 0.5662606621781985,
      "step": 550
    },
    {
      "epoch": 0.23991431631560156,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0004994577156392854,
      "loss": 1.9279,
      "mean_token_accuracy": 0.5633541295925776,
      "step": 560
    },
    {
      "epoch": 0.24419850053552303,
      "grad_norm": 0.640625,
      "learning_rate": 0.0004993763100468144,
      "loss": 1.8886,
      "mean_token_accuracy": 0.5679133623838425,
      "step": 570
    },
    {
      "epoch": 0.2484826847554445,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0004992892215501618,
      "loss": 1.8938,
      "mean_token_accuracy": 0.5673948327700297,
      "step": 580
    },
    {
      "epoch": 0.25276686897536593,
      "grad_norm": 0.671875,
      "learning_rate": 0.0004991964521339408,
      "loss": 1.8795,
      "mean_token_accuracy": 0.5750111728906632,
      "step": 590
    },
    {
      "epoch": 0.25705105319528737,
      "grad_norm": 0.640625,
      "learning_rate": 0.0004990980039122245,
      "loss": 1.8579,
      "mean_token_accuracy": 0.5755017310380935,
      "step": 600
    },
    {
      "epoch": 0.26133523741520887,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0004989938791284971,
      "loss": 1.8579,
      "mean_token_accuracy": 0.5754131337006887,
      "step": 610
    },
    {
      "epoch": 0.2656194216351303,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0004988840801556029,
      "loss": 1.9225,
      "mean_token_accuracy": 0.5679664716124535,
      "step": 620
    },
    {
      "epoch": 0.26990360585505174,
      "grad_norm": 0.625,
      "learning_rate": 0.0004987686094956922,
      "loss": 1.8843,
      "mean_token_accuracy": 0.5681375061472257,
      "step": 630
    },
    {
      "epoch": 0.27418779007497324,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0004986474697801647,
      "loss": 1.8747,
      "mean_token_accuracy": 0.5750736912091573,
      "step": 640
    },
    {
      "epoch": 0.2784719742948947,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000498520663769609,
      "loss": 1.8875,
      "mean_token_accuracy": 0.568752312163512,
      "step": 650
    },
    {
      "epoch": 0.2827561585148161,
      "grad_norm": 0.609375,
      "learning_rate": 0.0004983881943537396,
      "loss": 1.8393,
      "mean_token_accuracy": 0.5768392990032832,
      "step": 660
    },
    {
      "epoch": 0.2870403427347376,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0004982500645513319,
      "loss": 1.8389,
      "mean_token_accuracy": 0.5770889202753703,
      "step": 670
    },
    {
      "epoch": 0.29132452695465905,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0004981062775101524,
      "loss": 1.812,
      "mean_token_accuracy": 0.5807933017611504,
      "step": 680
    },
    {
      "epoch": 0.2956087111745805,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0004979568365068878,
      "loss": 1.8463,
      "mean_token_accuracy": 0.571879476805528,
      "step": 690
    },
    {
      "epoch": 0.299892895394502,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0004978017449470692,
      "loss": 1.8659,
      "mean_token_accuracy": 0.575709896783034,
      "step": 700
    },
    {
      "epoch": 0.3041770796144234,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0004976410063649963,
      "loss": 1.8381,
      "mean_token_accuracy": 0.577228785554568,
      "step": 710
    },
    {
      "epoch": 0.30846126383434486,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0004974746244236546,
      "loss": 1.7498,
      "mean_token_accuracy": 0.5971198469400406,
      "step": 720
    },
    {
      "epoch": 0.31274544805426635,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0004973026029146343,
      "loss": 1.797,
      "mean_token_accuracy": 0.5817218641440074,
      "step": 730
    },
    {
      "epoch": 0.3170296322741878,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0004971249457580418,
      "loss": 1.8144,
      "mean_token_accuracy": 0.5801170518000921,
      "step": 740
    },
    {
      "epoch": 0.32131381649410923,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0004969416570024118,
      "loss": 1.8323,
      "mean_token_accuracy": 0.5770084033409755,
      "step": 750
    },
    {
      "epoch": 0.3255980007140307,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0004967527408246142,
      "loss": 1.8112,
      "mean_token_accuracy": 0.5806623538335164,
      "step": 760
    },
    {
      "epoch": 0.32988218493395216,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0004965582015297593,
      "loss": 1.7814,
      "mean_token_accuracy": 0.590585180123647,
      "step": 770
    },
    {
      "epoch": 0.3341663691538736,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0004963580435510999,
      "loss": 1.7749,
      "mean_token_accuracy": 0.5853669941425323,
      "step": 780
    },
    {
      "epoch": 0.3384505533737951,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0004961522714499296,
      "loss": 1.8028,
      "mean_token_accuracy": 0.5840177754561107,
      "step": 790
    },
    {
      "epoch": 0.34273473759371653,
      "grad_norm": 0.53125,
      "learning_rate": 0.0004959408899154796,
      "loss": 1.8328,
      "mean_token_accuracy": 0.5787926654020945,
      "step": 800
    },
    {
      "epoch": 0.34701892181363797,
      "grad_norm": 0.546875,
      "learning_rate": 0.0004957239037648111,
      "loss": 1.8215,
      "mean_token_accuracy": 0.5796164035797119,
      "step": 810
    },
    {
      "epoch": 0.35130310603355946,
      "grad_norm": 0.546875,
      "learning_rate": 0.0004955013179427064,
      "loss": 1.813,
      "mean_token_accuracy": 0.5805202250679334,
      "step": 820
    },
    {
      "epoch": 0.3555872902534809,
      "grad_norm": 0.515625,
      "learning_rate": 0.0004952731375215554,
      "loss": 1.7699,
      "mean_token_accuracy": 0.5885967900355656,
      "step": 830
    },
    {
      "epoch": 0.35987147447340234,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0004950393677012406,
      "loss": 1.7783,
      "mean_token_accuracy": 0.5840276171763737,
      "step": 840
    },
    {
      "epoch": 0.36415565869332384,
      "grad_norm": 0.53125,
      "learning_rate": 0.0004948000138090178,
      "loss": 1.758,
      "mean_token_accuracy": 0.5915057013432184,
      "step": 850
    },
    {
      "epoch": 0.3684398429132453,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000494555081299396,
      "loss": 1.7502,
      "mean_token_accuracy": 0.5921384165684382,
      "step": 860
    },
    {
      "epoch": 0.3727240271331667,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0004943045757540116,
      "loss": 1.7696,
      "mean_token_accuracy": 0.5876839280128479,
      "step": 870
    },
    {
      "epoch": 0.3770082113530882,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0004940485028815028,
      "loss": 1.7923,
      "mean_token_accuracy": 0.5851048608620961,
      "step": 880
    },
    {
      "epoch": 0.38129239557300965,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0004937868685173779,
      "loss": 1.7734,
      "mean_token_accuracy": 0.5879452233513196,
      "step": 890
    },
    {
      "epoch": 0.3855765797929311,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0004935196786238832,
      "loss": 1.7807,
      "mean_token_accuracy": 0.5856667757034302,
      "step": 900
    },
    {
      "epoch": 0.3898607640128526,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0004932469392898675,
      "loss": 1.7573,
      "mean_token_accuracy": 0.589621431628863,
      "step": 910
    },
    {
      "epoch": 0.394144948232774,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0004929686567306424,
      "loss": 1.7764,
      "mean_token_accuracy": 0.5870031158129374,
      "step": 920
    },
    {
      "epoch": 0.39842913245269546,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0004926848372878412,
      "loss": 1.7499,
      "mean_token_accuracy": 0.5911330193281173,
      "step": 930
    },
    {
      "epoch": 0.40271331667261695,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0004923954874292743,
      "loss": 1.7605,
      "mean_token_accuracy": 0.5885690222183864,
      "step": 940
    },
    {
      "epoch": 0.4069975008925384,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0004921006137487819,
      "loss": 1.7249,
      "mean_token_accuracy": 0.5983182614048322,
      "step": 950
    },
    {
      "epoch": 0.4112816851124598,
      "grad_norm": 0.5,
      "learning_rate": 0.0004918002229660836,
      "loss": 1.7411,
      "mean_token_accuracy": 0.5920816282431285,
      "step": 960
    },
    {
      "epoch": 0.41556586933238127,
      "grad_norm": 0.45703125,
      "learning_rate": 0.000491494321926625,
      "loss": 1.7022,
      "mean_token_accuracy": 0.6039901932080587,
      "step": 970
    },
    {
      "epoch": 0.41985005355230276,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0004911829176014227,
      "loss": 1.7546,
      "mean_token_accuracy": 0.5963041971127192,
      "step": 980
    },
    {
      "epoch": 0.4241342377722242,
      "grad_norm": 0.484375,
      "learning_rate": 0.0004908660170869041,
      "loss": 1.7394,
      "mean_token_accuracy": 0.5946206981937091,
      "step": 990
    },
    {
      "epoch": 0.42841842199214564,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0004905436276047468,
      "loss": 1.7065,
      "mean_token_accuracy": 0.5999080528815587,
      "step": 1000
    },
    {
      "epoch": 0.43270260621206713,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0004902157565017131,
      "loss": 1.7317,
      "mean_token_accuracy": 0.5951716423034668,
      "step": 1010
    },
    {
      "epoch": 0.43698679043198857,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0004898824112494834,
      "loss": 1.7353,
      "mean_token_accuracy": 0.5931165516376495,
      "step": 1020
    },
    {
      "epoch": 0.44127097465191,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0004895435994444855,
      "loss": 1.7234,
      "mean_token_accuracy": 0.592176150282224,
      "step": 1030
    },
    {
      "epoch": 0.4455551588718315,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0004891993288077216,
      "loss": 1.7373,
      "mean_token_accuracy": 0.5937283883492152,
      "step": 1040
    },
    {
      "epoch": 0.44983934309175294,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0004888496071845921,
      "loss": 1.7339,
      "mean_token_accuracy": 0.5963114351034164,
      "step": 1050
    },
    {
      "epoch": 0.4541235273116744,
      "grad_norm": 0.515625,
      "learning_rate": 0.0004884944425447174,
      "loss": 1.7075,
      "mean_token_accuracy": 0.5985465884208679,
      "step": 1060
    },
    {
      "epoch": 0.4584077115315959,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00048813384298175533,
      "loss": 1.7108,
      "mean_token_accuracy": 0.5949873934189479,
      "step": 1070
    },
    {
      "epoch": 0.4626918957515173,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000487767816713218,
      "loss": 1.7319,
      "mean_token_accuracy": 0.5978627324104309,
      "step": 1080
    },
    {
      "epoch": 0.46697607997143875,
      "grad_norm": 0.515625,
      "learning_rate": 0.00048739637208028343,
      "loss": 1.7215,
      "mean_token_accuracy": 0.5949975033601125,
      "step": 1090
    },
    {
      "epoch": 0.47126026419136025,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0004870195175476059,
      "loss": 1.6915,
      "mean_token_accuracy": 0.6006935626268387,
      "step": 1100
    },
    {
      "epoch": 0.4755444484112817,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00048663726170312304,
      "loss": 1.7224,
      "mean_token_accuracy": 0.5965896293520927,
      "step": 1110
    },
    {
      "epoch": 0.4798286326312031,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0004862496132578601,
      "loss": 1.7188,
      "mean_token_accuracy": 0.5954587419827779,
      "step": 1120
    },
    {
      "epoch": 0.4841128168511246,
      "grad_norm": 0.46875,
      "learning_rate": 0.0004858565810457315,
      "loss": 1.7029,
      "mean_token_accuracy": 0.5984790166219075,
      "step": 1130
    },
    {
      "epoch": 0.48839700107104606,
      "grad_norm": 0.46875,
      "learning_rate": 0.00048545817402333944,
      "loss": 1.6707,
      "mean_token_accuracy": 0.6039733906586965,
      "step": 1140
    },
    {
      "epoch": 0.4926811852909675,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00048505440126976975,
      "loss": 1.7063,
      "mean_token_accuracy": 0.5990351974964142,
      "step": 1150
    },
    {
      "epoch": 0.496965369510889,
      "grad_norm": 0.443359375,
      "learning_rate": 0.000484645271986385,
      "loss": 1.7019,
      "mean_token_accuracy": 0.5967398678263028,
      "step": 1160
    },
    {
      "epoch": 0.5012495537308104,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00048423079549661513,
      "loss": 1.7185,
      "mean_token_accuracy": 0.5945661654074986,
      "step": 1170
    },
    {
      "epoch": 0.5055337379507319,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00048381098124574453,
      "loss": 1.6953,
      "mean_token_accuracy": 0.5991637865702312,
      "step": 1180
    },
    {
      "epoch": 0.5098179221706534,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000483385838800697,
      "loss": 1.6724,
      "mean_token_accuracy": 0.6055593649546306,
      "step": 1190
    },
    {
      "epoch": 0.5141021063905747,
      "grad_norm": 0.453125,
      "learning_rate": 0.0004829553778498177,
      "loss": 1.69,
      "mean_token_accuracy": 0.6024374475081762,
      "step": 1200
    },
    {
      "epoch": 0.5183862906104962,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0004825196082026525,
      "loss": 1.6896,
      "mean_token_accuracy": 0.6015072325865428,
      "step": 1210
    },
    {
      "epoch": 0.5226704748304177,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00048207853978972425,
      "loss": 1.6911,
      "mean_token_accuracy": 0.5975394507249197,
      "step": 1220
    },
    {
      "epoch": 0.5269546590503391,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00048163218266230657,
      "loss": 1.6506,
      "mean_token_accuracy": 0.6080829570690791,
      "step": 1230
    },
    {
      "epoch": 0.5312388432702606,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00048118054699219486,
      "loss": 1.7084,
      "mean_token_accuracy": 0.5956538558006287,
      "step": 1240
    },
    {
      "epoch": 0.5355230274901821,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00048072364307147434,
      "loss": 1.6769,
      "mean_token_accuracy": 0.6048657476902009,
      "step": 1250
    },
    {
      "epoch": 0.5398072117101035,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00048026148131228544,
      "loss": 1.6391,
      "mean_token_accuracy": 0.6106082250674566,
      "step": 1260
    },
    {
      "epoch": 0.544091395930025,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00047979407224658704,
      "loss": 1.652,
      "mean_token_accuracy": 0.609614963332812,
      "step": 1270
    },
    {
      "epoch": 0.5483755801499465,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0004793214265259158,
      "loss": 1.6686,
      "mean_token_accuracy": 0.6056814392407736,
      "step": 1280
    },
    {
      "epoch": 0.5526597643698679,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0004788435549211439,
      "loss": 1.7061,
      "mean_token_accuracy": 0.600483645995458,
      "step": 1290
    },
    {
      "epoch": 0.5569439485897894,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00047836046832223336,
      "loss": 1.6521,
      "mean_token_accuracy": 0.607241137822469,
      "step": 1300
    },
    {
      "epoch": 0.5612281328097108,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00047787217773798775,
      "loss": 1.6408,
      "mean_token_accuracy": 0.6109602769215902,
      "step": 1310
    },
    {
      "epoch": 0.5655123170296322,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00047737869429580177,
      "loss": 1.6651,
      "mean_token_accuracy": 0.6036148915688196,
      "step": 1320
    },
    {
      "epoch": 0.5697965012495537,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0004768800292414073,
      "loss": 1.6612,
      "mean_token_accuracy": 0.6085895299911499,
      "step": 1330
    },
    {
      "epoch": 0.5740806854694752,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00047637619393861726,
      "loss": 1.6645,
      "mean_token_accuracy": 0.6032974988222122,
      "step": 1340
    },
    {
      "epoch": 0.5783648696893966,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00047586719986906644,
      "loss": 1.6206,
      "mean_token_accuracy": 0.6132790346940359,
      "step": 1350
    },
    {
      "epoch": 0.5826490539093181,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00047535305863195023,
      "loss": 1.6523,
      "mean_token_accuracy": 0.6070671379566193,
