{
  "best_metric": 10.32548999786377,
  "best_model_checkpoint": "miner_id_24/checkpoint-400",
  "epoch": 0.058722061144346166,
  "eval_steps": 50,
  "global_step": 400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00014680515286086542,
      "grad_norm": 0.08386878669261932,
      "learning_rate": 1e-05,
      "loss": 10.3866,
      "step": 1
    },
    {
      "epoch": 0.00014680515286086542,
      "eval_loss": 10.377557754516602,
      "eval_runtime": 41.0485,
      "eval_samples_per_second": 279.499,
      "eval_steps_per_second": 69.893,
      "step": 1
    },
    {
      "epoch": 0.00029361030572173084,
      "grad_norm": 0.07220980525016785,
      "learning_rate": 2e-05,
      "loss": 10.3816,
      "step": 2
    },
    {
      "epoch": 0.00044041545858259626,
      "grad_norm": 0.06736605614423752,
      "learning_rate": 3e-05,
      "loss": 10.38,
      "step": 3
    },
    {
      "epoch": 0.0005872206114434617,
      "grad_norm": 0.06941753625869751,
      "learning_rate": 4e-05,
      "loss": 10.3777,
      "step": 4
    },
    {
      "epoch": 0.0007340257643043271,
      "grad_norm": 0.04316747188568115,
      "learning_rate": 5e-05,
      "loss": 10.3809,
      "step": 5
    },
    {
      "epoch": 0.0008808309171651925,
      "grad_norm": 0.05745130404829979,
      "learning_rate": 6e-05,
      "loss": 10.3809,
      "step": 6
    },
    {
      "epoch": 0.0010276360700260578,
      "grad_norm": 0.05900517851114273,
      "learning_rate": 7e-05,
      "loss": 10.377,
      "step": 7
    },
    {
      "epoch": 0.0011744412228869234,
      "grad_norm": 0.05711089074611664,
      "learning_rate": 8e-05,
      "loss": 10.3778,
      "step": 8
    },
    {
      "epoch": 0.0013212463757477887,
      "grad_norm": 0.05510803684592247,
      "learning_rate": 9e-05,
      "loss": 10.3779,
      "step": 9
    },
    {
      "epoch": 0.0014680515286086542,
      "grad_norm": 0.049112264066934586,
      "learning_rate": 0.0001,
      "loss": 10.3796,
      "step": 10
    },
    {
      "epoch": 0.0016148566814695195,
      "grad_norm": 0.04304530471563339,
      "learning_rate": 9.99983777858264e-05,
      "loss": 10.3803,
      "step": 11
    },
    {
      "epoch": 0.001761661834330385,
      "grad_norm": 0.045705340802669525,
      "learning_rate": 9.999351124856874e-05,
      "loss": 10.379,
      "step": 12
    },
    {
      "epoch": 0.0019084669871912503,
      "grad_norm": 0.05853036418557167,
      "learning_rate": 9.998540070400966e-05,
      "loss": 10.3789,
      "step": 13
    },
    {
      "epoch": 0.0020552721400521157,
      "grad_norm": 0.04991290718317032,
      "learning_rate": 9.997404667843075e-05,
      "loss": 10.3783,
      "step": 14
    },
    {
      "epoch": 0.0022020772929129814,
      "grad_norm": 0.05034385249018669,
      "learning_rate": 9.995944990857849e-05,
      "loss": 10.3741,
      "step": 15
    },
    {
      "epoch": 0.0023488824457738467,
      "grad_norm": 0.060934100300073624,
      "learning_rate": 9.994161134161634e-05,
      "loss": 10.374,
      "step": 16
    },
    {
      "epoch": 0.002495687598634712,
      "grad_norm": 0.057455215603113174,
      "learning_rate": 9.992053213506334e-05,
      "loss": 10.3744,
      "step": 17
    },
    {
      "epoch": 0.0026424927514955773,
      "grad_norm": 0.055931784212589264,
      "learning_rate": 9.989621365671902e-05,
      "loss": 10.3768,
      "step": 18
    },
    {
      "epoch": 0.002789297904356443,
      "grad_norm": 0.06356349587440491,
      "learning_rate": 9.986865748457457e-05,
      "loss": 10.3743,
      "step": 19
    },
    {
      "epoch": 0.0029361030572173084,
      "grad_norm": 0.06787053495645523,
      "learning_rate": 9.983786540671051e-05,
      "loss": 10.3727,
      "step": 20
    },
    {
      "epoch": 0.0030829082100781737,
      "grad_norm": 0.06242568790912628,
      "learning_rate": 9.980383942118066e-05,
      "loss": 10.3719,
      "step": 21
    },
    {
      "epoch": 0.003229713362939039,
      "grad_norm": 0.07491078972816467,
      "learning_rate": 9.976658173588244e-05,
      "loss": 10.372,
      "step": 22
    },
    {
      "epoch": 0.0033765185157999048,
      "grad_norm": 0.07711949944496155,
      "learning_rate": 9.972609476841367e-05,
      "loss": 10.3709,
      "step": 23
    },
    {
      "epoch": 0.00352332366866077,
      "grad_norm": 0.06691116839647293,
      "learning_rate": 9.968238114591566e-05,
      "loss": 10.373,
      "step": 24
    },
    {
      "epoch": 0.0036701288215216354,
      "grad_norm": 0.08010495454072952,
      "learning_rate": 9.96354437049027e-05,
      "loss": 10.3715,
      "step": 25
    },
    {
      "epoch": 0.0038169339743825007,
      "grad_norm": 0.08388824760913849,
      "learning_rate": 9.95852854910781e-05,
      "loss": 10.3715,
      "step": 26
    },
    {
      "epoch": 0.003963739127243366,
      "grad_norm": 0.08033105731010437,
      "learning_rate": 9.953190975913647e-05,
      "loss": 10.3747,
      "step": 27
    },
    {
      "epoch": 0.004110544280104231,
      "grad_norm": 0.09592588245868683,
      "learning_rate": 9.947531997255256e-05,
      "loss": 10.3661,
      "step": 28
    },
    {
      "epoch": 0.0042573494329650975,
      "grad_norm": 0.0965273454785347,
      "learning_rate": 9.941551980335652e-05,
      "loss": 10.3677,
      "step": 29
    },
    {
      "epoch": 0.004404154585825963,
      "grad_norm": 0.10320152342319489,
      "learning_rate": 9.935251313189564e-05,
      "loss": 10.3663,
      "step": 30
    },
    {
      "epoch": 0.004550959738686828,
      "grad_norm": 0.1014312133193016,
      "learning_rate": 9.928630404658255e-05,
      "loss": 10.3673,
      "step": 31
    },
    {
      "epoch": 0.004697764891547693,
      "grad_norm": 0.09669344127178192,
      "learning_rate": 9.921689684362989e-05,
      "loss": 10.3698,
      "step": 32
    },
    {
      "epoch": 0.004844570044408559,
      "grad_norm": 0.12578672170639038,
      "learning_rate": 9.914429602677162e-05,
      "loss": 10.3669,
      "step": 33
    },
    {
      "epoch": 0.004991375197269424,
      "grad_norm": 0.10694191604852676,
      "learning_rate": 9.906850630697068e-05,
      "loss": 10.3643,
      "step": 34
    },
    {
      "epoch": 0.005138180350130289,
      "grad_norm": 0.1083347275853157,
      "learning_rate": 9.898953260211338e-05,
      "loss": 10.3663,
      "step": 35
    },
    {
      "epoch": 0.005284985502991155,
      "grad_norm": 0.11611084640026093,
      "learning_rate": 9.890738003669029e-05,
      "loss": 10.3648,
      "step": 36
    },
    {
      "epoch": 0.005431790655852021,
      "grad_norm": 0.1346110850572586,
      "learning_rate": 9.882205394146361e-05,
      "loss": 10.3648,
      "step": 37
    },
    {
      "epoch": 0.005578595808712886,
      "grad_norm": 0.12299930304288864,
      "learning_rate": 9.87335598531214e-05,
      "loss": 10.3647,
      "step": 38
    },
    {
      "epoch": 0.0057254009615737515,
      "grad_norm": 0.1609814614057541,
      "learning_rate": 9.864190351391822e-05,
      "loss": 10.3652,
      "step": 39
    },
    {
      "epoch": 0.005872206114434617,
      "grad_norm": 0.11613401025533676,
      "learning_rate": 9.85470908713026e-05,
      "loss": 10.3575,
      "step": 40
    },
    {
      "epoch": 0.006019011267295482,
      "grad_norm": 0.11813950538635254,
      "learning_rate": 9.844912807753104e-05,
      "loss": 10.3616,
      "step": 41
    },
    {
      "epoch": 0.006165816420156347,
      "grad_norm": 0.1686146855354309,
      "learning_rate": 9.834802148926882e-05,
      "loss": 10.3625,
      "step": 42
    },
    {
      "epoch": 0.006312621573017213,
      "grad_norm": 0.13682858645915985,
      "learning_rate": 9.824377766717759e-05,
      "loss": 10.3568,
      "step": 43
    },
    {
      "epoch": 0.006459426725878078,
      "grad_norm": 0.15394780039787292,
      "learning_rate": 9.813640337548954e-05,
      "loss": 10.3548,
      "step": 44
    },
    {
      "epoch": 0.006606231878738943,
      "grad_norm": 0.10382828116416931,
      "learning_rate": 9.802590558156862e-05,
      "loss": 10.3559,
      "step": 45
    },
    {
      "epoch": 0.0067530370315998095,
      "grad_norm": 0.14495818316936493,
      "learning_rate": 9.791229145545831e-05,
      "loss": 10.3555,
      "step": 46
    },
    {
      "epoch": 0.006899842184460675,
      "grad_norm": 0.13232113420963287,
      "learning_rate": 9.779556836941645e-05,
      "loss": 10.3513,
      "step": 47
    },
    {
      "epoch": 0.00704664733732154,
      "grad_norm": 0.1268262416124344,
      "learning_rate": 9.767574389743682e-05,
      "loss": 10.3501,
      "step": 48
    },
    {
      "epoch": 0.0071934524901824054,
      "grad_norm": 0.14365290105342865,
      "learning_rate": 9.755282581475769e-05,
      "loss": 10.3493,
      "step": 49
    },
    {
      "epoch": 0.007340257643043271,
      "grad_norm": 0.3059467077255249,
      "learning_rate": 9.742682209735727e-05,
      "loss": 10.3693,
      "step": 50
    },
    {
      "epoch": 0.007340257643043271,
      "eval_loss": 10.354704856872559,
      "eval_runtime": 40.872,
      "eval_samples_per_second": 280.706,
      "eval_steps_per_second": 70.195,
      "step": 50
    },
    {
      "epoch": 0.007487062795904136,
      "grad_norm": 0.2364913523197174,
      "learning_rate": 9.729774092143627e-05,
      "loss": 10.3608,
      "step": 51
    },
    {
      "epoch": 0.007633867948765001,
      "grad_norm": 0.15694783627986908,
      "learning_rate": 9.716559066288715e-05,
      "loss": 10.3607,
      "step": 52
    },
    {
      "epoch": 0.007780673101625867,
      "grad_norm": 0.20694315433502197,
      "learning_rate": 9.703037989675087e-05,
      "loss": 10.3585,
      "step": 53
    },
    {
      "epoch": 0.007927478254486732,
      "grad_norm": 0.15253427624702454,
      "learning_rate": 9.689211739666023e-05,
      "loss": 10.357,
      "step": 54
    },
    {
      "epoch": 0.008074283407347598,
      "grad_norm": 0.15416797995567322,
      "learning_rate": 9.675081213427076e-05,
      "loss": 10.3544,
      "step": 55
    },
    {
      "epoch": 0.008221088560208463,
      "grad_norm": 0.1405656933784485,
      "learning_rate": 9.66064732786784e-05,
      "loss": 10.3543,
      "step": 56
    },
    {
      "epoch": 0.008367893713069329,
      "grad_norm": 0.10970322787761688,
      "learning_rate": 9.645911019582467e-05,
      "loss": 10.3543,
      "step": 57
    },
    {
      "epoch": 0.008514698865930195,
      "grad_norm": 0.1166229397058487,
      "learning_rate": 9.630873244788883e-05,
      "loss": 10.3498,
      "step": 58
    },
    {
      "epoch": 0.00866150401879106,
      "grad_norm": 0.08719631284475327,
      "learning_rate": 9.615534979266745e-05,
      "loss": 10.3521,
      "step": 59
    },
    {
      "epoch": 0.008808309171651926,
      "grad_norm": 0.10392676293849945,
      "learning_rate": 9.599897218294122e-05,
      "loss": 10.3479,
      "step": 60
    },
    {
      "epoch": 0.00895511432451279,
      "grad_norm": 0.0706394761800766,
      "learning_rate": 9.583960976582913e-05,
      "loss": 10.3506,
      "step": 61
    },
    {
      "epoch": 0.009101919477373656,
      "grad_norm": 0.06449541449546814,
      "learning_rate": 9.567727288213005e-05,
      "loss": 10.35,
      "step": 62
    },
    {
      "epoch": 0.00924872463023452,
      "grad_norm": 0.07657317072153091,
      "learning_rate": 9.551197206565173e-05,
      "loss": 10.3509,
      "step": 63
    },
    {
      "epoch": 0.009395529783095387,
      "grad_norm": 0.06846413761377335,
      "learning_rate": 9.534371804252728e-05,
      "loss": 10.3481,
      "step": 64
    },
    {
      "epoch": 0.009542334935956251,
      "grad_norm": 0.08481397479772568,
      "learning_rate": 9.517252173051911e-05,
      "loss": 10.3481,
      "step": 65
    },
    {
      "epoch": 0.009689140088817117,
      "grad_norm": 0.07156714051961899,
      "learning_rate": 9.49983942383106e-05,
      "loss": 10.3493,
      "step": 66
    },
    {
      "epoch": 0.009835945241677984,
      "grad_norm": 0.06248435378074646,
      "learning_rate": 9.482134686478519e-05,
      "loss": 10.349,
      "step": 67
    },
    {
      "epoch": 0.009982750394538848,
      "grad_norm": 0.06693941354751587,
      "learning_rate": 9.464139109829321e-05,
      "loss": 10.3484,
      "step": 68
    },
    {
      "epoch": 0.010129555547399714,
      "grad_norm": 0.07589535415172577,
      "learning_rate": 9.445853861590647e-05,
      "loss": 10.3489,
      "step": 69
    },
    {
      "epoch": 0.010276360700260579,
      "grad_norm": 0.07633531838655472,