      "step": 1360
    },
    {
      "epoch": 0.5869332381292396,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00047483378194376004,
      "loss": 1.6394,
      "mean_token_accuracy": 0.608396037419637,
      "step": 1370
    },
    {
      "epoch": 0.591217422349161,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00047430938163801623,
      "loss": 1.6279,
      "mean_token_accuracy": 0.6083019236723582,
      "step": 1380
    },
    {
      "epoch": 0.5955016065690825,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00047377986966499867,
      "loss": 1.6447,
      "mean_token_accuracy": 0.6057626972595851,
      "step": 1390
    },
    {
      "epoch": 0.599785790789004,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00047324525809147437,
      "loss": 1.6669,
      "mean_token_accuracy": 0.6031167497237523,
      "step": 1400
    },
    {
      "epoch": 0.6040699750089253,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0004727055591004221,
      "loss": 1.6684,
      "mean_token_accuracy": 0.6013229255874951,
      "step": 1410
    },
    {
      "epoch": 0.6083541592288468,
      "grad_norm": 0.453125,
      "learning_rate": 0.00047216078499075556,
      "loss": 1.6047,
      "mean_token_accuracy": 0.613938628633817,
      "step": 1420
    },
    {
      "epoch": 0.6126383434487683,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0004716109481770422,
      "loss": 1.646,
      "mean_token_accuracy": 0.6082295358181,
      "step": 1430
    },
    {
      "epoch": 0.6169225276686897,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000471056061189221,
      "loss": 1.5866,
      "mean_token_accuracy": 0.6149321556091308,
      "step": 1440
    },
    {
      "epoch": 0.6212067118886112,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0004704961366723165,
      "loss": 1.6708,
      "mean_token_accuracy": 0.607517758011818,
      "step": 1450
    },
    {
      "epoch": 0.6254908961085327,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000469931187386151,
      "loss": 1.6741,
      "mean_token_accuracy": 0.607824856042862,
      "step": 1460
    },
    {
      "epoch": 0.6297750803284541,
      "grad_norm": 0.4375,
      "learning_rate": 0.0004693612262050535,
      "loss": 1.6384,
      "mean_token_accuracy": 0.6065444548924764,
      "step": 1470
    },
    {
      "epoch": 0.6340592645483756,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0004687862661175664,
      "loss": 1.6302,
      "mean_token_accuracy": 0.611606694261233,
      "step": 1480
    },
    {
      "epoch": 0.6383434487682971,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0004682063202261495,
      "loss": 1.608,
      "mean_token_accuracy": 0.617574452360471,
      "step": 1490
    },
    {
      "epoch": 0.6426276329882185,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0004676214017468815,
      "loss": 1.6193,
      "mean_token_accuracy": 0.6125218907992045,
      "step": 1500
    },
    {
      "epoch": 0.64691181720814,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00046703152400915873,
      "loss": 1.6094,
      "mean_token_accuracy": 0.6132543057203292,
      "step": 1510
    },
    {
      "epoch": 0.6511960014280614,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0004664367004553914,
      "loss": 1.644,
      "mean_token_accuracy": 0.6055496126413346,
      "step": 1520
    },
    {
      "epoch": 0.6554801856479828,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0004658369446406974,
      "loss": 1.6027,
      "mean_token_accuracy": 0.6137914508581161,
      "step": 1530
    },
    {
      "epoch": 0.6597643698679043,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000465232270232593,
      "loss": 1.5611,
      "mean_token_accuracy": 0.6206978172063827,
      "step": 1540
    },
    {
      "epoch": 0.6640485540878258,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0004646226910106821,
      "loss": 1.5668,
      "mean_token_accuracy": 0.6236391812562943,
      "step": 1550
    },
    {
      "epoch": 0.6683327383077472,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0004640082208663415,
      "loss": 1.6394,
      "mean_token_accuracy": 0.6060982346534729,
      "step": 1560
    },
    {
      "epoch": 0.6726169225276687,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0004633888738024048,
      "loss": 1.5921,
      "mean_token_accuracy": 0.6156363248825073,
      "step": 1570
    },
    {
      "epoch": 0.6769011067475902,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00046276466393284295,
      "loss": 1.6083,
      "mean_token_accuracy": 0.6118316878875096,
      "step": 1580
    },
    {
      "epoch": 0.6811852909675116,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00046213560548244296,
      "loss": 1.5921,
      "mean_token_accuracy": 0.6157896757125855,
      "step": 1590
    },
    {
      "epoch": 0.6854694751874331,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0004615017127864834,
      "loss": 1.612,
      "mean_token_accuracy": 0.6124424457550048,
      "step": 1600
    },
    {
      "epoch": 0.6897536594073546,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00046086300029040805,
      "loss": 1.5575,
      "mean_token_accuracy": 0.620477185646693,
      "step": 1610
    },
    {
      "epoch": 0.6940378436272759,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0004602194825494965,
      "loss": 1.5634,
      "mean_token_accuracy": 0.623723766207695,
      "step": 1620
    },
    {
      "epoch": 0.6983220278471974,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00045957117422853257,
      "loss": 1.5947,
      "mean_token_accuracy": 0.6120685537656149,
      "step": 1630
    },
    {
      "epoch": 0.7026062120671189,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004589180901014699,
      "loss": 1.5554,
      "mean_token_accuracy": 0.6264029294252396,
      "step": 1640
    },
    {
      "epoch": 0.7068903962870403,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004582602450510955,
      "loss": 1.6142,
      "mean_token_accuracy": 0.6149922668933868,
      "step": 1650
    },
    {
      "epoch": 0.7111745805069618,
      "grad_norm": 0.375,
      "learning_rate": 0.00045759765406869077,
      "loss": 1.5881,
      "mean_token_accuracy": 0.6184295862913132,
      "step": 1660
    },
    {
      "epoch": 0.7154587647268833,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00045693033225368917,
      "loss": 1.5637,
      "mean_token_accuracy": 0.618203051884969,
      "step": 1670
    },
    {
      "epoch": 0.7197429489468047,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0004562582948133331,
      "loss": 1.5764,
      "mean_token_accuracy": 0.6192536850770315,
      "step": 1680
    },
    {
      "epoch": 0.7240271331667262,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004555815570623264,
      "loss": 1.6028,
      "mean_token_accuracy": 0.61400941212972,
      "step": 1690
    },
    {
      "epoch": 0.7283113173866477,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000454900134422486,
      "loss": 1.5918,
      "mean_token_accuracy": 0.6173286845286687,
      "step": 1700
    },
    {
      "epoch": 0.732595501606569,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004542140424223904,
      "loss": 1.5623,
      "mean_token_accuracy": 0.6222762515147527,
      "step": 1710
    },
    {
      "epoch": 0.7368796858264905,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0004535232966970253,
      "loss": 1.5609,
      "mean_token_accuracy": 0.6209416131178538,
      "step": 1720
    },
    {
      "epoch": 0.741163870046412,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0004528279129874281,
      "loss": 1.6066,
      "mean_token_accuracy": 0.6145075420538585,
      "step": 1730
    },
    {
      "epoch": 0.7454480542663334,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00045212790714032843,
      "loss": 1.5767,
      "mean_token_accuracy": 0.6191084752480189,
      "step": 1740
    },
    {
      "epoch": 0.7497322384862549,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0004514232951077875,
      "loss": 1.5814,
      "mean_token_accuracy": 0.612170214454333,
      "step": 1750
    },
    {
      "epoch": 0.7540164227061764,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00045071409294683443,
      "loss": 1.5696,
      "mean_token_accuracy": 0.6165617436170578,
      "step": 1760
    },
    {
      "epoch": 0.7583006069260978,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00045000031681910024,
      "loss": 1.5513,
      "mean_token_accuracy": 0.6239033748706182,
      "step": 1770
    },
    {
      "epoch": 0.7625847911460193,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0004492819829904498,
      "loss": 1.6032,
      "mean_token_accuracy": 0.614315361281236,
      "step": 1780
    },
    {
      "epoch": 0.7668689753659408,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004485591078306109,
      "loss": 1.5622,
      "mean_token_accuracy": 0.6223886062701544,
      "step": 1790
    },
    {
      "epoch": 0.7711531595858622,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0004478317078128013,
      "loss": 1.5891,
      "mean_token_accuracy": 0.6170624762773513,
      "step": 1800
    },
    {
      "epoch": 0.7754373438057837,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004470997995133534,
      "loss": 1.5589,
      "mean_token_accuracy": 0.6224645217259724,
      "step": 1810
    },
    {
      "epoch": 0.7797215280257052,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0004463633996113365,
      "loss": 1.5597,
      "mean_token_accuracy": 0.6208334664503733,
      "step": 1820
    },
    {
      "epoch": 0.7840057122456265,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00044562252488817644,
      "loss": 1.5748,
      "mean_token_accuracy": 0.6175821512937546,
      "step": 1830
    },
    {
      "epoch": 0.788289896465548,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00044487719222727353,
      "loss": 1.5298,
      "mean_token_accuracy": 0.6264814734458923,
      "step": 1840
    },
    {
      "epoch": 0.7925740806854695,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0004441274186136176,
      "loss": 1.5153,
      "mean_token_accuracy": 0.6300516416629155,
      "step": 1850
    },
    {
      "epoch": 0.7968582649053909,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0004433732211334011,
      "loss": 1.556,
      "mean_token_accuracy": 0.6228625237941742,
      "step": 1860
    },
    {
      "epoch": 0.8011424491253124,
      "grad_norm": 0.375,
      "learning_rate": 0.0004426146169736295,
      "loss": 1.5473,
      "mean_token_accuracy": 0.6202453782161077,
      "step": 1870
    },
    {
      "epoch": 0.8054266333452339,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004418516234217297,
      "loss": 1.5485,
      "mean_token_accuracy": 0.6223239193360011,
      "step": 1880
    },
    {
      "epoch": 0.8097108175651553,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00044108425786515626,
      "loss": 1.5764,
      "mean_token_accuracy": 0.6173768778642018,
      "step": 1890
    },
    {
      "epoch": 0.8139950017850768,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00044031253779099505,
      "loss": 1.5518,
      "mean_token_accuracy": 0.6269906918207805,
      "step": 1900
    },
    {
      "epoch": 0.8182791860049982,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00043953648078556465,
      "loss": 1.53,
      "mean_token_accuracy": 0.6300149967273077,
      "step": 1910
    },
    {
      "epoch": 0.8225633702249197,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004387561045340155,
      "loss": 1.5431,
      "mean_token_accuracy": 0.6231356263160706,
      "step": 1920
    },
    {
      "epoch": 0.8268475544448411,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00043797142681992744,
      "loss": 1.5386,
      "mean_token_accuracy": 0.626959690451622,
      "step": 1930
    },
    {
      "epoch": 0.8311317386647625,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004371824655249037,
      "loss": 1.5707,
      "mean_token_accuracy": 0.6185725013415019,
      "step": 1940
    },
    {
      "epoch": 0.835415922884684,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0004363892386281639,
      "loss": 1.5308,
      "mean_token_accuracy": 0.6247758358716965,
      "step": 1950
    },
    {
      "epoch": 0.8397001071046055,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004355917642061342,
      "loss": 1.5746,
      "mean_token_accuracy": 0.6213243504365286,
      "step": 1960
    },
    {
      "epoch": 0.8439842913245269,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004347900604320353,
      "loss": 1.5369,
      "mean_token_accuracy": 0.6267731686433157,
      "step": 1970
    },
    {
      "epoch": 0.8482684755444484,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004339841455754684,
      "loss": 1.5302,
      "mean_token_accuracy": 0.6294100970029831,
      "step": 1980
    },
    {
      "epoch": 0.8525526597643699,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004331740380019988,
      "loss": 1.5668,
      "mean_token_accuracy": 0.6167016873757044,
      "step": 1990
    },
    {
      "epoch": 0.8568368439842913,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004323597561727374,
      "loss": 1.5192,
      "mean_token_accuracy": 0.6278371940056483,
      "step": 2000
    },
    {
      "epoch": 0.8611210282042128,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004315413186439201,
      "loss": 1.5214,
      "mean_token_accuracy": 0.6299047251542409,
      "step": 2010
    },
    {
      "epoch": 0.8654052124241343,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004307187440664846,
      "loss": 1.5366,
      "mean_token_accuracy": 0.6220711261034012,
      "step": 2020
    },
    {
      "epoch": 0.8696893966440556,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00042989205118564575,
      "loss": 1.5766,
      "mean_token_accuracy": 0.6220267007748286,
      "step": 2030
    },
    {
      "epoch": 0.8739735808639771,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00042906125884046827,
      "loss": 1.5452,
      "mean_token_accuracy": 0.628257621328036,
      "step": 2040
    },
    {
      "epoch": 0.8782577650838986,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00042822638596343735,
      "loss": 1.5187,
      "mean_token_accuracy": 0.6320155799388886,
      "step": 2050
    },
    {
      "epoch": 0.88254194930382,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004273874515800271,
      "loss": 1.4877,
      "mean_token_accuracy": 0.6335129290819168,
      "step": 2060
    },
    {
      "epoch": 0.8868261335237415,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004265444748082674,
      "loss": 1.5414,
      "mean_token_accuracy": 0.6232473621765773,
      "step": 2070
    },
    {
      "epoch": 0.891110317743663,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00042569747485830784,
      "loss": 1.5354,
      "mean_token_accuracy": 0.6265609403451283,
      "step": 2080
    },
    {
      "epoch": 0.8953945019635844,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00042484647103198007,
      "loss": 1.5011,
      "mean_token_accuracy": 0.632336409886678,
      "step": 2090
    },
    {
      "epoch": 0.8996786861835059,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004239914827223579,
      "loss": 1.5261,
      "mean_token_accuracy": 0.6247933119535446,
      "step": 2100
    },
    {
      "epoch": 0.9039628704034274,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004231325294133155,
      "loss": 1.5346,
      "mean_token_accuracy": 0.6297541747490565,
      "step": 2110
    },
    {
      "epoch": 0.9082470546233488,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004222696306790833,
      "loss": 1.5457,
      "mean_token_accuracy": 0.6180067032575607,
      "step": 2120
    },
    {
      "epoch": 0.9125312388432703,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000421402806183802,
      "loss": 1.4974,
      "mean_token_accuracy": 0.6331047038237254,
      "step": 2130
    },
    {
      "epoch": 0.9168154230631917,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00042053207568107414,
      "loss": 1.5353,
      "mean_token_accuracy": 0.6293119370937348,
      "step": 2140
    },
    {
      "epoch": 0.9210996072831131,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004196574590135144,
      "loss": 1.5305,
      "mean_token_accuracy": 0.6253014832735062,
      "step": 2150
    },
    {
      "epoch": 0.9253837915030346,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0004187789761122972,
      "loss": 1.5138,
      "mean_token_accuracy": 0.6260040392478307,
      "step": 2160
    },
    {
      "epoch": 0.9296679757229561,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0004178966469967024,
      "loss": 1.5205,
      "mean_token_accuracy": 0.6270245303710301,
      "step": 2170
    },
    {