      "learning_rate": 9.42728012826605e-05,
      "loss": 10.3516,
      "step": 70
    },
    {
      "epoch": 0.010423165853121445,
      "grad_norm": 0.06408070772886276,
      "learning_rate": 9.408419115078471e-05,
      "loss": 10.3511,
      "step": 71
    },
    {
      "epoch": 0.01056997100598231,
      "grad_norm": 0.05269797518849373,
      "learning_rate": 9.389272045892024e-05,
      "loss": 10.3531,
      "step": 72
    },
    {
      "epoch": 0.010716776158843175,
      "grad_norm": 0.07646088302135468,
      "learning_rate": 9.36984016313259e-05,
      "loss": 10.3479,
      "step": 73
    },
    {
      "epoch": 0.010863581311704042,
      "grad_norm": 0.06696692854166031,
      "learning_rate": 9.350124727707197e-05,
      "loss": 10.3479,
      "step": 74
    },
    {
      "epoch": 0.011010386464564906,
      "grad_norm": 0.07607897371053696,
      "learning_rate": 9.330127018922194e-05,
      "loss": 10.3493,
      "step": 75
    },
    {
      "epoch": 0.011157191617425772,
      "grad_norm": 0.076313816010952,
      "learning_rate": 9.309848334400246e-05,
      "loss": 10.3479,
      "step": 76
    },
    {
      "epoch": 0.011303996770286637,
      "grad_norm": 0.08129531145095825,
      "learning_rate": 9.289289989996133e-05,
      "loss": 10.3481,
      "step": 77
    },
    {
      "epoch": 0.011450801923147503,
      "grad_norm": 0.08560651540756226,
      "learning_rate": 9.268453319711363e-05,
      "loss": 10.3489,
      "step": 78
    },
    {
      "epoch": 0.011597607076008367,
      "grad_norm": 0.08942073583602905,
      "learning_rate": 9.247339675607605e-05,
      "loss": 10.3415,
      "step": 79
    },
    {
      "epoch": 0.011744412228869234,
      "grad_norm": 0.08754400163888931,
      "learning_rate": 9.225950427718975e-05,
      "loss": 10.349,
      "step": 80
    },
    {
      "epoch": 0.011891217381730098,
      "grad_norm": 0.10151137411594391,
      "learning_rate": 9.204286963963111e-05,
      "loss": 10.3458,
      "step": 81
    },
    {
      "epoch": 0.012038022534590964,
      "grad_norm": 0.09648429602384567,
      "learning_rate": 9.182350690051133e-05,
      "loss": 10.3426,
      "step": 82
    },
    {
      "epoch": 0.01218482768745183,
      "grad_norm": 0.09070229530334473,
      "learning_rate": 9.160143029396422e-05,
      "loss": 10.3474,
      "step": 83
    },
    {
      "epoch": 0.012331632840312695,
      "grad_norm": 0.08309106528759003,
      "learning_rate": 9.13766542302225e-05,
      "loss": 10.345,
      "step": 84
    },
    {
      "epoch": 0.012478437993173561,
      "grad_norm": 0.0835483968257904,
      "learning_rate": 9.114919329468282e-05,
      "loss": 10.3441,
      "step": 85
    },
    {
      "epoch": 0.012625243146034425,
      "grad_norm": 0.09185004979372025,
      "learning_rate": 9.091906224695935e-05,
      "loss": 10.3438,
      "step": 86
    },
    {
      "epoch": 0.012772048298895292,
      "grad_norm": 0.10591776669025421,
      "learning_rate": 9.068627601992598e-05,
      "loss": 10.3421,
      "step": 87
    },
    {
      "epoch": 0.012918853451756156,
      "grad_norm": 0.08486839383840561,
      "learning_rate": 9.045084971874738e-05,
      "loss": 10.3446,
      "step": 88
    },
    {
      "epoch": 0.013065658604617022,
      "grad_norm": 0.10287255793809891,
      "learning_rate": 9.021279861989885e-05,
      "loss": 10.3445,
      "step": 89
    },
    {
      "epoch": 0.013212463757477887,
      "grad_norm": 0.09322993457317352,
      "learning_rate": 8.997213817017507e-05,
      "loss": 10.3436,
      "step": 90
    },
    {
      "epoch": 0.013359268910338753,
      "grad_norm": 0.10085370391607285,
      "learning_rate": 8.972888398568772e-05,
      "loss": 10.3445,
      "step": 91
    },
    {
      "epoch": 0.013506074063199619,
      "grad_norm": 0.10401784628629684,
      "learning_rate": 8.948305185085225e-05,
      "loss": 10.3419,
      "step": 92
    },
    {
      "epoch": 0.013652879216060483,
      "grad_norm": 0.09502251446247101,
      "learning_rate": 8.92346577173636e-05,
      "loss": 10.3387,
      "step": 93
    },
    {
      "epoch": 0.01379968436892135,
      "grad_norm": 0.12163512408733368,
      "learning_rate": 8.898371770316111e-05,
      "loss": 10.3343,
      "step": 94
    },
    {
      "epoch": 0.013946489521782214,
      "grad_norm": 0.09053828567266464,
      "learning_rate": 8.873024809138272e-05,
      "loss": 10.3404,
      "step": 95
    },
    {
      "epoch": 0.01409329467464308,
      "grad_norm": 0.11755872517824173,
      "learning_rate": 8.847426532930831e-05,
      "loss": 10.3402,
      "step": 96
    },
    {
      "epoch": 0.014240099827503945,
      "grad_norm": 0.11396406590938568,
      "learning_rate": 8.821578602729242e-05,
      "loss": 10.3441,
      "step": 97
    },
    {
      "epoch": 0.014386904980364811,
      "grad_norm": 0.12473347783088684,
      "learning_rate": 8.795482695768658e-05,
      "loss": 10.3407,
      "step": 98
    },
    {
      "epoch": 0.014533710133225677,
      "grad_norm": 0.14885030686855316,
      "learning_rate": 8.769140505375085e-05,
      "loss": 10.3442,
      "step": 99
    },
    {
      "epoch": 0.014680515286086541,
      "grad_norm": 0.30355480313301086,
      "learning_rate": 8.742553740855506e-05,
      "loss": 10.3534,
      "step": 100
    },
    {
      "epoch": 0.014680515286086541,
      "eval_loss": 10.34232234954834,
      "eval_runtime": 40.9959,
      "eval_samples_per_second": 279.857,
      "eval_steps_per_second": 69.983,
      "step": 100
    },
    {
      "epoch": 0.014827320438947408,
      "grad_norm": 0.21847106516361237,
      "learning_rate": 8.715724127386972e-05,
      "loss": 10.3458,
      "step": 101
    },
    {
      "epoch": 0.014974125591808272,
      "grad_norm": 0.10899436473846436,
      "learning_rate": 8.688653405904652e-05,
      "loss": 10.3467,
      "step": 102
    },
    {
      "epoch": 0.015120930744669138,
      "grad_norm": 0.11821383237838745,
      "learning_rate": 8.661343332988869e-05,
      "loss": 10.3513,
      "step": 103
    },
    {
      "epoch": 0.015267735897530003,
      "grad_norm": 0.0735338032245636,
      "learning_rate": 8.633795680751116e-05,
      "loss": 10.3427,
      "step": 104
    },
    {
      "epoch": 0.015414541050390869,
      "grad_norm": 0.09628330916166306,
      "learning_rate": 8.606012236719073e-05,
      "loss": 10.3445,
      "step": 105
    },
    {
      "epoch": 0.015561346203251733,
      "grad_norm": 0.07052254676818848,
      "learning_rate": 8.577994803720606e-05,
      "loss": 10.3406,
      "step": 106
    },
    {
      "epoch": 0.0157081513561126,
      "grad_norm": 0.09761352837085724,
      "learning_rate": 8.549745199766792e-05,
      "loss": 10.3424,
      "step": 107
    },
    {
      "epoch": 0.015854956508973464,
      "grad_norm": 0.08755065500736237,
      "learning_rate": 8.521265257933948e-05,
      "loss": 10.3399,
      "step": 108
    },
    {
      "epoch": 0.01600176166183433,
      "grad_norm": 0.08864150941371918,
      "learning_rate": 8.492556826244687e-05,
      "loss": 10.3416,
      "step": 109
    },
    {
      "epoch": 0.016148566814695196,
      "grad_norm": 0.07106677442789078,
      "learning_rate": 8.463621767547998e-05,
      "loss": 10.3408,
      "step": 110
    },
    {
      "epoch": 0.016295371967556063,
      "grad_norm": 0.08989017456769943,
      "learning_rate": 8.434461959398376e-05,
      "loss": 10.3448,
      "step": 111
    },
    {
      "epoch": 0.016442177120416925,
      "grad_norm": 0.06086868792772293,
      "learning_rate": 8.405079293933986e-05,
      "loss": 10.3404,
      "step": 112
    },
    {
      "epoch": 0.01658898227327779,
      "grad_norm": 0.06935160607099533,
      "learning_rate": 8.375475677753881e-05,
      "loss": 10.3394,
      "step": 113
    },
    {
      "epoch": 0.016735787426138658,
      "grad_norm": 0.08700171113014221,
      "learning_rate": 8.345653031794292e-05,
      "loss": 10.3408,
      "step": 114
    },
    {
      "epoch": 0.016882592578999524,
      "grad_norm": 0.07914917916059494,
      "learning_rate": 8.315613291203976e-05,
      "loss": 10.3415,
      "step": 115
    },
    {
      "epoch": 0.01702939773186039,
      "grad_norm": 0.0641724243760109,
      "learning_rate": 8.285358405218655e-05,
      "loss": 10.3399,
      "step": 116
    },
    {
      "epoch": 0.017176202884721253,
      "grad_norm": 0.061687447130680084,
      "learning_rate": 8.25489033703452e-05,
      "loss": 10.3452,
      "step": 117
    },
    {
      "epoch": 0.01732300803758212,
      "grad_norm": 0.08324383944272995,
      "learning_rate": 8.224211063680853e-05,
      "loss": 10.3374,
      "step": 118
    },
    {
      "epoch": 0.017469813190442985,
      "grad_norm": 0.07736765593290329,
      "learning_rate": 8.19332257589174e-05,
      "loss": 10.3373,
      "step": 119
    },
    {
      "epoch": 0.01761661834330385,
      "grad_norm": 0.08318845182657242,
      "learning_rate": 8.162226877976887e-05,
      "loss": 10.3377,
      "step": 120
    },
    {
      "epoch": 0.017763423496164714,
      "grad_norm": 0.05588462948799133,
      "learning_rate": 8.130925987691569e-05,
      "loss": 10.3347,
      "step": 121
    },
    {
      "epoch": 0.01791022864902558,
      "grad_norm": 0.07511571049690247,
      "learning_rate": 8.099421936105702e-05,
      "loss": 10.3428,
      "step": 122
    },
    {
      "epoch": 0.018057033801886446,
      "grad_norm": 0.07930921018123627,
      "learning_rate": 8.067716767472045e-05,
      "loss": 10.3353,
      "step": 123
    },
    {
      "epoch": 0.018203838954747312,
      "grad_norm": 0.07638075947761536,
      "learning_rate": 8.035812539093557e-05,
      "loss": 10.3366,
      "step": 124
    },
    {
      "epoch": 0.01835064410760818,
      "grad_norm": 0.09952006489038467,
      "learning_rate": 8.003711321189895e-05,
      "loss": 10.3379,
      "step": 125
    },
    {
      "epoch": 0.01849744926046904,
      "grad_norm": 0.08625619858503342,
      "learning_rate": 7.971415196763088e-05,
      "loss": 10.3391,
      "step": 126
    },
    {
      "epoch": 0.018644254413329907,
      "grad_norm": 0.07221750169992447,
      "learning_rate": 7.938926261462366e-05,
      "loss": 10.339,
      "step": 127
    },
    {
      "epoch": 0.018791059566190774,
      "grad_norm": 0.08416017144918442,
      "learning_rate": 7.906246623448183e-05,
      "loss": 10.341,
      "step": 128
    },
    {
      "epoch": 0.01893786471905164,
      "grad_norm": 0.0774596631526947,
      "learning_rate": 7.873378403255419e-05,
      "loss": 10.3398,
      "step": 129
    },
    {
      "epoch": 0.019084669871912503,
      "grad_norm": 0.07498623430728912,
      "learning_rate": 7.840323733655778e-05,
      "loss": 10.336,
      "step": 130
    },
    {
      "epoch": 0.01923147502477337,
      "grad_norm": 0.07914584130048752,
      "learning_rate": 7.807084759519405e-05,
      "loss": 10.3362,
      "step": 131
    },
    {
      "epoch": 0.019378280177634235,
      "grad_norm": 0.08211899548768997,
      "learning_rate": 7.773663637675694e-05,
      "loss": 10.3364,
      "step": 132
    },
    {
      "epoch": 0.0195250853304951,
      "grad_norm": 0.07826335728168488,
      "learning_rate": 7.740062536773352e-05,
      "loss": 10.3351,
      "step": 133
    },
    {
      "epoch": 0.019671890483355967,
      "grad_norm": 0.07519382983446121,
      "learning_rate": 7.706283637139658e-05,
      "loss": 10.3374,
      "step": 134
    },
    {
      "epoch": 0.01981869563621683,
      "grad_norm": 0.08778757601976395,
      "learning_rate": 7.672329130639005e-05,
      "loss": 10.3368,
      "step": 135
    },
    {
      "epoch": 0.019965500789077696,
      "grad_norm": 0.08082710951566696,
      "learning_rate": 7.638201220530665e-05,
      "loss": 10.3305,
      "step": 136
    },
    {
      "epoch": 0.020112305941938562,
      "grad_norm": 0.09035288542509079,
      "learning_rate": 7.603902121325813e-05,
      "loss": 10.3329,
      "step": 137
    },
    {
      "epoch": 0.02025911109479943,
      "grad_norm": 0.09213805943727493,
      "learning_rate": 7.569434058643844e-05,
      "loss": 10.3346,
      "step": 138
    },
    {
      "epoch": 0.02040591624766029,
      "grad_norm": 0.08065856993198395,
      "learning_rate": 7.534799269067953e-05,
      "loss": 10.3371,
      "step": 139
    },
    {
      "epoch": 0.020552721400521157,
      "grad_norm": 0.07647357136011124,
      "learning_rate": 7.500000000000001e-05,
      "loss": 10.336,
      "step": 140
    },
    {
      "epoch": 0.020699526553382024,
      "grad_norm": 0.09344423562288284,