      "epoch": 0.9339521599428775,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004170104917736591,
      "loss": 1.5428,
      "mean_token_accuracy": 0.6261491070191065,
      "step": 2180
    },
    {
      "epoch": 0.938236344162799,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00041612053063728793,
      "loss": 1.4962,
      "mean_token_accuracy": 0.6338411172231039,
      "step": 2190
    },
    {
      "epoch": 0.9425205283827205,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00041522678386844003,
      "loss": 1.5215,
      "mean_token_accuracy": 0.6297219822804133,
      "step": 2200
    },
    {
      "epoch": 0.9468047126026419,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004143292718342355,
      "loss": 1.5052,
      "mean_token_accuracy": 0.6313861280679702,
      "step": 2210
    },
    {
      "epoch": 0.9510888968225634,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004134280149875991,
      "loss": 1.5522,
      "mean_token_accuracy": 0.6244584927956264,
      "step": 2220
    },
    {
      "epoch": 0.9553730810424849,
      "grad_norm": 0.34375,
      "learning_rate": 0.000412523033866794,
      "loss": 1.5116,
      "mean_token_accuracy": 0.6311194211244583,
      "step": 2230
    },
    {
      "epoch": 0.9596572652624062,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000411614349094954,
      "loss": 1.5068,
      "mean_token_accuracy": 0.6359375943740209,
      "step": 2240
    },
    {
      "epoch": 0.9639414494823277,
      "grad_norm": 0.34375,
      "learning_rate": 0.00041070198137961334,
      "loss": 1.5228,
      "mean_token_accuracy": 0.630420845746994,
      "step": 2250
    },
    {
      "epoch": 0.9682256337022492,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00040978595151223496,
      "loss": 1.5304,
      "mean_token_accuracy": 0.6266944895188014,
      "step": 2260
    },
    {
      "epoch": 0.9725098179221706,
      "grad_norm": 0.34375,
      "learning_rate": 0.00040886628036773665,
      "loss": 1.4539,
      "mean_token_accuracy": 0.6405547618865967,
      "step": 2270
    },
    {
      "epoch": 0.9767940021420921,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0004079429889040153,
      "loss": 1.4749,
      "mean_token_accuracy": 0.6339232285817464,
      "step": 2280
    },
    {
      "epoch": 0.9810781863620136,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004070160981614693,
      "loss": 1.4821,
      "mean_token_accuracy": 0.6370660841464997,
      "step": 2290
    },
    {
      "epoch": 0.985362370581935,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00040608562926251914,
      "loss": 1.5096,
      "mean_token_accuracy": 0.6285079121589661,
      "step": 2300
    },
    {
      "epoch": 0.9896465548018565,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000405151603411126,
      "loss": 1.5167,
      "mean_token_accuracy": 0.6253816932439804,
      "step": 2310
    },
    {
      "epoch": 0.993930739021778,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004042140418923085,
      "loss": 1.487,
      "mean_token_accuracy": 0.6365177949269613,
      "step": 2320
    },
    {
      "epoch": 0.9982149232416994,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004032729660716579,
      "loss": 1.4693,
      "mean_token_accuracy": 0.6372685492038727,
      "step": 2330
    },
    {
      "epoch": 1.0021420921099606,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00040232839739485067,
      "loss": 1.4216,
      "mean_token_accuracy": 0.6438040440732783,
      "step": 2340
    },
    {
      "epoch": 1.0064262763298821,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004013803573871605,
      "loss": 1.3392,
      "mean_token_accuracy": 0.6590171555678049,
      "step": 2350
    },
    {
      "epoch": 1.0107104605498036,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00040042886765296714,
      "loss": 1.3092,
      "mean_token_accuracy": 0.6659611910581589,
      "step": 2360
    },
    {
      "epoch": 1.0149946447697251,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0003994739498752645,
      "loss": 1.338,
      "mean_token_accuracy": 0.6604650288820266,
      "step": 2370
    },
    {
      "epoch": 1.0192788289896466,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0003985156258151662,
      "loss": 1.2883,
      "mean_token_accuracy": 0.6699223518371582,
      "step": 2380
    },
    {
      "epoch": 1.023563013209568,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00039755391731140986,
      "loss": 1.3335,
      "mean_token_accuracy": 0.6621675411860148,
      "step": 2390
    },
    {
      "epoch": 1.0278471974294894,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00039658884627985947,
      "loss": 1.3185,
      "mean_token_accuracy": 0.6633899579445521,
      "step": 2400
    },
    {
      "epoch": 1.0321313816494109,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00039562043471300573,
      "loss": 1.3408,
      "mean_token_accuracy": 0.6567703276872635,
      "step": 2410
    },
    {
      "epoch": 1.0364155658693324,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00039464870467946516,
      "loss": 1.3053,
      "mean_token_accuracy": 0.6645698845386505,
      "step": 2420
    },
    {
      "epoch": 1.0406997500892539,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00039367367832347707,
      "loss": 1.3634,
      "mean_token_accuracy": 0.6515278299649556,
      "step": 2430
    },
    {
      "epoch": 1.0449839343091754,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00039269537786439866,
      "loss": 1.3179,
      "mean_token_accuracy": 0.6596842000881831,
      "step": 2440
    },
    {
      "epoch": 1.0492681185290968,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0003917138255961993,
      "loss": 1.3042,
      "mean_token_accuracy": 0.6642681717872619,
      "step": 2450
    },
    {
      "epoch": 1.0535523027490181,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0003907290438869517,
      "loss": 1.328,
      "mean_token_accuracy": 0.6599580854177475,
      "step": 2460
    },
    {
      "epoch": 1.0578364869689396,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00038974105517832315,
      "loss": 1.304,
      "mean_token_accuracy": 0.6646572331587474,
      "step": 2470
    },
    {
      "epoch": 1.062120671188861,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00038874988198506287,
      "loss": 1.3355,
      "mean_token_accuracy": 0.6583087474107743,
      "step": 2480
    },
    {
      "epoch": 1.0664048554087826,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00038775554689449013,
      "loss": 1.3209,
      "mean_token_accuracy": 0.6637792021036149,
      "step": 2490
    },
    {
      "epoch": 1.070689039628704,
      "grad_norm": 0.390625,
      "learning_rate": 0.00038675807256597863,
      "loss": 1.3437,
      "mean_token_accuracy": 0.6541990071535111,
      "step": 2500
    },
    {
      "epoch": 1.0749732238486256,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0003857574817304407,
      "loss": 1.3368,
      "mean_token_accuracy": 0.6589954853057861,
      "step": 2510
    },
    {
      "epoch": 1.0792574080685469,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0003847537971898093,
      "loss": 1.3431,
      "mean_token_accuracy": 0.6575679163138072,
      "step": 2520
    },
    {
      "epoch": 1.0835415922884684,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0003837470418165176,
      "loss": 1.3559,
      "mean_token_accuracy": 0.6553231467803319,
      "step": 2530
    },
    {
      "epoch": 1.0878257765083899,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000382737238552979,
      "loss": 1.3401,
      "mean_token_accuracy": 0.6568175663550695,
      "step": 2540
    },
    {
      "epoch": 1.0921099607283113,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00038172441041106316,
      "loss": 1.3171,
      "mean_token_accuracy": 0.6621425936619441,
      "step": 2550
    },
    {
      "epoch": 1.0963941449482328,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0003807085804715723,
      "loss": 1.3206,
      "mean_token_accuracy": 0.6625715295473734,
      "step": 2560
    },
    {
      "epoch": 1.1006783291681543,
      "grad_norm": 0.31640625,
      "learning_rate": 0.000379689771883715,
      "loss": 1.3234,
      "mean_token_accuracy": 0.660184133052826,
      "step": 2570
    },
    {
      "epoch": 1.1049625133880756,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00037866800786457864,
      "loss": 1.3085,
      "mean_token_accuracy": 0.6675328105688095,
      "step": 2580
    },
    {
      "epoch": 1.109246697607997,
      "grad_norm": 0.34375,
      "learning_rate": 0.00037764331169860046,
      "loss": 1.3138,
      "mean_token_accuracy": 0.6605868438879648,
      "step": 2590
    },
    {
      "epoch": 1.1135308818279186,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0003766157067370366,
      "loss": 1.3211,
      "mean_token_accuracy": 0.6628186653057734,
      "step": 2600
    },
    {
      "epoch": 1.11781506604784,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00037558521639743036,
      "loss": 1.3113,
      "mean_token_accuracy": 0.6645319203535716,
      "step": 2610
    },
    {
      "epoch": 1.1220992502677616,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0003745518641630785,
      "loss": 1.3303,
      "mean_token_accuracy": 0.659632471203804,
      "step": 2620
    },
    {
      "epoch": 1.1263834344876829,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0003735156735824957,
      "loss": 1.3333,
      "mean_token_accuracy": 0.6590901345014573,
      "step": 2630
    },
    {
      "epoch": 1.1306676187076043,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0003724766682688784,
      "loss": 1.3414,
      "mean_token_accuracy": 0.6546030879020691,
      "step": 2640
    },
    {
      "epoch": 1.1349518029275258,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00037143487189956635,
      "loss": 1.3257,
      "mean_token_accuracy": 0.6572333713372548,
      "step": 2650
    },
    {
      "epoch": 1.1392359871474473,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0003703903082155035,
      "loss": 1.3372,
      "mean_token_accuracy": 0.6556382944186528,
      "step": 2660
    },
    {
      "epoch": 1.1435201713673688,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0003693430010206962,
      "loss": 1.321,
      "mean_token_accuracy": 0.6620635330677033,
      "step": 2670
    },
    {
      "epoch": 1.1478043555872903,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0003682929741816717,
      "loss": 1.3186,
      "mean_token_accuracy": 0.6653014043966929,
      "step": 2680
    },
    {
      "epoch": 1.1520885398072118,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00036724025162693317,
      "loss": 1.3106,
      "mean_token_accuracy": 0.666445321838061,
      "step": 2690
    },
    {
      "epoch": 1.156372724027133,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00036618485734641584,
      "loss": 1.3426,
      "mean_token_accuracy": 0.6579491098721822,
      "step": 2700
    },
    {
      "epoch": 1.1606569082470546,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0003651268153909386,
      "loss": 1.3593,
      "mean_token_accuracy": 0.6558250943819682,
      "step": 2710
    },
    {
      "epoch": 1.164941092466976,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00036406614987165737,
      "loss": 1.3385,
      "mean_token_accuracy": 0.6573931157588959,
      "step": 2720
    },
    {
      "epoch": 1.1692252766868976,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00036300288495951487,
      "loss": 1.303,
      "mean_token_accuracy": 0.6656403839588165,
      "step": 2730
    },
    {
      "epoch": 1.173509460906819,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0003619370448846901,
      "loss": 1.321,
      "mean_token_accuracy": 0.6617445478836695,
      "step": 2740
    },
    {
      "epoch": 1.1777936451267403,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000360868653936046,
      "loss": 1.3403,
      "mean_token_accuracy": 0.6576227575540543,
      "step": 2750
    },
    {
      "epoch": 1.1820778293466618,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00035979773646057603,
      "loss": 1.3161,
      "mean_token_accuracy": 0.6614106744527817,
      "step": 2760
    },
    {
      "epoch": 1.1863620135665833,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0003587243168628491,
      "loss": 1.3549,
      "mean_token_accuracy": 0.6531448741753896,
      "step": 2770
    },
    {
      "epoch": 1.1906461977865048,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00035764841960445433,
      "loss": 1.3016,
      "mean_token_accuracy": 0.6694233824809392,
      "step": 2780
    },
    {
      "epoch": 1.1949303820064263,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0003565700692034421,
      "loss": 1.3076,
      "mean_token_accuracy": 0.6626476556062698,
      "step": 2790
    },
    {
      "epoch": 1.1992145662263478,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00035548929023376677,
      "loss": 1.2832,
      "mean_token_accuracy": 0.6730219264825185,
      "step": 2800
    },
    {
      "epoch": 1.2034987504462693,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00035440610732472564,
      "loss": 1.3105,
      "mean_token_accuracy": 0.6644942373037338,
      "step": 2810
    },
    {
      "epoch": 1.2077829346661906,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00035332054516039834,
      "loss": 1.3084,
      "mean_token_accuracy": 0.6659467816352844,
      "step": 2820
    },
    {
      "epoch": 1.212067118886112,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0003522326284790842,
      "loss": 1.3263,
      "mean_token_accuracy": 0.6606633335351944,
      "step": 2830
    },
    {
      "epoch": 1.2163513031060336,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000351142382072738,
      "loss": 1.3269,
      "mean_token_accuracy": 0.6555161933104198,
      "step": 2840
    },
    {
      "epoch": 1.220635487325955,
      "grad_norm": 0.328125,
      "learning_rate": 0.0003500498307864057,
      "loss": 1.3094,
      "mean_token_accuracy": 0.6611082951227824,
      "step": 2850
    },
    {
      "epoch": 1.2249196715458766,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00034895499951765805,
      "loss": 1.3298,
      "mean_token_accuracy": 0.6599132279555003,
      "step": 2860
    },
    {
      "epoch": 1.2292038557657978,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0003478579132160226,
      "loss": 1.3194,
      "mean_token_accuracy": 0.6632148762543996,
      "step": 2870
    },
    {
      "epoch": 1.2334880399857193,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00034675859688241607,
      "loss": 1.3216,
      "mean_token_accuracy": 0.6630672425031662,
      "step": 2880
    },
    {
      "epoch": 1.2377722242056408,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00034565707556857405,
      "loss": 1.3175,
      "mean_token_accuracy": 0.6651228954394658,
      "step": 2890
    },
    {
      "epoch": 1.2420564084255623,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0003445533743764804,
      "loss": 1.2983,
      "mean_token_accuracy": 0.6665589342514674,
      "step": 2900
    },
    {
      "epoch": 1.2463405926454838,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00034344751845779485,
      "loss": 1.332,
      "mean_token_accuracy": 0.6578145096699397,
      "step": 2910
    },
    {
      "epoch": 1.250624776865405,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00034233953301328026,
      "loss": 1.3221,
      "mean_token_accuracy": 0.6605254203081131,
      "step": 2920
    },
    {
      "epoch": 1.2549089610853268,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0003412294432922278,
      "loss": 1.3093,
      "mean_token_accuracy": 0.665352334578832,
      "step": 2930
    },
    {
      "epoch": 1.259193145305248,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00034011727459188224,
      "loss": 1.2938,
      "mean_token_accuracy": 0.6671251505613327,
      "step": 2940
    },
    {
      "epoch": 1.2634773295251696,
      "grad_norm": 0.33203125,
      "learning_rate": 0.000339003052256865,
      "loss": 1.283,
      "mean_token_accuracy": 0.666038570801417,
      "step": 2950
    },
    {
      "epoch": 1.267761513745091,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0003378868016785966,
      "loss": 1.3174,
      "mean_token_accuracy": 0.6619399875402451,
      "step": 2960
    },
    {
      "epoch": 1.2720456979650125,
      "grad_norm": 0.3203125,
      "learning_rate": 0.000336768548294718,
      "loss": 1.3422,
      "mean_token_accuracy": 0.6565162986516953,
      "step": 2970
    },
    {
      "epoch": 1.276329882184934,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00033564831758851145,
      "loss": 1.2966,
      "mean_token_accuracy": 0.6664889295895894,
      "step": 2980
    },
    {
      "epoch": 1.2806140664048553,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0003345261350883189,