      "learning_rate": 7.465038509514688e-05,
      "loss": 10.3362,
      "step": 141
    },
    {
      "epoch": 0.02084633170624289,
      "grad_norm": 0.0886511355638504,
      "learning_rate": 7.42991706621303e-05,
      "loss": 10.3326,
      "step": 142
    },
    {
      "epoch": 0.020993136859103756,
      "grad_norm": 0.1064351350069046,
      "learning_rate": 7.394637949075154e-05,
      "loss": 10.334,
      "step": 143
    },
    {
      "epoch": 0.02113994201196462,
      "grad_norm": 0.09669923037290573,
      "learning_rate": 7.35920344731241e-05,
      "loss": 10.328,
      "step": 144
    },
    {
      "epoch": 0.021286747164825485,
      "grad_norm": 0.08189446479082108,
      "learning_rate": 7.323615860218843e-05,
      "loss": 10.3302,
      "step": 145
    },
    {
      "epoch": 0.02143355231768635,
      "grad_norm": 0.09737958014011383,
      "learning_rate": 7.287877497021978e-05,
      "loss": 10.3316,
      "step": 146
    },
    {
      "epoch": 0.021580357470547217,
      "grad_norm": 0.09141288697719574,
      "learning_rate": 7.251990676732984e-05,
      "loss": 10.3307,
      "step": 147
    },
    {
      "epoch": 0.021727162623408083,
      "grad_norm": 0.08963339030742645,
      "learning_rate": 7.215957727996207e-05,
      "loss": 10.3276,
      "step": 148
    },
    {
      "epoch": 0.021873967776268946,
      "grad_norm": 0.12152159214019775,
      "learning_rate": 7.179780988938051e-05,
      "loss": 10.332,
      "step": 149
    },
    {
      "epoch": 0.022020772929129812,
      "grad_norm": 0.3284924626350403,
      "learning_rate": 7.143462807015271e-05,
      "loss": 10.3525,
      "step": 150
    },
    {
      "epoch": 0.022020772929129812,
      "eval_loss": 10.33497428894043,
      "eval_runtime": 40.8105,
      "eval_samples_per_second": 281.129,
      "eval_steps_per_second": 70.301,
      "step": 150
    },
    {
      "epoch": 0.02216757808199068,
      "grad_norm": 0.17632122337818146,
      "learning_rate": 7.107005538862646e-05,
      "loss": 10.3456,
      "step": 151
    },
    {
      "epoch": 0.022314383234851545,
      "grad_norm": 0.09494099766016006,
      "learning_rate": 7.07041155014006e-05,
      "loss": 10.3476,
      "step": 152
    },
    {
      "epoch": 0.022461188387712407,
      "grad_norm": 0.1040123701095581,
      "learning_rate": 7.033683215379002e-05,
      "loss": 10.3457,
      "step": 153
    },
    {
      "epoch": 0.022607993540573273,
      "grad_norm": 0.10032442212104797,
      "learning_rate": 6.996822917828477e-05,
      "loss": 10.3345,
      "step": 154
    },
    {
      "epoch": 0.02275479869343414,
      "grad_norm": 0.05919092521071434,
      "learning_rate": 6.959833049300377e-05,
      "loss": 10.3359,
      "step": 155
    },
    {
      "epoch": 0.022901603846295006,
      "grad_norm": 0.09216248989105225,
      "learning_rate": 6.922716010014255e-05,
      "loss": 10.3382,
      "step": 156
    },
    {
      "epoch": 0.023048408999155872,
      "grad_norm": 0.05931238457560539,
      "learning_rate": 6.885474208441603e-05,
      "loss": 10.3347,
      "step": 157
    },
    {
      "epoch": 0.023195214152016735,
      "grad_norm": 0.06898997724056244,
      "learning_rate": 6.848110061149556e-05,
      "loss": 10.335,
      "step": 158
    },
    {
      "epoch": 0.0233420193048776,
      "grad_norm": 0.07491658627986908,
      "learning_rate": 6.810625992644085e-05,
      "loss": 10.3296,
      "step": 159
    },
    {
      "epoch": 0.023488824457738467,
      "grad_norm": 0.06421888619661331,
      "learning_rate": 6.773024435212678e-05,
      "loss": 10.3307,
      "step": 160
    },
    {
      "epoch": 0.023635629610599333,
      "grad_norm": 0.06266214698553085,
      "learning_rate": 6.735307828766515e-05,
      "loss": 10.3375,
      "step": 161
    },
    {
      "epoch": 0.023782434763460196,
      "grad_norm": 0.0847383365035057,
      "learning_rate": 6.697478620682137e-05,
      "loss": 10.3355,
      "step": 162
    },
    {
      "epoch": 0.023929239916321062,
      "grad_norm": 0.056205783039331436,
      "learning_rate": 6.659539265642643e-05,
      "loss": 10.3348,
      "step": 163
    },
    {
      "epoch": 0.02407604506918193,
      "grad_norm": 0.06446053087711334,
      "learning_rate": 6.621492225478414e-05,
      "loss": 10.3265,
      "step": 164
    },
    {
      "epoch": 0.024222850222042795,
      "grad_norm": 0.06294980645179749,
      "learning_rate": 6.583339969007363e-05,
      "loss": 10.3351,
      "step": 165
    },
    {
      "epoch": 0.02436965537490366,
      "grad_norm": 0.07819314301013947,
      "learning_rate": 6.545084971874738e-05,
      "loss": 10.3297,
      "step": 166
    },
    {
      "epoch": 0.024516460527764523,
      "grad_norm": 0.062054529786109924,
      "learning_rate": 6.506729716392481e-05,
      "loss": 10.3312,
      "step": 167
    },
    {
      "epoch": 0.02466326568062539,
      "grad_norm": 0.07110702246427536,
      "learning_rate": 6.468276691378155e-05,
      "loss": 10.3359,
      "step": 168
    },
    {
      "epoch": 0.024810070833486256,
      "grad_norm": 0.07958277314901352,
      "learning_rate": 6.429728391993446e-05,
      "loss": 10.3323,
      "step": 169
    },
    {
      "epoch": 0.024956875986347122,
      "grad_norm": 0.10255732387304306,
      "learning_rate": 6.391087319582264e-05,
      "loss": 10.3349,
      "step": 170
    },
    {
      "epoch": 0.025103681139207985,
      "grad_norm": 0.06388706713914871,
      "learning_rate": 6.35235598150842e-05,
      "loss": 10.3284,
      "step": 171
    },
    {
      "epoch": 0.02525048629206885,
      "grad_norm": 0.07001566141843796,
      "learning_rate": 6.313536890992935e-05,
      "loss": 10.3324,
      "step": 172
    },
    {
      "epoch": 0.025397291444929717,
      "grad_norm": 0.06101684272289276,
      "learning_rate": 6.274632566950967e-05,
      "loss": 10.33,
      "step": 173
    },
    {
      "epoch": 0.025544096597790583,
      "grad_norm": 0.07129354029893875,
      "learning_rate": 6.235645533828349e-05,
      "loss": 10.3331,
      "step": 174
    },
    {
      "epoch": 0.02569090175065145,
      "grad_norm": 0.06856873631477356,
      "learning_rate": 6.19657832143779e-05,
      "loss": 10.3294,
      "step": 175
    },
    {
      "epoch": 0.025837706903512312,