      "loss": 1.3002,
      "mean_token_accuracy": 0.6681903342405955,
      "step": 2990
    },
    {
      "epoch": 1.2848982506247768,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00033340202636696103,
      "loss": 1.3375,
      "mean_token_accuracy": 0.6571354617675146,
      "step": 3000
    },
    {
      "epoch": 1.2891824348446983,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0003322760170411539,
      "loss": 1.3163,
      "mean_token_accuracy": 0.6622751196225484,
      "step": 3010
    },
    {
      "epoch": 1.2934666190646198,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00033114813277092557,
      "loss": 1.3095,
      "mean_token_accuracy": 0.6604131867488225,
      "step": 3020
    },
    {
      "epoch": 1.2977508032845413,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00033001839925903123,
      "loss": 1.269,
      "mean_token_accuracy": 0.670791240533193,
      "step": 3030
    },
    {
      "epoch": 1.3020349875044626,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00032888684225036735,
      "loss": 1.324,
      "mean_token_accuracy": 0.6640015542507172,
      "step": 3040
    },
    {
      "epoch": 1.3063191717243843,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0003277534875313851,
      "loss": 1.2934,
      "mean_token_accuracy": 0.6645639598369598,
      "step": 3050
    },
    {
      "epoch": 1.3106033559443055,
      "grad_norm": 0.314453125,
      "learning_rate": 0.000326618360929503,
      "loss": 1.2911,
      "mean_token_accuracy": 0.662736972173055,
      "step": 3060
    },
    {
      "epoch": 1.314887540164227,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0003254814883125176,
      "loss": 1.3102,
      "mean_token_accuracy": 0.6641101191441218,
      "step": 3070
    },
    {
      "epoch": 1.3191717243841485,
      "grad_norm": 0.34375,
      "learning_rate": 0.00032434289558801486,
      "loss": 1.2946,
      "mean_token_accuracy": 0.669797545671463,
      "step": 3080
    },
    {
      "epoch": 1.32345590860407,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00032320260870277907,
      "loss": 1.3234,
      "mean_token_accuracy": 0.662629238764445,
      "step": 3090
    },
    {
      "epoch": 1.3277400928239915,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00032206065364220204,
      "loss": 1.2875,
      "mean_token_accuracy": 0.6695479412873586,
      "step": 3100
    },
    {
      "epoch": 1.3320242770439128,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0003209170564296907,
      "loss": 1.3185,
      "mean_token_accuracy": 0.6618772675593694,
      "step": 3110
    },
    {
      "epoch": 1.3363084612638343,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00031977184312607406,
      "loss": 1.2956,
      "mean_token_accuracy": 0.6677195648352305,
      "step": 3120
    },
    {
      "epoch": 1.3405926454837558,
      "grad_norm": 0.34375,
      "learning_rate": 0.0003186250398290093,
      "loss": 1.31,
      "mean_token_accuracy": 0.662284501393636,
      "step": 3130
    },
    {
      "epoch": 1.3448768297036773,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0003174766726723873,
      "loss": 1.3053,
      "mean_token_accuracy": 0.6651754269997279,
      "step": 3140
    },
    {
      "epoch": 1.3491610139235988,
      "grad_norm": 0.326171875,
      "learning_rate": 0.000316326767825737,
      "loss": 1.3318,
      "mean_token_accuracy": 0.658694452047348,
      "step": 3150
    },
    {
      "epoch": 1.35344519814352,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0003151753514936285,
      "loss": 1.2949,
      "mean_token_accuracy": 0.6696987201770147,
      "step": 3160
    },
    {
      "epoch": 1.3577293823634418,
      "grad_norm": 0.3125,
      "learning_rate": 0.00031402244991507656,
      "loss": 1.3184,
      "mean_token_accuracy": 0.6619691550731659,
      "step": 3170
    },
    {
      "epoch": 1.362013566583363,
      "grad_norm": 0.328125,
      "learning_rate": 0.00031286808936294266,
      "loss": 1.2695,
      "mean_token_accuracy": 0.6740480273962021,
      "step": 3180
    },
    {
      "epoch": 1.3662977508032845,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00031171229614333567,
      "loss": 1.2973,
      "mean_token_accuracy": 0.6682296107212703,
      "step": 3190
    },
    {
      "epoch": 1.370581935023206,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0003105550965950132,
      "loss": 1.2879,
      "mean_token_accuracy": 0.6695340464512507,
      "step": 3200
    },
    {
      "epoch": 1.3748661192431275,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0003093965170887804,
      "loss": 1.3382,
      "mean_token_accuracy": 0.6577111542224884,
      "step": 3210
    },
    {
      "epoch": 1.379150303463049,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00030823658402689004,
      "loss": 1.3005,
      "mean_token_accuracy": 0.6613259871800741,
      "step": 3220
    },
    {
      "epoch": 1.3834344876829703,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0003070753238424401,
      "loss": 1.2846,
      "mean_token_accuracy": 0.6712899088859559,
      "step": 3230
    },
    {
      "epoch": 1.3877186719028918,
      "grad_norm": 0.3125,
      "learning_rate": 0.0003059127629987715,
      "loss": 1.3022,
      "mean_token_accuracy": 0.6612801601489385,
      "step": 3240
    },
    {
      "epoch": 1.3920028561228133,
      "grad_norm": 0.328125,
      "learning_rate": 0.00030474892798886574,
      "loss": 1.2673,
      "mean_token_accuracy": 0.6732950339714686,
      "step": 3250
    },
    {
      "epoch": 1.3962870403427348,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00030358384533473993,
      "loss": 1.2745,
      "mean_token_accuracy": 0.6716413120428721,
      "step": 3260
    },
    {
      "epoch": 1.4005712245626563,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0003024175415868436,
      "loss": 1.2964,
      "mean_token_accuracy": 0.6657003333171209,
      "step": 3270
    },
    {
      "epoch": 1.4048554087825775,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00030125004332345293,
      "loss": 1.3154,
      "mean_token_accuracy": 0.6634772340456645,
      "step": 3280
    },
    {
      "epoch": 1.409139593002499,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0003000813771500652,
      "loss": 1.2857,
      "mean_token_accuracy": 0.6668003877003987,
      "step": 3290
    },
    {
      "epoch": 1.4134237772224205,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00029891156969879276,
      "loss": 1.2846,
      "mean_token_accuracy": 0.6660936713218689,
      "step": 3300
    },
    {
      "epoch": 1.417707961442342,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00029774064762775584,
      "loss": 1.3119,
      "mean_token_accuracy": 0.6623692701260249,
      "step": 3310
    },
    {
      "epoch": 1.4219921456622635,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00029656863762047507,
      "loss": 1.2528,
      "mean_token_accuracy": 0.6746878981590271,
      "step": 3320
    },
    {
      "epoch": 1.426276329882185,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002953955663852637,
      "loss": 1.2755,
      "mean_token_accuracy": 0.6691719969113668,
      "step": 3330
    },
    {
      "epoch": 1.4305605141021065,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00029422146065461846,
      "loss": 1.291,
      "mean_token_accuracy": 0.6642946968475978,
      "step": 3340
    },
    {
      "epoch": 1.4348446983220278,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002930463471846109,
      "loss": 1.2858,
      "mean_token_accuracy": 0.6680060108502706,
      "step": 3350
    },
    {
      "epoch": 1.4391288825419493,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00029187025275427726,
      "loss": 1.3238,
      "mean_token_accuracy": 0.6617356767257054,
      "step": 3360
    },
    {
      "epoch": 1.4434130667618708,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002906932041650083,
      "loss": 1.2809,
      "mean_token_accuracy": 0.6699890126784642,
      "step": 3370
    },
    {
      "epoch": 1.4476972509817922,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00028951522823993884,
      "loss": 1.2368,
      "mean_token_accuracy": 0.6795000006755193,
      "step": 3380
    },
    {
      "epoch": 1.4519814352017137,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002883363518233361,
      "loss": 1.2939,
      "mean_token_accuracy": 0.6654811183611552,
      "step": 3390
    },
    {
      "epoch": 1.456265619421635,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002871566017799881,
      "loss": 1.2918,
      "mean_token_accuracy": 0.6694387882947922,
      "step": 3400
    },
    {
      "epoch": 1.4605498036415565,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002859760049945915,
      "loss": 1.2996,
      "mean_token_accuracy": 0.6622726023197174,
      "step": 3410
    },
    {
      "epoch": 1.464833987861478,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00028479458837113886,
      "loss": 1.2907,
      "mean_token_accuracy": 0.6693145235379537,
      "step": 3420
    },
    {
      "epoch": 1.4691181720813995,
      "grad_norm": 0.328125,
      "learning_rate": 0.00028361237883230595,
      "loss": 1.2589,
      "mean_token_accuracy": 0.6747862150271734,
      "step": 3430
    },
    {
      "epoch": 1.473402356301321,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00028242940331883726,
      "loss": 1.301,
      "mean_token_accuracy": 0.665789802869161,
      "step": 3440
    },
    {
      "epoch": 1.4776865405212425,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00028124568878893323,
      "loss": 1.2522,
      "mean_token_accuracy": 0.6749315430720647,
      "step": 3450
    },
    {
      "epoch": 1.481970724741164,
      "grad_norm": 0.330078125,
      "learning_rate": 0.000280061262217635,
      "loss": 1.2831,
      "mean_token_accuracy": 0.6701455026865005,
      "step": 3460
    },
    {
      "epoch": 1.4862549089610853,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002788761505962102,
      "loss": 1.2951,
      "mean_token_accuracy": 0.6682408134142558,
      "step": 3470
    },
    {
      "epoch": 1.4905390931810067,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00027769038093153765,
      "loss": 1.2898,
      "mean_token_accuracy": 0.6709642231464386,
      "step": 3480
    },
    {
      "epoch": 1.4948232774009282,
      "grad_norm": 0.318359375,
      "learning_rate": 0.000276503980245492,
      "loss": 1.2709,
      "mean_token_accuracy": 0.6721013089021047,
      "step": 3490
    },
    {
      "epoch": 1.4991074616208497,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002753169755743277,
      "loss": 1.2677,
      "mean_token_accuracy": 0.6691457152366638,
      "step": 3500
    },
    {
      "epoch": 1.5033916458407712,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002741293939680637,
      "loss": 1.2841,
      "mean_token_accuracy": 0.6697168608506521,
      "step": 3510
    },
    {
      "epoch": 1.5076758300606925,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00027294126248986563,
      "loss": 1.2824,
      "mean_token_accuracy": 0.6717678278684616,
      "step": 3520
    },
    {
      "epoch": 1.5119600142806142,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002717526082154304,
      "loss": 1.2587,
      "mean_token_accuracy": 0.6737531036138534,
      "step": 3530
    },
    {
      "epoch": 1.5162441985005355,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00027056345823236837,
      "loss": 1.265,
      "mean_token_accuracy": 0.6732283522685368,
      "step": 3540
    },
    {
      "epoch": 1.520528382720457,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002693738396395866,
      "loss": 1.2674,
      "mean_token_accuracy": 0.671101305882136,
      "step": 3550
    },
    {
      "epoch": 1.5248125669403785,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00026818377954667083,
      "loss": 1.294,
      "mean_token_accuracy": 0.6672694971164067,
      "step": 3560
    },
    {
      "epoch": 1.5290967511602997,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002669933050732679,
      "loss": 1.2372,
      "mean_token_accuracy": 0.6788262248039245,
      "step": 3570
    },
    {
      "epoch": 1.5333809353802215,
      "grad_norm": 0.3203125,
      "learning_rate": 0.000265802443348468,
      "loss": 1.2665,
      "mean_token_accuracy": 0.6717091699441274,
      "step": 3580
    },
    {
      "epoch": 1.5376651196001427,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002646112215101858,
      "loss": 1.2526,
      "mean_token_accuracy": 0.6772685488065083,
      "step": 3590
    },
    {
      "epoch": 1.5419493038200642,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002634196667045428,
      "loss": 1.2361,
      "mean_token_accuracy": 0.6766181766986847,
      "step": 3600
    },
    {
      "epoch": 1.5462334880399857,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002622278060852481,
      "loss": 1.2519,
      "mean_token_accuracy": 0.6780080666144689,
      "step": 3610
    },
    {
      "epoch": 1.5505176722599072,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00026103566681297973,
      "loss": 1.2568,
      "mean_token_accuracy": 0.6755953232447306,
      "step": 3620
    },
    {
      "epoch": 1.5548018564798287,
      "grad_norm": 0.3125,
      "learning_rate": 0.00025984327605476607,
      "loss": 1.2607,
      "mean_token_accuracy": 0.6782088299592336,
      "step": 3630
    },
    {
      "epoch": 1.55908604069975,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002586506609833662,
      "loss": 1.2753,
      "mean_token_accuracy": 0.6692374924818675,
      "step": 3640
    },
    {
      "epoch": 1.5633702249196717,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00025745784877665123,
      "loss": 1.2738,
      "mean_token_accuracy": 0.6710573007663091,
      "step": 3650
    },
    {
      "epoch": 1.567654409139593,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00025626486661698447,
      "loss": 1.2333,
      "mean_token_accuracy": 0.6815130422512691,
      "step": 3660
    },
    {
      "epoch": 1.5719385933595145,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002550717416906022,
      "loss": 1.2424,
      "mean_token_accuracy": 0.6788204977909724,
      "step": 3670
    },
    {
      "epoch": 1.576222777579436,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00025387850118699433,
      "loss": 1.2701,
      "mean_token_accuracy": 0.6694417973359426,
      "step": 3680
    },
    {
      "epoch": 1.5805069617993572,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00025268517229828436,
      "loss": 1.2538,
      "mean_token_accuracy": 0.6743627349535625,
      "step": 3690
    },
    {
      "epoch": 1.584791146019279,
      "grad_norm": 0.328125,
      "learning_rate": 0.00025149178221861015,
      "loss": 1.2538,
      "mean_token_accuracy": 0.6739106665054957,
      "step": 3700
    },
    {
      "epoch": 1.5890753302392002,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002502983581435038,
      "loss": 1.2443,
      "mean_token_accuracy": 0.6750338355700175,
      "step": 3710
    },
    {
      "epoch": 1.5933595144591217,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00024910492726927237,
      "loss": 1.2376,
      "mean_token_accuracy": 0.6762576440970103,
      "step": 3720
    },
    {
      "epoch": 1.5976436986790432,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002479115167923776,
      "loss": 1.2506,
      "mean_token_accuracy": 0.6760666708151499,
      "step": 3730
    },
    {
      "epoch": 1.6019278828989647,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002467181539088166,
      "loss": 1.2515,
      "mean_token_accuracy": 0.6759268889824549,
      "step": 3740
    },
    {
      "epoch": 1.6062120671188862,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002455248658135018,
      "loss": 1.2563,
      "mean_token_accuracy": 0.6760497013727824,
      "step": 3750
    },
    {
      "epoch": 1.6104962513388075,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002443316796996414,
      "loss": 1.2115,
      "mean_token_accuracy": 0.6829163481791815,
      "step": 3760
    },
    {
      "epoch": 1.6147804355587292,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00024313862275811954,
      "loss": 1.2524,
      "mean_token_accuracy": 0.6750467717647552,
      "step": 3770
    },
    {
      "epoch": 1.6190646197786505,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00024194572217687657,
      "loss": 1.2558,
      "mean_token_accuracy": 0.6752542575200399,
      "step": 3780
    },
    {
      "epoch": 1.623348803998572,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00024075300514028996,
      "loss": 1.2014,
      "mean_token_accuracy": 0.6858539601167043,
      "step": 3790
    },
    {
      "epoch": 1.6276329882184934,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00023956049882855435,