      "grad_norm": 0.06160217523574829,
      "learning_rate": 6.157433464794716e-05,
      "loss": 10.3284,
      "step": 176
    },
    {
      "epoch": 0.025984512056373178,
      "grad_norm": 0.06369267404079437,
      "learning_rate": 6.118213503952779e-05,
      "loss": 10.3318,
      "step": 177
    },
    {
      "epoch": 0.026131317209234044,
      "grad_norm": 0.059929605573415756,
      "learning_rate": 6.078920983839031e-05,
      "loss": 10.3271,
      "step": 178
    },
    {
      "epoch": 0.02627812236209491,
      "grad_norm": 0.07750251144170761,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 10.3276,
      "step": 179
    },
    {
      "epoch": 0.026424927514955773,
      "grad_norm": 0.07519747316837311,
      "learning_rate": 6.0001284688802226e-05,
      "loss": 10.3329,
      "step": 180
    },
    {
      "epoch": 0.02657173266781664,
      "grad_norm": 0.10204313695430756,
      "learning_rate": 5.960633586768543e-05,
      "loss": 10.3308,
      "step": 181
    },
    {
      "epoch": 0.026718537820677506,
      "grad_norm": 0.08898138999938965,
      "learning_rate": 5.921076370520058e-05,
      "loss": 10.3307,
      "step": 182
    },
    {
      "epoch": 0.026865342973538372,
      "grad_norm": 0.08795037120580673,
      "learning_rate": 5.8814593869458455e-05,
      "loss": 10.3247,
      "step": 183
    },
    {
      "epoch": 0.027012148126399238,
      "grad_norm": 0.06972832977771759,
      "learning_rate": 5.841785206735192e-05,
      "loss": 10.3352,
      "step": 184
    },
    {
      "epoch": 0.0271589532792601,
      "grad_norm": 0.07827477902173996,
      "learning_rate": 5.8020564042888015e-05,
      "loss": 10.33,
      "step": 185
    },
    {
      "epoch": 0.027305758432120967,
      "grad_norm": 0.08251015096902847,
      "learning_rate": 5.762275557551727e-05,
      "loss": 10.3295,
      "step": 186
    },
    {
      "epoch": 0.027452563584981833,
      "grad_norm": 0.07507099956274033,
      "learning_rate": 5.7224452478461064e-05,
      "loss": 10.3332,
      "step": 187
    },
    {
      "epoch": 0.0275993687378427,
      "grad_norm": 0.0892663300037384,
      "learning_rate": 5.682568059703659e-05,
      "loss": 10.3298,
      "step": 188
    },
    {
      "epoch": 0.027746173890703562,
      "grad_norm": 0.07587073743343353,
      "learning_rate": 5.642646580697973e-05,
      "loss": 10.3349,
      "step": 189
    },
    {
      "epoch": 0.027892979043564428,
      "grad_norm": 0.07973253726959229,
      "learning_rate": 5.602683401276615e-05,
      "loss": 10.3334,
      "step": 190
    },
    {
      "epoch": 0.028039784196425294,
      "grad_norm": 0.09443668276071548,
      "learning_rate": 5.562681114593028e-05,
      "loss": 10.3226,
      "step": 191
    },
    {
      "epoch": 0.02818658934928616,
      "grad_norm": 0.08195987343788147,
      "learning_rate": 5.522642316338268e-05,
      "loss": 10.3305,
      "step": 192
    },
    {
      "epoch": 0.028333394502147027,
      "grad_norm": 0.09204565733671188,
      "learning_rate": 5.482569604572576e-05,
      "loss": 10.3267,
      "step": 193
    },
    {
      "epoch": 0.02848019965500789,
      "grad_norm": 0.09184598922729492,
      "learning_rate": 5.442465579556793e-05,
      "loss": 10.3302,
      "step": 194
    },
    {
      "epoch": 0.028627004807868756,
      "grad_norm": 0.10049549490213394,
      "learning_rate": 5.402332843583631e-05,
      "loss": 10.3289,
      "step": 195
    },
    {
      "epoch": 0.028773809960729622,
      "grad_norm": 0.09506766498088837,
      "learning_rate": 5.3621740008088126e-05,
      "loss": 10.3329,
      "step": 196
    },
    {
      "epoch": 0.028920615113590488,
      "grad_norm": 0.1400189846754074,
      "learning_rate": 5.321991657082097e-05,
      "loss": 10.3292,
      "step": 197
    },
    {
      "epoch": 0.029067420266451354,
      "grad_norm": 0.13115336000919342,
      "learning_rate": 5.281788419778187e-05,
      "loss": 10.3279,
      "step": 198
    },
    {
      "epoch": 0.029214225419312217,
      "grad_norm": 0.12796978652477264,
      "learning_rate": 5.2415668976275355e-05,
      "loss": 10.3205,
      "step": 199
    },
    {
      "epoch": 0.029361030572173083,
      "grad_norm": 0.23809519410133362,
      "learning_rate": 5.201329700547076e-05,
      "loss": 10.346,
      "step": 200
    },
    {
      "epoch": 0.029361030572173083,
      "eval_loss": 10.330036163330078,
      "eval_runtime": 40.6183,
      "eval_samples_per_second": 282.459,
      "eval_steps_per_second": 70.633,
      "step": 200
    },
    {
      "epoch": 0.02950783572503395,
      "grad_norm": 0.18717199563980103,
      "learning_rate": 5.161079439470866e-05,
      "loss": 10.3518,
      "step": 201
    },
    {
      "epoch": 0.029654640877894815,
      "grad_norm": 0.09647821635007858,
      "learning_rate": 5.1208187261806615e-05,
      "loss": 10.3394,
      "step": 202
    },
    {
      "epoch": 0.029801446030755678,
      "grad_norm": 0.06907595694065094,
      "learning_rate": 5.080550173136457e-05,
      "loss": 10.3366,
      "step": 203
    },
    {
      "epoch": 0.029948251183616544,
      "grad_norm": 0.1104382574558258,
      "learning_rate": 5.0402763933069496e-05,
      "loss": 10.3378,
      "step": 204
    },
    {
      "epoch": 0.03009505633647741,
      "grad_norm": 0.07312840223312378,
      "learning_rate": 5e-05,
      "loss": 10.3343,
      "step": 205
    },
    {
      "epoch": 0.030241861489338277,
      "grad_norm": 0.06366623193025589,
      "learning_rate": 4.9597236066930516e-05,
      "loss": 10.3377,
      "step": 206
    },
    {
      "epoch": 0.030388666642199143,
      "grad_norm": 0.06475561857223511,
      "learning_rate": 4.919449826863544e-05,
      "loss": 10.3265,
      "step": 207
    },
    {
      "epoch": 0.030535471795060005,
      "grad_norm": 0.06041239947080612,
      "learning_rate": 4.87918127381934e-05,
      "loss": 10.3348,
      "step": 208
    },
    {
      "epoch": 0.03068227694792087,
      "grad_norm": 0.07550211995840073,
      "learning_rate": 4.8389205605291365e-05,
      "loss": 10.333,
      "step": 209
    },
    {
      "epoch": 0.030829082100781738,
      "grad_norm": 0.06893938034772873,
      "learning_rate": 4.798670299452926e-05,
      "loss": 10.3299,
      "step": 210