      "loss": 1.2438,
      "mean_token_accuracy": 0.6779787172873815,
      "step": 3800
    },
    {
      "epoch": 1.6319171724384147,
      "grad_norm": 0.34375,
      "learning_rate": 0.00023836823041706214,
      "loss": 1.2752,
      "mean_token_accuracy": 0.6702134589354197,
      "step": 3810
    },
    {
      "epoch": 1.6362013566583364,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00023717622707578444,
      "loss": 1.2694,
      "mean_token_accuracy": 0.6743356436491013,
      "step": 3820
    },
    {
      "epoch": 1.6404855408782577,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00023598451596865185,
      "loss": 1.2574,
      "mean_token_accuracy": 0.6762406408786774,
      "step": 3830
    },
    {
      "epoch": 1.6447697250981792,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00023479312425293532,
      "loss": 1.2388,
      "mean_token_accuracy": 0.6752854565779368,
      "step": 3840
    },
    {
      "epoch": 1.6490539093181007,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00023360207907862753,
      "loss": 1.2227,
      "mean_token_accuracy": 0.6849334806203842,
      "step": 3850
    },
    {
      "epoch": 1.653338093538022,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00023241140758782387,
      "loss": 1.2652,
      "mean_token_accuracy": 0.674446169535319,
      "step": 3860
    },
    {
      "epoch": 1.6576222777579437,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00023122113691410396,
      "loss": 1.2314,
      "mean_token_accuracy": 0.6808192272981007,
      "step": 3870
    },
    {
      "epoch": 1.661906461977865,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00023003129418191356,
      "loss": 1.2636,
      "mean_token_accuracy": 0.6753036012252172,
      "step": 3880
    },
    {
      "epoch": 1.6661906461977865,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00022884190650594648,
      "loss": 1.2169,
      "mean_token_accuracy": 0.6835887253284454,
      "step": 3890
    },
    {
      "epoch": 1.670474830417708,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00022765300099052607,
      "loss": 1.2232,
      "mean_token_accuracy": 0.679744150241216,
      "step": 3900
    },
    {
      "epoch": 1.6747590146376294,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00022646460472898824,
      "loss": 1.2182,
      "mean_token_accuracy": 0.6807852476835251,
      "step": 3910
    },
    {
      "epoch": 1.679043198857551,
      "grad_norm": 0.3125,
      "learning_rate": 0.00022527674480306382,
      "loss": 1.2026,
      "mean_token_accuracy": 0.6893678406874338,
      "step": 3920
    },
    {
      "epoch": 1.6833273830774722,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00022408944828226113,
      "loss": 1.234,
      "mean_token_accuracy": 0.6771794279416402,
      "step": 3930
    },
    {
      "epoch": 1.687611567297394,
      "grad_norm": 0.296875,
      "learning_rate": 0.00022290274222324971,
      "loss": 1.235,
      "mean_token_accuracy": 0.6769244899352391,
      "step": 3940
    },
    {
      "epoch": 1.6918957515173152,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00022171665366924303,
      "loss": 1.2525,
      "mean_token_accuracy": 0.6755237986644109,
      "step": 3950
    },
    {
      "epoch": 1.6961799357372367,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002205312096493829,
      "loss": 1.2201,
      "mean_token_accuracy": 0.6836084206899007,
      "step": 3960
    },
    {
      "epoch": 1.7004641199571582,
      "grad_norm": 0.296875,
      "learning_rate": 0.00021934643717812281,
      "loss": 1.2055,
      "mean_token_accuracy": 0.685323445002238,
      "step": 3970
    },
    {
      "epoch": 1.7047483041770795,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002181623632546129,
      "loss": 1.2321,
      "mean_token_accuracy": 0.6818149715662003,
      "step": 3980
    },
    {
      "epoch": 1.7090324883970012,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00021697901486208458,
      "loss": 1.2101,
      "mean_token_accuracy": 0.6855930884679159,
      "step": 3990
    },
    {
      "epoch": 1.7133166726169224,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002157964189672353,
      "loss": 1.2379,
      "mean_token_accuracy": 0.6786120980978012,
      "step": 4000
    },
    {
      "epoch": 1.717600856836844,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002146146025196144,
      "loss": 1.2587,
      "mean_token_accuracy": 0.676125301917394,
      "step": 4010
    },
    {
      "epoch": 1.7218850410567654,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00021343359245100873,
      "loss": 1.2412,
      "mean_token_accuracy": 0.6767153064409892,
      "step": 4020
    },
    {
      "epoch": 1.726169225276687,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002122534156748289,
      "loss": 1.2144,
      "mean_token_accuracy": 0.68453202744325,
      "step": 4030
    },
    {
      "epoch": 1.7304534094966084,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00021107409908549632,
      "loss": 1.2382,
      "mean_token_accuracy": 0.6767129331827164,
      "step": 4040
    },
    {
      "epoch": 1.7347375937165297,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00020989566955782992,
      "loss": 1.2235,
      "mean_token_accuracy": 0.6808336655298869,
      "step": 4050
    },
    {
      "epoch": 1.7390217779364514,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00020871815394643385,
      "loss": 1.2331,
      "mean_token_accuracy": 0.6781981885433197,
      "step": 4060
    },
    {
      "epoch": 1.7433059621563727,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00020754157908508536,
      "loss": 1.2358,
      "mean_token_accuracy": 0.6778120845556259,
      "step": 4070
    },
    {
      "epoch": 1.7475901463762942,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00020636597178612365,
      "loss": 1.2209,
      "mean_token_accuracy": 0.6791574577490489,
      "step": 4080
    },
    {
      "epoch": 1.7518743305962157,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00020519135883983878,
      "loss": 1.2202,
      "mean_token_accuracy": 0.6824594676494599,
      "step": 4090
    },
    {
      "epoch": 1.756158514816137,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002040177670138607,
      "loss": 1.2279,
      "mean_token_accuracy": 0.6825836052497228,
      "step": 4100
    },
    {
      "epoch": 1.7604426990360587,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002028452230525497,
      "loss": 1.1914,
      "mean_token_accuracy": 0.6880497256914775,
      "step": 4110
    },
    {
      "epoch": 1.76472688325598,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00020167375367638707,
      "loss": 1.2336,
      "mean_token_accuracy": 0.6778925875822703,
      "step": 4120
    },
    {
      "epoch": 1.7690110674759014,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002005033855813655,
      "loss": 1.248,
      "mean_token_accuracy": 0.6761407842238744,
      "step": 4130
    },
    {
      "epoch": 1.773295251695823,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0001993341454383817,
      "loss": 1.2239,
      "mean_token_accuracy": 0.6766098787387212,
      "step": 4140
    },
    {
      "epoch": 1.7775794359157444,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0001981660598926277,
      "loss": 1.2034,
      "mean_token_accuracy": 0.6846154699722926,
      "step": 4150
    },
    {
      "epoch": 1.781863620135666,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00019699915556298413,
      "loss": 1.1972,
      "mean_token_accuracy": 0.6866144865751267,
      "step": 4160
    },
    {
      "epoch": 1.7861478043555872,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0001958334590414136,
      "loss": 1.2012,
      "mean_token_accuracy": 0.6865098079045614,
      "step": 4170
    },
    {
      "epoch": 1.790431988575509,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00019466899689235434,
      "loss": 1.2154,
      "mean_token_accuracy": 0.6773028880357742,
      "step": 4180
    },
    {
      "epoch": 1.7947161727954302,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00019350579565211563,
      "loss": 1.2082,
      "mean_token_accuracy": 0.6842084477345148,
      "step": 4190
    },
    {
      "epoch": 1.7990003570153517,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0001923438818282721,
      "loss": 1.2321,
      "mean_token_accuracy": 0.680533907810847,
      "step": 4200
    },
    {
      "epoch": 1.8032845412352732,
      "grad_norm": 0.275390625,
      "learning_rate": 0.00019118328189906037,
      "loss": 1.2276,
      "mean_token_accuracy": 0.6793491671482722,
      "step": 4210
    },
    {
      "epoch": 1.8075687254551944,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00019002402231277533,
      "loss": 1.1895,
      "mean_token_accuracy": 0.6883792887131374,
      "step": 4220
    },
    {
      "epoch": 1.8118529096751161,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00018886612948716737,
      "loss": 1.196,
      "mean_token_accuracy": 0.6824484934409459,
      "step": 4230
    },
    {
      "epoch": 1.8161370938950374,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00018770962980884086,
      "loss": 1.2068,
      "mean_token_accuracy": 0.6837878266970316,
      "step": 4240
    },
    {
      "epoch": 1.820421278114959,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0001865545496326523,
      "loss": 1.2161,
      "mean_token_accuracy": 0.6800723781188329,
      "step": 4250
    },
    {
      "epoch": 1.8247054623348804,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00018540091528110973,
      "loss": 1.1971,
      "mean_token_accuracy": 0.6874994953473409,
      "step": 4260
    },
    {
      "epoch": 1.828989646554802,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0001842487530437732,
      "loss": 1.1987,
      "mean_token_accuracy": 0.6840162913004557,
      "step": 4270
    },
    {
      "epoch": 1.8332738307747234,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00018309808917665562,
      "loss": 1.1892,
      "mean_token_accuracy": 0.6898854543765386,
      "step": 4280
    },
    {
      "epoch": 1.8375580149946447,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00018194894990162424,
      "loss": 1.1957,
      "mean_token_accuracy": 0.6860898315906525,
      "step": 4290
    },
    {
      "epoch": 1.8418421992145664,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00018080136140580328,
      "loss": 1.2323,
      "mean_token_accuracy": 0.6818597843249639,
      "step": 4300
    },
    {
      "epoch": 1.8461263834344876,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00017965534984097696,
      "loss": 1.2124,
      "mean_token_accuracy": 0.6838676472504933,
      "step": 4310
    },
    {
      "epoch": 1.8504105676544091,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00017851094132299362,
      "loss": 1.1997,
      "mean_token_accuracy": 0.6875764499107997,
      "step": 4320
    },
    {
      "epoch": 1.8546947518743306,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00017736816193117066,
      "loss": 1.1956,
      "mean_token_accuracy": 0.6873005121946335,
      "step": 4330
    },
    {
      "epoch": 1.858978936094252,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0001762270377077005,
      "loss": 1.1768,
      "mean_token_accuracy": 0.6917450726032257,
      "step": 4340
    },
    {
      "epoch": 1.8632631203141736,
      "grad_norm": 0.296875,
      "learning_rate": 0.0001750875946570564,
      "loss": 1.2165,
      "mean_token_accuracy": 0.6850099762280782,
      "step": 4350
    },
    {
      "epoch": 1.867547304534095,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00017394985874540032,
      "loss": 1.1952,
      "mean_token_accuracy": 0.6887429515520732,
      "step": 4360
    },
    {
      "epoch": 1.8718314887540164,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00017281385589999133,
      "loss": 1.1916,
      "mean_token_accuracy": 0.6882669021685918,
      "step": 4370
    },
    {
      "epoch": 1.8761156729739379,
      "grad_norm": 0.296875,
      "learning_rate": 0.00017167961200859432,
      "loss": 1.2191,
      "mean_token_accuracy": 0.6836802691221238,
      "step": 4380
    },
    {
      "epoch": 1.8803998571938594,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00017054715291889072,
      "loss": 1.1771,
      "mean_token_accuracy": 0.689866092801094,
      "step": 4390
    },
    {
      "epoch": 1.8846840414137809,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00016941650443788857,
      "loss": 1.1708,
      "mean_token_accuracy": 0.6898455142974853,
      "step": 4400
    },
    {
      "epoch": 1.8889682256337021,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00016828769233133528,
      "loss": 1.2134,
      "mean_token_accuracy": 0.6872386793295543,
      "step": 4410
    },
    {
      "epoch": 1.8932524098536239,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00016716074232312993,
      "loss": 1.2143,
      "mean_token_accuracy": 0.6822999050219853,
      "step": 4420
    },
    {
      "epoch": 1.8975365940735451,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00016603568009473715,
      "loss": 1.1919,
      "mean_token_accuracy": 0.6847109029690425,
      "step": 4430
    },
    {
      "epoch": 1.9018207782934666,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00016491253128460222,
      "loss": 1.1944,
      "mean_token_accuracy": 0.6889555603265762,
      "step": 4440
    },
    {
      "epoch": 1.9061049625133881,
      "grad_norm": 0.3125,
      "learning_rate": 0.00016379132148756638,
      "loss": 1.2015,
      "mean_token_accuracy": 0.6848962704340616,
      "step": 4450
    },
    {
      "epoch": 1.9103891467333094,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00016267207625428375,
      "loss": 1.2066,
      "mean_token_accuracy": 0.6844677517811457,
      "step": 4460
    },
    {
      "epoch": 1.9146733309532311,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00016155482109063898,
      "loss": 1.1895,
      "mean_token_accuracy": 0.689970392982165,
      "step": 4470
    },
    {
      "epoch": 1.9189575151731524,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00016043958145716615,
      "loss": 1.1808,
      "mean_token_accuracy": 0.6905927946170171,
      "step": 4480
    },
    {
      "epoch": 1.9232416993930739,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00015932638276846853,
      "loss": 1.1895,
      "mean_token_accuracy": 0.6876069366931915,
      "step": 4490
    },
    {
      "epoch": 1.9275258836129954,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00015821525039263945,
      "loss": 1.1847,
      "mean_token_accuracy": 0.6875915179649988,
      "step": 4500
    },
    {
      "epoch": 1.9318100678329169,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00015710620965068395,
      "loss": 1.2023,
      "mean_token_accuracy": 0.6880812575419744,
      "step": 4510
    },
    {
      "epoch": 1.9360942520528384,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00015599928581594197,
      "loss": 1.1944,
      "mean_token_accuracy": 0.6877446641524633,
      "step": 4520
    },
    {
      "epoch": 1.9403784362727596,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00015489450411351247,
      "loss": 1.2015,
      "mean_token_accuracy": 0.6858119696378708,
      "step": 4530
    },
    {
      "epoch": 1.9446626204926813,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00015379188971967854,
      "loss": 1.1747,
      "mean_token_accuracy": 0.6891631742318471,
      "step": 4540
    },
    {
      "epoch": 1.9489468047126026,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00015269146776133346,
      "loss": 1.1804,
      "mean_token_accuracy": 0.6899756520986557,
      "step": 4550
    },
    {
      "epoch": 1.9532309889325241,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00015159326331540835,
      "loss": 1.1935,
      "mean_token_accuracy": 0.6860665520032246,
      "step": 4560
    },
    {
      "epoch": 1.9575151731524456,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00015049730140830064,
      "loss": 1.1958,
      "mean_token_accuracy": 0.6895552823940913,
      "step": 4570
    },
    {
      "epoch": 1.9617993573723669,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0001494036070153036,
      "loss": 1.1593,
      "mean_token_accuracy": 0.6954157501459122,
      "step": 4580
    },
    {
      "epoch": 1.9660835415922886,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0001483122050600376,
      "loss": 1.1817,
      "mean_token_accuracy": 0.6880984852711359,
      "step": 4590
    },
    {
      "epoch": 1.9703677258122099,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00014722312041388162,
      "loss": 1.1708,
      "mean_token_accuracy": 0.6934361755847931,
      "step": 4600
    },
    {
      "epoch": 1.9746519100321314,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00014613637789540683,
      "loss": 1.1526,
      "mean_token_accuracy": 0.6956786572933197,
      "step": 4610
    },
    {
      "epoch": 1.9789360942520529,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0001450520022698108,
      "loss": 1.1659,
      "mean_token_accuracy": 0.6891820311546326,
      "step": 4620
    },
    {
      "epoch": 1.9832202784719741,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0001439700182483532,
      "loss": 1.1926,
      "mean_token_accuracy": 0.6883834769328435,
      "step": 4630
    },
    {
      "epoch": 1.9875044626918958,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00014289045048779316,
      "loss": 1.179,
      "mean_token_accuracy": 0.6905747185150782,
      "step": 4640
    },
    {
      "epoch": 1.9917886469118171,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00014181332358982615,
      "loss": 1.1518,
      "mean_token_accuracy": 0.6962420682112376,
      "step": 4650
    },
    {
      "epoch": 1.9960728311317386,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00014073866210052478,
      "loss": 1.1798,
      "mean_token_accuracy": 0.6898052622874578,
      "step": 4660
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013966649050977853,
      "loss": 1.1623,
      "mean_token_accuracy": 0.6901761282574047,
      "step": 4670
    },
    {
      "epoch": 2.0042841842199213,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00013859683325073563,
      "loss": 0.848,
      "mean_token_accuracy": 0.7650491803884506,
      "step": 4680
    },
    {
      "epoch": 2.008568368439843,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00013752971469924727,
      "loss": 0.863,
      "mean_token_accuracy": 0.7565648088852565,
      "step": 4690
    },
    {
      "epoch": 2.0128525526597643,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00013646515917331055,
      "loss": 0.8361,
      "mean_token_accuracy": 0.7650597403446834,
      "step": 4700
    },
    {
      "epoch": 2.017136736879686,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00013540319093251565,
      "loss": 0.8601,
      "mean_token_accuracy": 0.7594423244396845,
      "step": 4710
    },
    {
      "epoch": 2.0214209210996072,
      "grad_norm": 0.28125,
      "learning_rate": 0.00013434383417749247,
      "loss": 0.8135,
      "mean_token_accuracy": 0.7731389890114466,
      "step": 4720
    },
    {
      "epoch": 2.0257051053195285,
      "grad_norm": 0.296875,
      "learning_rate": 0.0001332871130493587,
      "loss": 0.8556,
      "mean_token_accuracy": 0.7604803055524826,
      "step": 4730
    },
    {
      "epoch": 2.0299892895394502,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0001322330516291709,
      "loss": 0.842,
      "mean_token_accuracy": 0.7640973548094432,
      "step": 4740
    },
    {
      "epoch": 2.0342734737593715,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0001311816739373742,
      "loss": 0.8381,
      "mean_token_accuracy": 0.7674211412668228,
      "step": 4750
    },
    {
      "epoch": 2.0385576579792932,
      "grad_norm": 0.296875,
      "learning_rate": 0.00013013300393325611,
      "loss": 0.8126,
      "mean_token_accuracy": 0.7714175979296366,
      "step": 4760
    },
    {
      "epoch": 2.0428418421992145,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00012908706551440004,
      "loss": 0.8481,
      "mean_token_accuracy": 0.7640273501475652,
      "step": 4770
    },
    {
      "epoch": 2.047126026419136,
      "grad_norm": 0.3125,
      "learning_rate": 0.00012804388251614037,
      "loss": 0.8311,
      "mean_token_accuracy": 0.7653455446163814,
      "step": 4780
    },
    {
      "epoch": 2.0514102106390575,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00012700347871102036,
      "loss": 0.8258,
      "mean_token_accuracy": 0.7644449760516484,
      "step": 4790
    },
    {
      "epoch": 2.0556943948589788,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00012596587780824923,
      "loss": 0.8364,
      "mean_token_accuracy": 0.7657805611689885,
      "step": 4800
    },
    {
      "epoch": 2.0599785790789005,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0001249311034531623,
      "loss": 0.8471,
      "mean_token_accuracy": 0.7625262240568796,
      "step": 4810
    },
    {
      "epoch": 2.0642627632988217,
      "grad_norm": 0.3125,
      "learning_rate": 0.00012389917922668245,
      "loss": 0.823,
      "mean_token_accuracy": 0.769603114326795,
      "step": 4820
    },
    {
      "epoch": 2.0685469475187435,
      "grad_norm": 0.328125,
      "learning_rate": 0.0001228701286447824,
      "loss": 0.8242,
      "mean_token_accuracy": 0.7667522599299749,
      "step": 4830
    },
    {
      "epoch": 2.0728311317386647,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00012184397515794888,
      "loss": 0.843,
      "mean_token_accuracy": 0.7642757395903269,
      "step": 4840
    },
    {
      "epoch": 2.077115315958586,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00012082074215064836,
      "loss": 0.8298,
      "mean_token_accuracy": 0.7669574290513992,
      "step": 4850
    },
    {
      "epoch": 2.0813995001785077,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00011980045294079384,
      "loss": 0.8241,
      "mean_token_accuracy": 0.7687270969152451,
      "step": 4860
    },
    {
      "epoch": 2.085683684398429,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00011878313077921388,
      "loss": 0.8395,
      "mean_token_accuracy": 0.76190112332503,
      "step": 4870
    },
    {
      "epoch": 2.0899678686183507,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00011776879884912247,
      "loss": 0.8234,
      "mean_token_accuracy": 0.7668329626321793,
      "step": 4880
    },
    {
      "epoch": 2.094252052838272,
      "grad_norm": 0.3125,
      "learning_rate": 0.00011675748026559091,
      "loss": 0.8221,
      "mean_token_accuracy": 0.7703782876332601,
      "step": 4890
    },
    {
      "epoch": 2.0985362370581937,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00011574919807502091,
      "loss": 0.8332,
      "mean_token_accuracy": 0.7663482298453649,
      "step": 4900
    },
    {
      "epoch": 2.102820421278115,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00011474397525461919,
      "loss": 0.8166,
      "mean_token_accuracy": 0.7704643438259761,
      "step": 4910
    },
    {
      "epoch": 2.1071046054980362,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0001137418347118744,
      "loss": 0.8254,
      "mean_token_accuracy": 0.7690837909777959,
      "step": 4920
    },
    {
      "epoch": 2.111388789717958,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00011274279928403475,
      "loss": 0.8058,
      "mean_token_accuracy": 0.7719497382640839,
      "step": 4930
    },
    {
      "epoch": 2.1156729739378792,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00011174689173758759,
      "loss": 0.8294,
      "mean_token_accuracy": 0.7664976229270299,
      "step": 4940
    },
    {
      "epoch": 2.119957158157801,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00011075413476774066,
      "loss": 0.8292,
      "mean_token_accuracy": 0.7666595439116161,
      "step": 4950
    },
    {
      "epoch": 2.124241342377722,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00010976455099790491,
      "loss": 0.8326,
      "mean_token_accuracy": 0.7672491510709126,
      "step": 4960
    },
    {
      "epoch": 2.1285255265976435,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00010877816297917881,
      "loss": 0.8007,
      "mean_token_accuracy": 0.7737640182177226,
      "step": 4970
    },
    {
      "epoch": 2.132809710817565,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00010779499318983463,
      "loss": 0.8092,
      "mean_token_accuracy": 0.7724389503399531,
      "step": 4980
    },
    {
      "epoch": 2.1370938950374865,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00010681506403480617,
      "loss": 0.8111,
      "mean_token_accuracy": 0.7739120999972026,
      "step": 4990
    },
    {
      "epoch": 2.141378079257408,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00010583839784517812,
      "loss": 0.82,
      "mean_token_accuracy": 0.7729650676250458,
      "step": 5000
    },
    {
      "epoch": 2.1456622634773295,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00010486501687767719,
      "loss": 0.792,
      "mean_token_accuracy": 0.776722161968549,
      "step": 5010
    },
    {
      "epoch": 2.149946447697251,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00010389494331416477,
      "loss": 0.8301,
      "mean_token_accuracy": 0.7654323528210322,
      "step": 5020
    },
    {
      "epoch": 2.1542306319171725,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0001029281992611317,
      "loss": 0.7938,
      "mean_token_accuracy": 0.7787847359975179,
      "step": 5030
    },
    {
      "epoch": 2.1585148161370937,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00010196480674919443,
      "loss": 0.8205,
      "mean_token_accuracy": 0.7693845987319946,
      "step": 5040
    },
    {
      "epoch": 2.1627990003570154,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0001010047877325928,
      "loss": 0.7904,
      "mean_token_accuracy": 0.7785027374823889,
      "step": 5050
    },
    {
      "epoch": 2.1670831845769367,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00010004816408869002,
      "loss": 0.8308,
      "mean_token_accuracy": 0.7703908115625382,
      "step": 5060
    },
    {
      "epoch": 2.1713673687968584,
      "grad_norm": 0.3203125,
      "learning_rate": 9.909495761747372e-05,
      "loss": 0.8253,
      "mean_token_accuracy": 0.7665383875370025,
      "step": 5070
    },
    {
      "epoch": 2.1756515530167797,
      "grad_norm": 0.31640625,
      "learning_rate": 9.814519004105946e-05,
      "loss": 0.8144,
      "mean_token_accuracy": 0.7718339473009109,
      "step": 5080
    },
    {
      "epoch": 2.179935737236701,
      "grad_norm": 0.30078125,
      "learning_rate": 9.719888300319602e-05,
      "loss": 0.8283,
      "mean_token_accuracy": 0.7683377832174301,
      "step": 5090
    },
    {
      "epoch": 2.1842199214566227,
      "grad_norm": 0.302734375,
      "learning_rate": 9.625605806877122e-05,
      "loss": 0.8472,
      "mean_token_accuracy": 0.7613804837067922,
      "step": 5100
    },
    {
      "epoch": 2.188504105676544,
      "grad_norm": 0.306640625,
      "learning_rate": 9.531673672332145e-05,
      "loss": 0.8098,
      "mean_token_accuracy": 0.7719518701235454,
      "step": 5110
    },
    {
      "epoch": 2.1927882898964657,
      "grad_norm": 0.314453125,
      "learning_rate": 9.438094037254172e-05,
      "loss": 0.7964,
      "mean_token_accuracy": 0.7749660849571228,
      "step": 5120
    },
    {
      "epoch": 2.197072474116387,
      "grad_norm": 0.314453125,
      "learning_rate": 9.344869034179735e-05,
      "loss": 0.8143,
      "mean_token_accuracy": 0.7723793685436249,
      "step": 5130
    },
    {
      "epoch": 2.2013566583363087,
      "grad_norm": 0.32421875,
      "learning_rate": 9.252000787563919e-05,
      "loss": 0.8087,
      "mean_token_accuracy": 0.7712360550959905,
      "step": 5140
    },
    {
      "epoch": 2.20564084255623,
      "grad_norm": 0.33203125,
      "learning_rate": 9.159491413731805e-05,
      "loss": 0.7954,
      "mean_token_accuracy": 0.7725468198458354,
      "step": 5150
    },
    {
      "epoch": 2.209925026776151,
      "grad_norm": 0.3125,
      "learning_rate": 9.067343020830352e-05,
      "loss": 0.8136,
      "mean_token_accuracy": 0.7698225736618042,
      "step": 5160
    },
    {
      "epoch": 2.214209210996073,
      "grad_norm": 0.310546875,
      "learning_rate": 8.975557708780316e-05,
      "loss": 0.8181,
      "mean_token_accuracy": 0.769992024699847,
      "step": 5170
    },
    {
      "epoch": 2.218493395215994,
      "grad_norm": 0.310546875,
      "learning_rate": 8.884137569228362e-05,
      "loss": 0.8098,
      "mean_token_accuracy": 0.770973147948583,
      "step": 5180
    },
    {
      "epoch": 2.222777579435916,
      "grad_norm": 0.3359375,
      "learning_rate": 8.793084685499498e-05,
      "loss": 0.7998,
      "mean_token_accuracy": 0.776837948958079,
      "step": 5190
    },
    {
      "epoch": 2.227061763655837,
      "grad_norm": 0.306640625,
      "learning_rate": 8.702401132549459e-05,
      "loss": 0.7891,
      "mean_token_accuracy": 0.7805549720923106,
      "step": 5200
    },
    {
      "epoch": 2.2313459478757585,
      "grad_norm": 0.322265625,
      "learning_rate": 8.612088976917554e-05,
      "loss": 0.8364,
      "mean_token_accuracy": 0.7643806467453639,
      "step": 5210
    },
    {
      "epoch": 2.23563013209568,
      "grad_norm": 0.283203125,
      "learning_rate": 8.522150276679494e-05,
      "loss": 0.8184,
      "mean_token_accuracy": 0.7683983782927195,
      "step": 5220
    },
    {
      "epoch": 2.2399143163156015,
      "grad_norm": 0.306640625,
      "learning_rate": 8.432587081400492e-05,
      "loss": 0.7959,
      "mean_token_accuracy": 0.778293655316035,
      "step": 5230
    },
    {
      "epoch": 2.244198500535523,
      "grad_norm": 0.3203125,
      "learning_rate": 8.343401432088629e-05,
      "loss": 0.7847,
      "mean_token_accuracy": 0.7779005120197932,
      "step": 5240
    },
    {
      "epoch": 2.2484826847554444,
      "grad_norm": 0.296875,
      "learning_rate": 8.254595361148262e-05,
      "loss": 0.7926,
      "mean_token_accuracy": 0.7749503721793493,
      "step": 5250
    },
    {
      "epoch": 2.2527668689753657,
      "grad_norm": 0.30859375,
      "learning_rate": 8.166170892333722e-05,
      "loss": 0.7907,
      "mean_token_accuracy": 0.7779211064179739,
      "step": 5260
    },
    {
      "epoch": 2.2570510531952874,
      "grad_norm": 0.318359375,
      "learning_rate": 8.078130040703238e-05,
      "loss": 0.7985,
      "mean_token_accuracy": 0.7758079042037328,
      "step": 5270
    },
    {
      "epoch": 2.2613352374152087,
      "grad_norm": 0.298828125,
      "learning_rate": 7.990474812572981e-05,
      "loss": 0.7876,
      "mean_token_accuracy": 0.7797471781571707,
      "step": 5280
    },
    {
      "epoch": 2.2656194216351304,
      "grad_norm": 0.306640625,
      "learning_rate": 7.903207205471358e-05,
      "loss": 0.8053,
      "mean_token_accuracy": 0.7768173178037008,
      "step": 5290
    },
    {
      "epoch": 2.2699036058550517,
      "grad_norm": 0.298828125,
      "learning_rate": 7.81632920809349e-05,
      "loss": 0.7792,
      "mean_token_accuracy": 0.7814870576063792,
      "step": 5300
    },
    {
      "epoch": 2.2741877900749734,
      "grad_norm": 0.322265625,
      "learning_rate": 7.729842800255865e-05,
      "loss": 0.8226,
      "mean_token_accuracy": 0.768639792005221,
      "step": 5310
    },
    {
      "epoch": 2.2784719742948947,
      "grad_norm": 0.298828125,
      "learning_rate": 7.64374995285127e-05,
      "loss": 0.7939,
      "mean_token_accuracy": 0.7751558671394984,
      "step": 5320
    },
    {
      "epoch": 2.282756158514816,
      "grad_norm": 0.30078125,
      "learning_rate": 7.558052627803846e-05,
      "loss": 0.7966,
      "mean_token_accuracy": 0.7758613834778468,
      "step": 5330
    },
    {
      "epoch": 2.2870403427347377,
      "grad_norm": 0.328125,
      "learning_rate": 7.472752778024383e-05,
      "loss": 0.7812,
      "mean_token_accuracy": 0.7802222698926926,
      "step": 5340
    },
    {
      "epoch": 2.291324526954659,
      "grad_norm": 0.287109375,
      "learning_rate": 7.387852347365829e-05,
      "loss": 0.7849,
      "mean_token_accuracy": 0.7787080804506937,
      "step": 5350
    },
    {
      "epoch": 2.2956087111745807,
      "grad_norm": 0.326171875,
      "learning_rate": 7.303353270578952e-05,
      "loss": 0.8034,
      "mean_token_accuracy": 0.770248160759608,
      "step": 5360
    },
    {
      "epoch": 2.299892895394502,
      "grad_norm": 0.318359375,
      "learning_rate": 7.219257473268312e-05,
      "loss": 0.8094,
      "mean_token_accuracy": 0.7722504367431005,
      "step": 5370
    },
    {
      "epoch": 2.3041770796144236,
      "grad_norm": 0.318359375,
      "learning_rate": 7.135566871848331e-05,
      "loss": 0.8008,
      "mean_token_accuracy": 0.7736594120661417,
      "step": 5380
    },
    {
      "epoch": 2.308461263834345,
      "grad_norm": 0.302734375,
      "learning_rate": 7.052283373499649e-05,
      "loss": 0.797,
      "mean_token_accuracy": 0.7745833595593771,
      "step": 5390
    },
    {
      "epoch": 2.312745448054266,
      "grad_norm": 0.306640625,
      "learning_rate": 6.969408876125637e-05,
      "loss": 0.7898,
      "mean_token_accuracy": 0.77644149462382,
      "step": 5400
    },
    {
      "epoch": 2.317029632274188,
      "grad_norm": 0.32421875,
      "learning_rate": 6.886945268309177e-05,
      "loss": 0.8053,
      "mean_token_accuracy": 0.7735361973444621,
      "step": 5410
    },
    {
      "epoch": 2.321313816494109,