    },
    {
      "epoch": 0.030975887253642604,
      "grad_norm": 0.08085527271032333,
      "learning_rate": 4.758433102372466e-05,
      "loss": 10.3272,
      "step": 211
    },
    {
      "epoch": 0.031122692406503467,
      "grad_norm": 0.06497251242399216,
      "learning_rate": 4.7182115802218126e-05,
      "loss": 10.3325,
      "step": 212
    },
    {
      "epoch": 0.03126949755936433,
      "grad_norm": 0.07956556230783463,
      "learning_rate": 4.678008342917903e-05,
      "loss": 10.3285,
      "step": 213
    },
    {
      "epoch": 0.0314163027122252,
      "grad_norm": 0.05742551013827324,
      "learning_rate": 4.6378259991911886e-05,
      "loss": 10.3318,
      "step": 214
    },
    {
      "epoch": 0.031563107865086065,
      "grad_norm": 0.06607367098331451,
      "learning_rate": 4.597667156416371e-05,
      "loss": 10.3273,
      "step": 215
    },
    {
      "epoch": 0.03170991301794693,
      "grad_norm": 0.06792084127664566,
      "learning_rate": 4.5575344204432084e-05,
      "loss": 10.3287,
      "step": 216
    },
    {
      "epoch": 0.0318567181708078,
      "grad_norm": 0.07235945761203766,
      "learning_rate": 4.5174303954274244e-05,
      "loss": 10.3319,
      "step": 217
    },
    {
      "epoch": 0.03200352332366866,
      "grad_norm": 0.07845742255449295,
      "learning_rate": 4.477357683661734e-05,
      "loss": 10.3228,
      "step": 218
    },
    {
      "epoch": 0.03215032847652952,
      "grad_norm": 0.07698619365692139,
      "learning_rate": 4.437318885406973e-05,
      "loss": 10.3321,
      "step": 219
    },
    {
      "epoch": 0.03229713362939039,
      "grad_norm": 0.0897485613822937,
      "learning_rate": 4.397316598723385e-05,
      "loss": 10.3334,
      "step": 220
    },
    {
      "epoch": 0.032443938782251255,
      "grad_norm": 0.0717536062002182,
      "learning_rate": 4.3573534193020274e-05,
      "loss": 10.3275,
      "step": 221
    },
    {
      "epoch": 0.032590743935112125,
      "grad_norm": 0.06742411851882935,
      "learning_rate": 4.317431940296343e-05,
      "loss": 10.3212,
      "step": 222
    },
    {
      "epoch": 0.03273754908797299,
      "grad_norm": 0.0724957212805748,
      "learning_rate": 4.277554752153895e-05,
      "loss": 10.3284,
      "step": 223
    },
    {
      "epoch": 0.03288435424083385,
      "grad_norm": 0.06968709826469421,
      "learning_rate": 4.237724442448273e-05,
      "loss": 10.3235,
      "step": 224
    },
    {
      "epoch": 0.03303115939369472,
      "grad_norm": 0.07918791472911835,
      "learning_rate": 4.197943595711198e-05,
      "loss": 10.3258,
      "step": 225
    },
    {
      "epoch": 0.03317796454655558,
      "grad_norm": 0.08635478466749191,
      "learning_rate": 4.1582147932648074e-05,
      "loss": 10.3227,
      "step": 226
    },
    {
      "epoch": 0.03332476969941645,
      "grad_norm": 0.07083820551633835,
      "learning_rate": 4.118540613054156e-05,
      "loss": 10.326,
      "step": 227
    },
    {
      "epoch": 0.033471574852277315,
      "grad_norm": 0.07480268180370331,
      "learning_rate": 4.078923629479943e-05,
      "loss": 10.3249,
      "step": 228
    },
    {
      "epoch": 0.03361838000513818,
      "grad_norm": 0.07721208781003952,
      "learning_rate": 4.039366413231458e-05,
      "loss": 10.3278,
      "step": 229
    },
    {
      "epoch": 0.03376518515799905,
      "grad_norm": 0.08692963421344757,
      "learning_rate": 3.9998715311197785e-05,
      "loss": 10.3241,
      "step": 230
    },
    {
      "epoch": 0.03391199031085991,
      "grad_norm": 0.07471178472042084,
      "learning_rate": 3.960441545911204e-05,
      "loss": 10.3234,
      "step": 231
    },
    {
      "epoch": 0.03405879546372078,
      "grad_norm": 0.07830685377120972,
      "learning_rate": 3.92107901616097e-05,
      "loss": 10.3258,
      "step": 232
    },
    {
      "epoch": 0.03420560061658164,
      "grad_norm": 0.0772634893655777,
      "learning_rate": 3.8817864960472236e-05,
      "loss": 10.3228,
      "step": 233
    },
    {
      "epoch": 0.034352405769442505,
      "grad_norm": 0.07172077149152756,
      "learning_rate": 3.842566535205286e-05,
      "loss": 10.3239,
      "step": 234
    },
    {
      "epoch": 0.034499210922303375,
      "grad_norm": 0.08126288652420044,
      "learning_rate": 3.803421678562213e-05,
      "loss": 10.3259,
      "step": 235
    },
    {
      "epoch": 0.03464601607516424,
      "grad_norm": 0.0885980948805809,
      "learning_rate": 3.764354466171652e-05,
      "loss": 10.3305,
      "step": 236
    },
    {
      "epoch": 0.0347928212280251,
      "grad_norm": 0.08453716337680817,
      "learning_rate": 3.725367433049033e-05,
      "loss": 10.3247,
      "step": 237
    },
    {
      "epoch": 0.03493962638088597,
      "grad_norm": 0.08464954048395157,
      "learning_rate": 3.6864631090070655e-05,
      "loss": 10.325,
      "step": 238
    },
    {
      "epoch": 0.03508643153374683,
      "grad_norm": 0.10857713967561722,
      "learning_rate": 3.6476440184915815e-05,
      "loss": 10.3297,
      "step": 239
    },
    {
      "epoch": 0.0352332366866077,
      "grad_norm": 0.08051791787147522,
      "learning_rate": 3.608912680417737e-05,
      "loss": 10.329,
      "step": 240
    },
    {
      "epoch": 0.035380041839468565,
      "grad_norm": 0.08642332255840302,
      "learning_rate": 3.570271608006555e-05,
      "loss": 10.3282,
      "step": 241
    },
    {
      "epoch": 0.03552684699232943,
      "grad_norm": 0.08665293455123901,
      "learning_rate": 3.531723308621847e-05,
      "loss": 10.3278,
      "step": 242
    },
    {
      "epoch": 0.0356736521451903,
      "grad_norm": 0.0906917154788971,
      "learning_rate": 3.493270283607522e-05,
      "loss": 10.3274,
      "step": 243
    },
    {
      "epoch": 0.03582045729805116,
      "grad_norm": 0.0852770134806633,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 10.3328,
      "step": 244
    },
    {
      "epoch": 0.03596726245091203,
      "grad_norm": 0.08750558644533157,
      "learning_rate": 3.4166600309926387e-05,
      "loss": 10.3286,
      "step": 245
    },
    {
      "epoch": 0.03611406760377289,
      "grad_norm": 0.10538952052593231,
      "learning_rate": 3.3785077745215873e-05,
      "loss": 10.3219,
      "step": 246
    },
    {