      "grad_norm": 0.302734375,
      "learning_rate": 6.804894429269582e-05,
      "loss": 0.7936,
      "mean_token_accuracy": 0.7798129876454671,
      "step": 5420
    },
    {
      "epoch": 2.325598000714031,
      "grad_norm": 0.3203125,
      "learning_rate": 6.723258228819815e-05,
      "loss": 0.796,
      "mean_token_accuracy": 0.7762234061956406,
      "step": 5430
    },
    {
      "epoch": 2.329882184933952,
      "grad_norm": 0.294921875,
      "learning_rate": 6.64203852732386e-05,
      "loss": 0.8006,
      "mean_token_accuracy": 0.7752728084723155,
      "step": 5440
    },
    {
      "epoch": 2.3341663691538734,
      "grad_norm": 0.30859375,
      "learning_rate": 6.561237175654325e-05,
      "loss": 0.7862,
      "mean_token_accuracy": 0.7770834614833196,
      "step": 5450
    },
    {
      "epoch": 2.338450553373795,
      "grad_norm": 0.337890625,
      "learning_rate": 6.480856015150272e-05,
      "loss": 0.8179,
      "mean_token_accuracy": 0.7689370075861613,
      "step": 5460
    },
    {
      "epoch": 2.3427347375937164,
      "grad_norm": 0.291015625,
      "learning_rate": 6.40089687757523e-05,
      "loss": 0.7858,
      "mean_token_accuracy": 0.7789360960324605,
      "step": 5470
    },
    {
      "epoch": 2.347018921813638,
      "grad_norm": 0.32421875,
      "learning_rate": 6.32136158507549e-05,
      "loss": 0.8105,
      "mean_token_accuracy": 0.7707864145437876,
      "step": 5480
    },
    {
      "epoch": 2.3513031060335594,
      "grad_norm": 0.306640625,
      "learning_rate": 6.242251950138564e-05,
      "loss": 0.7907,
      "mean_token_accuracy": 0.7774596979220708,
      "step": 5490
    },
    {
      "epoch": 2.3555872902534807,
      "grad_norm": 0.3203125,
      "learning_rate": 6.163569775551869e-05,
      "loss": 0.7875,
      "mean_token_accuracy": 0.780629759033521,
      "step": 5500
    },
    {
      "epoch": 2.3598714744734024,
      "grad_norm": 0.298828125,
      "learning_rate": 6.0853168543616694e-05,
      "loss": 0.8057,
      "mean_token_accuracy": 0.7729009737571081,
      "step": 5510
    },
    {
      "epoch": 2.3641556586933237,
      "grad_norm": 0.322265625,
      "learning_rate": 6.007494969832181e-05,
      "loss": 0.7709,
      "mean_token_accuracy": 0.7820465574661891,
      "step": 5520
    },
    {
      "epoch": 2.3684398429132454,
      "grad_norm": 0.30859375,
      "learning_rate": 5.9301058954049664e-05,
      "loss": 0.7857,
      "mean_token_accuracy": 0.7776114324728648,
      "step": 5530
    },
    {
      "epoch": 2.3727240271331667,
      "grad_norm": 0.3125,
      "learning_rate": 5.853151394658526e-05,
      "loss": 0.7841,
      "mean_token_accuracy": 0.777163389325142,
      "step": 5540
    },
    {
      "epoch": 2.3770082113530884,
      "grad_norm": 0.326171875,
      "learning_rate": 5.776633221268057e-05,
      "loss": 0.7933,
      "mean_token_accuracy": 0.7760427872339885,
      "step": 5550
    },
    {
      "epoch": 2.3812923955730096,
      "grad_norm": 0.2890625,
      "learning_rate": 5.7005531189655515e-05,
      "loss": 0.8003,
      "mean_token_accuracy": 0.7774786601463953,
      "step": 5560
    },
    {
      "epoch": 2.385576579792931,
      "grad_norm": 0.31640625,
      "learning_rate": 5.624912821500025e-05,
      "loss": 0.789,
      "mean_token_accuracy": 0.77956602871418,
      "step": 5570
    },
    {
      "epoch": 2.3898607640128526,
      "grad_norm": 0.322265625,
      "learning_rate": 5.5497140525979925e-05,
      "loss": 0.8054,
      "mean_token_accuracy": 0.7731809784968694,
      "step": 5580
    },
    {
      "epoch": 2.394144948232774,
      "grad_norm": 0.3359375,
      "learning_rate": 5.474958525924262e-05,
      "loss": 0.7969,
      "mean_token_accuracy": 0.7769083728392919,
      "step": 5590
    },
    {
      "epoch": 2.3984291324526956,
      "grad_norm": 0.326171875,
      "learning_rate": 5.4006479450427694e-05,
      "loss": 0.7873,
      "mean_token_accuracy": 0.7816483676433563,
      "step": 5600
    },
    {
      "epoch": 2.402713316672617,
      "grad_norm": 0.3359375,
      "learning_rate": 5.3267840033778516e-05,
      "loss": 0.7924,
      "mean_token_accuracy": 0.7759514023860296,
      "step": 5610
    },
    {
      "epoch": 2.4069975008925386,
      "grad_norm": 0.29296875,
      "learning_rate": 5.25336838417563e-05,
      "loss": 0.7693,
      "mean_token_accuracy": 0.7819558610518773,
      "step": 5620
    },
    {
      "epoch": 2.41128168511246,
      "grad_norm": 0.294921875,
      "learning_rate": 5.1804027604655995e-05,
      "loss": 0.7901,
      "mean_token_accuracy": 0.7765521576007207,
      "step": 5630
    },
    {
      "epoch": 2.415565869332381,
      "grad_norm": 0.30078125,
      "learning_rate": 5.1078887950226084e-05,
      "loss": 0.8054,
      "mean_token_accuracy": 0.7711812168359756,
      "step": 5640
    },
    {
      "epoch": 2.419850053552303,
      "grad_norm": 0.298828125,
      "learning_rate": 5.035828140328852e-05,
      "loss": 0.7997,
      "mean_token_accuracy": 0.7752634723981221,
      "step": 5650
    },
    {
      "epoch": 2.424134237772224,
      "grad_norm": 0.310546875,
      "learning_rate": 4.964222438536295e-05,
      "loss": 0.7682,
      "mean_token_accuracy": 0.7835259586572647,
      "step": 5660
    },
    {
      "epoch": 2.428418421992146,
      "grad_norm": 0.29296875,
      "learning_rate": 4.8930733214292227e-05,
      "loss": 0.7798,
      "mean_token_accuracy": 0.7836492071549098,
      "step": 5670
    },
    {
      "epoch": 2.432702606212067,
      "grad_norm": 0.33203125,
      "learning_rate": 4.822382410387027e-05,
      "loss": 0.7744,
      "mean_token_accuracy": 0.7835030923287074,
      "step": 5680
    },
    {
      "epoch": 2.4369867904319884,
      "grad_norm": 0.314453125,
      "learning_rate": 4.752151316347328e-05,
      "loss": 0.7777,
      "mean_token_accuracy": 0.7836812446514766,
      "step": 5690
    },
    {
      "epoch": 2.44127097465191,
      "grad_norm": 0.328125,
      "learning_rate": 4.682381639769195e-05,
      "loss": 0.7843,
      "mean_token_accuracy": 0.7796516954898834,
      "step": 5700
    },
    {
      "epoch": 2.4455551588718314,
      "grad_norm": 0.337890625,
      "learning_rate": 4.6130749705966924e-05,
      "loss": 0.7856,
      "mean_token_accuracy": 0.7769708921511967,
      "step": 5710
    },
    {
      "epoch": 2.449839343091753,
      "grad_norm": 0.3203125,
      "learning_rate": 4.544232888222674e-05,
      "loss": 0.7558,
      "mean_token_accuracy": 0.7886910130580266,
      "step": 5720
    },
    {
      "epoch": 2.4541235273116744,
      "grad_norm": 0.326171875,
      "learning_rate": 4.475856961452765e-05,
      "loss": 0.7777,
      "mean_token_accuracy": 0.7824478884538014,
      "step": 5730
    },
    {
      "epoch": 2.4584077115315957,
      "grad_norm": 0.3203125,
      "learning_rate": 4.407948748469615e-05,
      "loss": 0.7647,
      "mean_token_accuracy": 0.7870442191759746,
      "step": 5740
    },
    {
      "epoch": 2.4626918957515174,
      "grad_norm": 0.34375,
      "learning_rate": 4.340509796797401e-05,
      "loss": 0.8013,
      "mean_token_accuracy": 0.7732500980297724,
      "step": 5750
    },
    {
      "epoch": 2.4669760799714386,
      "grad_norm": 0.2890625,
      "learning_rate": 4.273541643266537e-05,
      "loss": 0.7809,
      "mean_token_accuracy": 0.7783005088567734,
      "step": 5760
    },
    {
      "epoch": 2.4712602641913604,
      "grad_norm": 0.30078125,
      "learning_rate": 4.2070458139786886e-05,
      "loss": 0.7664,
      "mean_token_accuracy": 0.7849383483330409,
      "step": 5770
    },
    {
      "epoch": 2.4755444484112816,
      "grad_norm": 0.3125,
      "learning_rate": 4.141023824271964e-05,
      "loss": 0.7862,
      "mean_token_accuracy": 0.7778257886568706,
      "step": 5780
    },
    {
      "epoch": 2.479828632631203,
      "grad_norm": 0.3125,
      "learning_rate": 4.075477178686382e-05,
      "loss": 0.7558,
      "mean_token_accuracy": 0.7865474035342535,
      "step": 5790
    },
    {
      "epoch": 2.4841128168511246,
      "grad_norm": 0.314453125,
      "learning_rate": 4.010407370929634e-05,
      "loss": 0.7903,
      "mean_token_accuracy": 0.776697979370753,
      "step": 5800
    },
    {
      "epoch": 2.488397001071046,
      "grad_norm": 0.3046875,
      "learning_rate": 3.94581588384296e-05,
      "loss": 0.7905,
      "mean_token_accuracy": 0.7786333004633585,
      "step": 5810
    },
    {
      "epoch": 2.4926811852909676,
      "grad_norm": 0.298828125,
      "learning_rate": 3.8817041893674407e-05,
      "loss": 0.7815,
      "mean_token_accuracy": 0.7784610877434412,
      "step": 5820
    },
    {
      "epoch": 2.496965369510889,
      "grad_norm": 0.310546875,
      "learning_rate": 3.818073748510406e-05,
      "loss": 0.7908,
      "mean_token_accuracy": 0.7784650901953379,
      "step": 5830
    },
    {
      "epoch": 2.50124955373081,
      "grad_norm": 0.3046875,
      "learning_rate": 3.754926011312137e-05,
      "loss": 0.7778,
      "mean_token_accuracy": 0.7789822975794475,
      "step": 5840
    },
    {
      "epoch": 2.505533737950732,
      "grad_norm": 0.333984375,
      "learning_rate": 3.692262416812869e-05,
      "loss": 0.7799,
      "mean_token_accuracy": 0.7808815310398738,
      "step": 5850
    },
    {
      "epoch": 2.5098179221706536,
      "grad_norm": 0.291015625,
      "learning_rate": 3.63008439301995e-05,
      "loss": 0.7668,
      "mean_token_accuracy": 0.7802990694840749,
      "step": 5860
    },
    {
      "epoch": 2.514102106390575,
      "grad_norm": 0.30859375,
      "learning_rate": 3.568393356875305e-05,
      "loss": 0.7787,
      "mean_token_accuracy": 0.7813133666912715,
      "step": 5870
    },
    {
      "epoch": 2.518386290610496,
      "grad_norm": 0.330078125,
      "learning_rate": 3.507190714223168e-05,
      "loss": 0.7769,
      "mean_token_accuracy": 0.7826229612032573,
      "step": 5880
    },
    {
      "epoch": 2.522670474830418,
      "grad_norm": 0.3125,
      "learning_rate": 3.446477859778041e-05,
      "loss": 0.78,
      "mean_token_accuracy": 0.7789117127656937,
      "step": 5890
    },
    {
      "epoch": 2.526954659050339,
      "grad_norm": 0.291015625,
      "learning_rate": 3.386256177092886e-05,
      "loss": 0.7882,
      "mean_token_accuracy": 0.7778934846321742,
      "step": 5900
    },
    {
      "epoch": 2.531238843270261,
      "grad_norm": 0.318359375,
      "learning_rate": 3.3265270385276296e-05,
      "loss": 0.7719,
      "mean_token_accuracy": 0.7834851761658986,
      "step": 5910
    },
    {
      "epoch": 2.535523027490182,
      "grad_norm": 0.31640625,
      "learning_rate": 3.267291805217851e-05,
      "loss": 0.7863,
      "mean_token_accuracy": 0.7806941578785579,
      "step": 5920
    },
    {
      "epoch": 2.5398072117101034,
      "grad_norm": 0.3125,
      "learning_rate": 3.208551827043804e-05,
      "loss": 0.7656,
      "mean_token_accuracy": 0.7854113181432089,
      "step": 5930
    },
    {
      "epoch": 2.544091395930025,
      "grad_norm": 0.33203125,
      "learning_rate": 3.150308442599631e-05,
      "loss": 0.7658,
      "mean_token_accuracy": 0.783687628308932,
      "step": 5940
    },
    {
      "epoch": 2.5483755801499464,
      "grad_norm": 0.3125,
      "learning_rate": 3.092562979162864e-05,
      "loss": 0.8225,
      "mean_token_accuracy": 0.7703000376621882,
      "step": 5950
    },
    {
      "epoch": 2.552659764369868,
      "grad_norm": 0.298828125,
      "learning_rate": 3.0353167526641745e-05,
      "loss": 0.7792,
      "mean_token_accuracy": 0.7798318127791087,
      "step": 5960
    },
    {
      "epoch": 2.5569439485897894,
      "grad_norm": 0.3046875,
      "learning_rate": 2.97857106765739e-05,
      "loss": 0.7706,
      "mean_token_accuracy": 0.7838211774826049,
      "step": 5970
    },
    {
      "epoch": 2.5612281328097106,
      "grad_norm": 0.326171875,
      "learning_rate": 2.9223272172897607e-05,
      "loss": 0.7815,
      "mean_token_accuracy": 0.7787001341581344,
      "step": 5980
    },
    {
      "epoch": 2.5655123170296323,
      "grad_norm": 0.33984375,
      "learning_rate": 2.866586483272507e-05,
      "loss": 0.7745,
      "mean_token_accuracy": 0.780657422542572,
      "step": 5990
    },
    {
      "epoch": 2.5697965012495536,
      "grad_norm": 0.294921875,
      "learning_rate": 2.8113501358515813e-05,
      "loss": 0.7755,
      "mean_token_accuracy": 0.7821098357439041,
      "step": 6000
    },
    {
      "epoch": 2.5740806854694753,
      "grad_norm": 0.345703125,
      "learning_rate": 2.7566194337787507e-05,
      "loss": 0.8095,
      "mean_token_accuracy": 0.7710189620653788,
      "step": 6010
    },
    {
      "epoch": 2.5783648696893966,
      "grad_norm": 0.3203125,
      "learning_rate": 2.7023956242828968e-05,
      "loss": 0.7781,
      "mean_token_accuracy": 0.7805722614129385,
      "step": 6020
    },
    {
      "epoch": 2.582649053909318,
      "grad_norm": 0.328125,
      "learning_rate": 2.6486799430415875e-05,
      "loss": 0.7774,
      "mean_token_accuracy": 0.7799272914727529,
      "step": 6030
    },
    {
      "epoch": 2.5869332381292396,
      "grad_norm": 0.333984375,
      "learning_rate": 2.595473614152932e-05,
      "loss": 0.7781,
      "mean_token_accuracy": 0.7814104864994685,
      "step": 6040
    },
    {
      "epoch": 2.591217422349161,
      "grad_norm": 0.33203125,
      "learning_rate": 2.5427778501076804e-05,
      "loss": 0.7885,
      "mean_token_accuracy": 0.7791467080513637,
      "step": 6050
    },
    {
      "epoch": 2.5955016065690826,
      "grad_norm": 0.291015625,
      "learning_rate": 2.490593851761591e-05,
      "loss": 0.7689,
      "mean_token_accuracy": 0.7836964478095373,
      "step": 6060
    },
    {
      "epoch": 2.599785790789004,
      "grad_norm": 0.322265625,
      "learning_rate": 2.4389228083080722e-05,
      "loss": 0.8215,
      "mean_token_accuracy": 0.7677028367916743,
      "step": 6070
    },
    {
      "epoch": 2.604069975008925,
      "grad_norm": 0.328125,
      "learning_rate": 2.387765897251057e-05,
      "loss": 0.7749,
      "mean_token_accuracy": 0.7838013221820196,
      "step": 6080
    },
    {
      "epoch": 2.608354159228847,
      "grad_norm": 0.314453125,
      "learning_rate": 2.3371242843782088e-05,
      "loss": 0.7725,
      "mean_token_accuracy": 0.7843282967805862,
      "step": 6090
    },
    {
      "epoch": 2.6126383434487686,
      "grad_norm": 0.328125,
      "learning_rate": 2.2869991237343207e-05,
      "loss": 0.797,
      "mean_token_accuracy": 0.7766178419192632,
      "step": 6100
    },
    {
      "epoch": 2.61692252766869,
      "grad_norm": 0.330078125,
      "learning_rate": 2.237391557595042e-05,
      "loss": 0.7642,
      "mean_token_accuracy": 0.7866159160931905,
      "step": 6110
    },
    {
      "epoch": 2.621206711888611,
      "grad_norm": 0.32421875,
      "learning_rate": 2.188302716440832e-05,
      "loss": 0.7788,
      "mean_token_accuracy": 0.7817911605040232,
      "step": 6120
    },
    {
      "epoch": 2.625490896108533,
      "grad_norm": 0.31640625,
      "learning_rate": 2.1397337189311915e-05,
      "loss": 0.7902,
      "mean_token_accuracy": 0.7791908890008926,
      "step": 6130
    },
    {
      "epoch": 2.629775080328454,
      "grad_norm": 0.33203125,
      "learning_rate": 2.091685671879187e-05,
      "loss": 0.7905,
      "mean_token_accuracy": 0.7808248698711395,
      "step": 6140
    },
    {
      "epoch": 2.634059264548376,
      "grad_norm": 0.314453125,
      "learning_rate": 2.044159670226245e-05,
      "loss": 0.7572,
      "mean_token_accuracy": 0.7873713413874308,
      "step": 6150
    },
    {
      "epoch": 2.638343448768297,
      "grad_norm": 0.333984375,
      "learning_rate": 1.9971567970171355e-05,
      "loss": 0.7987,
      "mean_token_accuracy": 0.7766677429278691,
      "step": 6160
    },
    {
      "epoch": 2.6426276329882183,
      "grad_norm": 0.298828125,
      "learning_rate": 1.95067812337536e-05,
      "loss": 0.7688,
      "mean_token_accuracy": 0.7839142779509226,
      "step": 6170
    },
    {
      "epoch": 2.64691181720814,
      "grad_norm": 0.330078125,
      "learning_rate": 1.9047247084787112e-05,
      "loss": 0.7602,
      "mean_token_accuracy": 0.7852412790060044,
      "step": 6180
    },
    {
      "epoch": 2.6511960014280613,
      "grad_norm": 0.341796875,
      "learning_rate": 1.8592975995351257e-05,
      "loss": 0.762,
      "mean_token_accuracy": 0.7860971083243687,
      "step": 6190
    },
    {
      "epoch": 2.655480185647983,
      "grad_norm": 0.306640625,
      "learning_rate": 1.81439783175886e-05,
      "loss": 0.7658,
      "mean_token_accuracy": 0.7843163589636485,
      "step": 6200
    },
    {
      "epoch": 2.6597643698679043,
      "grad_norm": 0.3125,
      "learning_rate": 1.7700264283468465e-05,
      "loss": 0.7793,
      "mean_token_accuracy": 0.7768728653589885,
      "step": 6210
    },
    {
      "epoch": 2.6640485540878256,
      "grad_norm": 0.322265625,
      "learning_rate": 1.7261844004554105e-05,
      "loss": 0.7862,
      "mean_token_accuracy": 0.7782338261604309,
      "step": 6220
    },
    {
      "epoch": 2.6683327383077473,
      "grad_norm": 0.341796875,
      "learning_rate": 1.6828727471772358e-05,
      "loss": 0.7873,
      "mean_token_accuracy": 0.7764876892169317,
      "step": 6230
    },
    {
      "epoch": 2.6726169225276686,
      "grad_norm": 0.31640625,
      "learning_rate": 1.6400924555185492e-05,
      "loss": 0.7774,
      "mean_token_accuracy": 0.7807328204313914,
      "step": 6240
    },
    {
      "epoch": 2.6769011067475903,
      "grad_norm": 0.310546875,
      "learning_rate": 1.5978445003766968e-05,
      "loss": 0.786,
      "mean_token_accuracy": 0.7790028562148412,
      "step": 6250
    },
    {
      "epoch": 2.6811852909675116,
      "grad_norm": 0.30859375,
      "learning_rate": 1.5561298445178617e-05,
      "loss": 0.7511,
      "mean_token_accuracy": 0.7876404513915379,
      "step": 6260
    },
    {
      "epoch": 2.685469475187433,
      "grad_norm": 0.337890625,
      "learning_rate": 1.5149494385551688e-05,
      "loss": 0.781,
      "mean_token_accuracy": 0.7786963661511739,
      "step": 6270
    },
    {
      "epoch": 2.6897536594073546,
      "grad_norm": 0.318359375,
      "learning_rate": 1.474304220927003e-05,
      "loss": 0.7837,
      "mean_token_accuracy": 0.780462098121643,
      "step": 6280
    },
    {
      "epoch": 2.694037843627276,
      "grad_norm": 0.318359375,
      "learning_rate": 1.4341951178756168e-05,
      "loss": 0.7817,
      "mean_token_accuracy": 0.7801312992970149,
      "step": 6290
    },
    {
      "epoch": 2.6983220278471975,
      "grad_norm": 0.3203125,
      "learning_rate": 1.3946230434260493e-05,
      "loss": 0.7837,
      "mean_token_accuracy": 0.7814756115277608,
      "step": 6300
    },
    {
      "epoch": 2.702606212067119,
      "grad_norm": 0.31640625,
      "learning_rate": 1.3555888993652732e-05,
      "loss": 0.7825,
      "mean_token_accuracy": 0.7780475705862046,
      "step": 6310
    },
    {
      "epoch": 2.70689039628704,
      "grad_norm": 0.333984375,
      "learning_rate": 1.3170935752216423e-05,
      "loss": 0.7856,
      "mean_token_accuracy": 0.7791298975547155,
      "step": 6320
    },
    {
      "epoch": 2.711174580506962,
      "grad_norm": 0.318359375,
      "learning_rate": 1.2791379482446407e-05,
      "loss": 0.7796,
      "mean_token_accuracy": 0.7799790789683659,
      "step": 6330
    },
    {
      "epoch": 2.7154587647268835,
      "grad_norm": 0.337890625,
      "learning_rate": 1.2417228833848798e-05,
      "loss": 0.7748,
      "mean_token_accuracy": 0.7825185318787893,
      "step": 6340
    },
    {
      "epoch": 2.719742948946805,
      "grad_norm": 0.326171875,
      "learning_rate": 1.2048492332743827e-05,
      "loss": 0.7963,
      "mean_token_accuracy": 0.7784641563892365,
      "step": 6350
    },
    {
      "epoch": 2.724027133166726,
      "grad_norm": 0.3203125,
      "learning_rate": 1.1685178382071698e-05,
      "loss": 0.7778,
      "mean_token_accuracy": 0.7813698927561442,
      "step": 6360
    },
    {
      "epoch": 2.728311317386648,
      "grad_norm": 0.291015625,
      "learning_rate": 1.1327295261200826e-05,
      "loss": 0.7605,
      "mean_token_accuracy": 0.7861085881789526,
      "step": 6370
    },
    {
      "epoch": 2.732595501606569,
      "grad_norm": 0.3125,
      "learning_rate": 1.0974851125739483e-05,
      "loss": 0.7647,
      "mean_token_accuracy": 0.7843359092871348,
      "step": 6380
    },
    {
      "epoch": 2.7368796858264908,
      "grad_norm": 0.318359375,
      "learning_rate": 1.0627854007349725e-05,
      "loss": 0.7771,
      "mean_token_accuracy": 0.7815760542949041,
      "step": 6390
    },
    {
      "epoch": 2.741163870046412,
      "grad_norm": 0.33203125,
      "learning_rate": 1.0286311813564487e-05,
      "loss": 0.7819,
      "mean_token_accuracy": 0.7804714103539785,
      "step": 6400
    },
    {
      "epoch": 2.7454480542663333,
      "grad_norm": 0.314453125,
      "learning_rate": 9.950232327607278e-06,
      "loss": 0.7887,
      "mean_token_accuracy": 0.7779040902853012,
      "step": 6410
    },
    {
      "epoch": 2.749732238486255,
      "grad_norm": 0.32421875,
      "learning_rate": 9.619623208214801e-06,
      "loss": 0.7795,
      "mean_token_accuracy": 0.7815526028474172,
      "step": 6420
    },
    {
      "epoch": 2.7540164227061763,
      "grad_norm": 0.314453125,
      "learning_rate": 9.29449198946264e-06,
      "loss": 0.7577,
      "mean_token_accuracy": 0.7885059833526611,
      "step": 6430
    },
    {
      "epoch": 2.758300606926098,
      "grad_norm": 0.330078125,
      "learning_rate": 8.974846080593262e-06,
      "loss": 0.7688,
      "mean_token_accuracy": 0.7825045188268026,
      "step": 6440
    },
    {
      "epoch": 2.7625847911460193,
      "grad_norm": 0.296875,
      "learning_rate": 8.66069276584741e-06,
      "loss": 0.7639,
      "mean_token_accuracy": 0.7843573103348415,
      "step": 6450
    },
    {
      "epoch": 2.7668689753659406,
      "grad_norm": 0.279296875,
      "learning_rate": 8.352039204298029e-06,
      "loss": 0.7457,
      "mean_token_accuracy": 0.7897063046693802,
      "step": 6460
    },
    {
      "epoch": 2.7711531595858623,
      "grad_norm": 0.318359375,
      "learning_rate": 8.048892429687066e-06,
      "loss": 0.7661,
      "mean_token_accuracy": 0.7825449128945668,
      "step": 6470
    },
    {
      "epoch": 2.7754373438057836,
      "grad_norm": 0.306640625,
      "learning_rate": 7.751259350265216e-06,
      "loss": 0.7502,
      "mean_token_accuracy": 0.7883096744616827,
      "step": 6480
    },
    {
      "epoch": 2.7797215280257053,
      "grad_norm": 0.31640625,
      "learning_rate": 7.459146748634516e-06,
      "loss": 0.7681,
      "mean_token_accuracy": 0.7826910416285197,
      "step": 6490
    },
    {
      "epoch": 2.7840057122456265,
      "grad_norm": 0.298828125,
      "learning_rate": 7.172561281593798e-06,
      "loss": 0.7842,
      "mean_token_accuracy": 0.7800470501184463,
      "step": 6500
    },
    {
      "epoch": 2.788289896465548,
      "grad_norm": 0.302734375,
      "learning_rate": 6.891509479986957e-06,
      "loss": 0.7857,
      "mean_token_accuracy": 0.7821651329596837,
      "step": 6510
    },
    {
      "epoch": 2.7925740806854695,
      "grad_norm": 0.330078125,
      "learning_rate": 6.615997748554148e-06,
      "loss": 0.7631,
      "mean_token_accuracy": 0.7860769232114156,
      "step": 6520
    },
    {
      "epoch": 2.796858264905391,
      "grad_norm": 0.318359375,
      "learning_rate": 6.346032365785709e-06,
      "loss": 0.8009,
      "mean_token_accuracy": 0.7765180408954621,
      "step": 6530
    },
    {
      "epoch": 2.8011424491253125,
      "grad_norm": 0.30078125,
      "learning_rate": 6.081619483779277e-06,
      "loss": 0.7809,
      "mean_token_accuracy": 0.7795830368995667,
      "step": 6540
    },
    {
      "epoch": 2.805426633345234,
      "grad_norm": 0.322265625,
      "learning_rate": 5.822765128099483e-06,
      "loss": 0.764,
      "mean_token_accuracy": 0.7845493823289871,
      "step": 6550
    },
    {
      "epoch": 2.809710817565155,
      "grad_norm": 0.3125,
      "learning_rate": 5.569475197640672e-06,
      "loss": 0.7747,
      "mean_token_accuracy": 0.780571989218394,
      "step": 6560
    },
    {
      "epoch": 2.8139950017850768,
      "grad_norm": 0.296875,
      "learning_rate": 5.321755464492456e-06,
      "loss": 0.7897,
      "mean_token_accuracy": 0.7762352069218953,
      "step": 6570
    },
    {
      "epoch": 2.818279186004998,
      "grad_norm": 0.296875,
      "learning_rate": 5.079611573808124e-06,
      "loss": 0.7919,
      "mean_token_accuracy": 0.7770307193199794,
      "step": 6580
    },
    {
      "epoch": 2.8225633702249198,
      "grad_norm": 0.314453125,
      "learning_rate": 4.843049043676079e-06,
      "loss": 0.7841,
      "mean_token_accuracy": 0.7784975161155064,
      "step": 6590
    },
    {
      "epoch": 2.826847554444841,
      "grad_norm": 0.34375,
      "learning_rate": 4.61207326499416e-06,
      "loss": 0.8001,
      "mean_token_accuracy": 0.7755351980527242,
      "step": 6600
    },
    {
      "epoch": 2.8311317386647623,
      "grad_norm": 0.3125,
      "learning_rate": 4.386689501346574e-06,
      "loss": 0.8,
      "mean_token_accuracy": 0.7723700026671092,
      "step": 6610
    },
    {
      "epoch": 2.835415922884684,
      "grad_norm": 0.306640625,
      "learning_rate": 4.166902888884105e-06,
      "loss": 0.7928,
      "mean_token_accuracy": 0.7796116421620051,
      "step": 6620
    },
    {
      "epoch": 2.8397001071046057,
      "grad_norm": 0.33203125,
      "learning_rate": 3.952718436207065e-06,
      "loss": 0.7813,
      "mean_token_accuracy": 0.7800139904022216,
      "step": 6630
    },
    {
      "epoch": 2.843984291324527,
      "grad_norm": 0.302734375,
      "learning_rate": 3.7441410242510796e-06,
      "loss": 0.7706,
      "mean_token_accuracy": 0.7841657598813375,
      "step": 6640
    },
    {
      "epoch": 2.8482684755444483,
      "grad_norm": 0.333984375,
      "learning_rate": 3.5411754061759614e-06,
      "loss": 0.7877,
      "mean_token_accuracy": 0.7804912636677425,
      "step": 6650
    },
    {
      "epoch": 2.85255265976437,
      "grad_norm": 0.306640625,
      "learning_rate": 3.3438262072572612e-06,
      "loss": 0.7821,
      "mean_token_accuracy": 0.7790777862071991,
      "step": 6660
    },
    {
      "epoch": 2.8568368439842913,
      "grad_norm": 0.4140625,
      "learning_rate": 3.1520979247810223e-06,
      "loss": 0.7877,
      "mean_token_accuracy": 0.7802624036868413,
      "step": 6670
    },
    {
      "epoch": 2.861121028204213,
      "grad_norm": 0.30859375,
      "learning_rate": 2.9659949279411404e-06,
      "loss": 0.7736,
      "mean_token_accuracy": 0.7815846174955368,
      "step": 6680
    },
    {
      "epoch": 2.8654052124241343,
      "grad_norm": 0.31640625,
      "learning_rate": 2.785521457739859e-06,
      "loss": 0.7598,
      "mean_token_accuracy": 0.7873433222373326,
      "step": 6690
    },
    {
      "epoch": 2.8696893966440555,
      "grad_norm": 0.294921875,
      "learning_rate": 2.610681626891237e-06,
      "loss": 0.773,
      "mean_token_accuracy": 0.7824377208948136,
      "step": 6700
    },
    {
      "epoch": 2.8739735808639773,
      "grad_norm": 0.298828125,
      "learning_rate": 2.4414794197272217e-06,
      "loss": 0.8005,
      "mean_token_accuracy": 0.7779265910387039,
      "step": 6710
    },
    {
      "epoch": 2.8782577650838985,
      "grad_norm": 0.322265625,
      "learning_rate": 2.277918692106973e-06,
      "loss": 0.7652,
      "mean_token_accuracy": 0.7827598293622334,
      "step": 6720
    },
    {
      "epoch": 2.8825419493038202,
      "grad_norm": 0.322265625,
      "learning_rate": 2.120003171328988e-06,
      "loss": 0.8107,
      "mean_token_accuracy": 0.7738327503204345,
      "step": 6730
    },
    {
      "epoch": 2.8868261335237415,
      "grad_norm": 0.3046875,
      "learning_rate": 1.9677364560460874e-06,
      "loss": 0.7964,
      "mean_token_accuracy": 0.7783457924922307,
      "step": 6740
    },
    {
      "epoch": 2.891110317743663,
      "grad_norm": 0.30859375,
      "learning_rate": 1.8211220161835629e-06,
      "loss": 0.7952,
      "mean_token_accuracy": 0.778786172469457,
      "step": 6750
    },
    {
      "epoch": 2.8953945019635845,
      "grad_norm": 0.33984375,
      "learning_rate": 1.6801631928599626e-06,
      "loss": 0.7956,
      "mean_token_accuracy": 0.7760281264781952,
      "step": 6760
    },
    {
      "epoch": 2.8996786861835058,
      "grad_norm": 0.306640625,
      "learning_rate": 1.5448631983109584e-06,
      "loss": 0.7765,
      "mean_token_accuracy": 0.7821375171343485,
      "step": 6770
    },
    {
      "epoch": 2.9039628704034275,
      "grad_norm": 0.333984375,
      "learning_rate": 1.41522511581621e-06,
      "loss": 0.7765,
      "mean_token_accuracy": 0.7809682806332906,
      "step": 6780
    },
    {
      "epoch": 2.9082470546233488,
      "grad_norm": 0.318359375,
      "learning_rate": 1.2912518996290866e-06,
      "loss": 0.7705,
      "mean_token_accuracy": 0.786265421907107,
      "step": 6790
    },
    {
      "epoch": 2.91253123884327,
      "grad_norm": 0.322265625,
      "learning_rate": 1.1729463749093338e-06,
      "loss": 0.7982,
      "mean_token_accuracy": 0.7765524844328563,
      "step": 6800
    },
    {
      "epoch": 2.9168154230631917,
      "grad_norm": 0.291015625,
      "learning_rate": 1.060311237658651e-06,
      "loss": 0.773,
      "mean_token_accuracy": 0.7809695929288865,
      "step": 6810
    },
    {
      "epoch": 2.921099607283113,
      "grad_norm": 0.302734375,
      "learning_rate": 9.533490546593248e-07,
      "loss": 0.774,
      "mean_token_accuracy": 0.7783287167549133,
      "step": 6820
    },
    {
      "epoch": 2.9253837915030347,
      "grad_norm": 0.314453125,
      "learning_rate": 8.520622634156927e-07,
      "loss": 0.8114,
      "mean_token_accuracy": 0.7732517321904501,
      "step": 6830
    },
    {
      "epoch": 2.929667975722956,
      "grad_norm": 0.30859375,
      "learning_rate": 7.564531720985756e-07,
      "loss": 0.7683,
      "mean_token_accuracy": 0.7860218505064647,
      "step": 6840
    },
    {
      "epoch": 2.9339521599428773,
      "grad_norm": 0.3125,
      "learning_rate": 6.665239594927929e-07,
      "loss": 0.7902,
      "mean_token_accuracy": 0.780942557255427,
      "step": 6850
    },
    {
      "epoch": 2.938236344162799,
      "grad_norm": 0.3203125,
      "learning_rate": 5.82276674947313e-07,
      "loss": 0.7628,
      "mean_token_accuracy": 0.7842933177947998,
      "step": 6860
    },
    {
      "epoch": 2.9425205283827207,
      "grad_norm": 0.302734375,
      "learning_rate": 5.037132383287624e-07,
      "loss": 0.7806,
      "mean_token_accuracy": 0.7809433400630951,
      "step": 6870
    },
    {
      "epoch": 2.946804712602642,
      "grad_norm": 0.326171875,
      "learning_rate": 4.308354399775172e-07,
      "loss": 0.7796,
      "mean_token_accuracy": 0.7829487164815266,
      "step": 6880
    },
    {
      "epoch": 2.9510888968225633,
      "grad_norm": 0.3125,
      "learning_rate": 3.636449406670128e-07,
      "loss": 0.7785,
      "mean_token_accuracy": 0.7833139419555664,
      "step": 6890
    },
    {
      "epoch": 2.955373081042485,
      "grad_norm": 0.3046875,
      "learning_rate": 3.021432715658023e-07,
      "loss": 0.776,
      "mean_token_accuracy": 0.7816599200169245,
      "step": 6900
    },
    {
      "epoch": 2.9596572652624062,
      "grad_norm": 0.31640625,
      "learning_rate": 2.4633183420275093e-07,
      "loss": 0.7895,
      "mean_token_accuracy": 0.7800428807735443,
      "step": 6910
    },
    {
      "epoch": 2.963941449482328,
      "grad_norm": 0.310546875,
      "learning_rate": 1.9621190043506155e-07,
      "loss": 0.7643,
      "mean_token_accuracy": 0.7820758432149887,
      "step": 6920
    },
    {
      "epoch": 2.9682256337022492,
      "grad_norm": 0.328125,
      "learning_rate": 1.5178461241918684e-07,
      "loss": 0.7962,
      "mean_token_accuracy": 0.7780212819576263,
      "step": 6930
    },
    {
      "epoch": 2.9725098179221705,
      "grad_norm": 0.33203125,
      "learning_rate": 1.1305098258504454e-07,
      "loss": 0.7694,
      "mean_token_accuracy": 0.7845108538866044,
      "step": 6940
    },
    {
      "epoch": 2.976794002142092,
      "grad_norm": 0.32421875,
      "learning_rate": 8.001189361273032e-08,
      "loss": 0.7709,
      "mean_token_accuracy": 0.7825704693794251,
      "step": 6950
    },
    {
      "epoch": 2.9810781863620135,
      "grad_norm": 0.33203125,
      "learning_rate": 5.266809841247833e-08,
      "loss": 0.7737,
      "mean_token_accuracy": 0.7819208929936091,
      "step": 6960
    },
    {
      "epoch": 2.985362370581935,
      "grad_norm": 0.310546875,
      "learning_rate": 3.1020220107480513e-08,
      "loss": 0.7593,
      "mean_token_accuracy": 0.7868289381265641,
      "step": 6970
    },
    {
      "epoch": 2.9896465548018565,
      "grad_norm": 0.31640625,
      "learning_rate": 1.506875201975899e-08,
      "loss": 0.8075,
      "mean_token_accuracy": 0.7752761413653692,
      "step": 6980
    },
    {
      "epoch": 2.9939307390217778,
      "grad_norm": 0.30859375,
      "learning_rate": 4.8140576588140415e-09,
      "loss": 0.7801,
      "mean_token_accuracy": 0.7793139110008875,
      "step": 6990
    },
    {
      "epoch": 2.9982149232416995,
      "grad_norm": 0.326171875,
      "learning_rate": 2.5637071346396034e-10,
      "loss": 0.7702,
      "mean_token_accuracy": 0.7825540274381637,
      "step": 7000
    }
  ],
  "logging_steps": 10,
  "max_steps": 7002,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 1.3733801424314171e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}