      "epoch": 0.036260872756633755,
      "grad_norm": 0.10317540913820267,
      "learning_rate": 3.340460734357359e-05,
      "loss": 10.3329,
      "step": 247
    },
    {
      "epoch": 0.036407677909494625,
      "grad_norm": 0.11254429072141647,
      "learning_rate": 3.3025213793178646e-05,
      "loss": 10.3298,
      "step": 248
    },
    {
      "epoch": 0.03655448306235549,
      "grad_norm": 0.17804968357086182,
      "learning_rate": 3.264692171233485e-05,
      "loss": 10.3316,
      "step": 249
    },
    {
      "epoch": 0.03670128821521636,
      "grad_norm": 0.21033170819282532,
      "learning_rate": 3.226975564787322e-05,
      "loss": 10.3361,
      "step": 250
    },
    {
      "epoch": 0.03670128821521636,
      "eval_loss": 10.327363967895508,
      "eval_runtime": 40.4885,
      "eval_samples_per_second": 283.364,
      "eval_steps_per_second": 70.86,
      "step": 250
    },
    {
      "epoch": 0.03684809336807722,
      "grad_norm": 0.16220684349536896,
      "learning_rate": 3.189374007355917e-05,
      "loss": 10.3386,
      "step": 251
    },
    {
      "epoch": 0.03699489852093808,
      "grad_norm": 0.09081603586673737,
      "learning_rate": 3.151889938850445e-05,
      "loss": 10.3442,
      "step": 252
    },
    {
      "epoch": 0.03714170367379895,
      "grad_norm": 0.08561962842941284,
      "learning_rate": 3.114525791558398e-05,
      "loss": 10.34,
      "step": 253
    },
    {
      "epoch": 0.037288508826659815,
      "grad_norm": 0.08950639516115189,
      "learning_rate": 3.0772839899857464e-05,
      "loss": 10.3282,
      "step": 254
    },
    {
      "epoch": 0.037435313979520685,
      "grad_norm": 0.09625831991434097,
      "learning_rate": 3.0401669506996256e-05,
      "loss": 10.3354,
      "step": 255
    },
    {
      "epoch": 0.03758211913238155,
      "grad_norm": 0.09835858643054962,
      "learning_rate": 3.003177082171523e-05,
      "loss": 10.3343,
      "step": 256
    },
    {
      "epoch": 0.03772892428524241,
      "grad_norm": 0.06951465457677841,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 10.3244,
      "step": 257
    },
    {
      "epoch": 0.03787572943810328,
      "grad_norm": 0.06446226686239243,
      "learning_rate": 2.9295884498599414e-05,
      "loss": 10.3361,
      "step": 258
    },
    {
      "epoch": 0.03802253459096414,
      "grad_norm": 0.06822808086872101,
      "learning_rate": 2.8929944611373554e-05,
      "loss": 10.3242,
      "step": 259
    },
    {
      "epoch": 0.038169339743825005,
      "grad_norm": 0.06012842431664467,
      "learning_rate": 2.8565371929847284e-05,
      "loss": 10.3321,
      "step": 260
    },
    {
      "epoch": 0.038316144896685875,
      "grad_norm": 0.06354397535324097,
      "learning_rate": 2.8202190110619493e-05,
      "loss": 10.3311,
      "step": 261
    },
    {
      "epoch": 0.03846295004954674,
      "grad_norm": 0.06928583979606628,
      "learning_rate": 2.784042272003794e-05,
      "loss": 10.3251,
      "step": 262
    },
    {
      "epoch": 0.03860975520240761,
      "grad_norm": 0.07280782610177994,
      "learning_rate": 2.7480093232670158e-05,
      "loss": 10.3254,
      "step": 263
    },
    {
      "epoch": 0.03875656035526847,
      "grad_norm": 0.06568915396928787,
      "learning_rate": 2.712122502978024e-05,
      "loss": 10.3338,
      "step": 264
    },
    {
      "epoch": 0.03890336550812933,
      "grad_norm": 0.07334387302398682,
      "learning_rate": 2.6763841397811573e-05,
      "loss": 10.3209,
      "step": 265
    },
    {
      "epoch": 0.0390501706609902,
      "grad_norm": 0.07109591364860535,
      "learning_rate": 2.64079655268759e-05,
      "loss": 10.3289,
      "step": 266
    },
    {
      "epoch": 0.039196975813851065,
      "grad_norm": 0.06435167044401169,
      "learning_rate": 2.605362050924848e-05,
      "loss": 10.3227,
      "step": 267
    },
    {
      "epoch": 0.039343780966711935,
      "grad_norm": 0.07095325738191605,
      "learning_rate": 2.57008293378697e-05,
      "loss": 10.3224,
      "step": 268
    },
    {
      "epoch": 0.0394905861195728,
      "grad_norm": 0.06839649379253387,
      "learning_rate": 2.534961490485313e-05,
      "loss": 10.3298,
      "step": 269
    },
    {
      "epoch": 0.03963739127243366,
      "grad_norm": 0.0642453208565712,
      "learning_rate": 2.500000000000001e-05,
      "loss": 10.3336,
      "step": 270
    },
    {
      "epoch": 0.03978419642529453,
      "grad_norm": 0.059606146067380905,
      "learning_rate": 2.4652007309320498e-05,
      "loss": 10.3326,
      "step": 271
    },
    {
      "epoch": 0.03993100157815539,
      "grad_norm": 0.06261233240365982,
      "learning_rate": 2.430565941356157e-05,
      "loss": 10.3299,
      "step": 272
    },
    {
      "epoch": 0.04007780673101626,
      "grad_norm": 0.06576954573392868,
      "learning_rate": 2.3960978786741877e-05,
      "loss": 10.3281,
      "step": 273
    },
    {
      "epoch": 0.040224611883877125,
      "grad_norm": 0.07377596199512482,
      "learning_rate": 2.361798779469336e-05,
      "loss": 10.3179,
      "step": 274
    },
    {
      "epoch": 0.04037141703673799,
      "grad_norm": 0.06469722092151642,
      "learning_rate": 2.3276708693609943e-05,
      "loss": 10.3234,
      "step": 275
    },
    {
      "epoch": 0.04051822218959886,
      "grad_norm": 0.08151844143867493,
      "learning_rate": 2.2937163628603435e-05,
      "loss": 10.3272,
      "step": 276
    },
    {
      "epoch": 0.04066502734245972,
      "grad_norm": 0.07202380895614624,
      "learning_rate": 2.259937463226651e-05,
      "loss": 10.3284,
      "step": 277
    },
    {
      "epoch": 0.04081183249532058,
      "grad_norm": 0.08032609522342682,
      "learning_rate": 2.2263363623243054e-05,
      "loss": 10.3283,
      "step": 278
    },
    {
      "epoch": 0.04095863764818145,
      "grad_norm": 0.09396550804376602,
      "learning_rate": 2.192915240480596e-05,
      "loss": 10.327,
      "step": 279
    },
    {
      "epoch": 0.041105442801042315,
      "grad_norm": 0.07072699815034866,
      "learning_rate": 2.1596762663442218e-05,
      "loss": 10.3271,
      "step": 280
    },
    {
      "epoch": 0.041252247953903184,
      "grad_norm": 0.07877470552921295,
      "learning_rate": 2.1266215967445824e-05,
      "loss": 10.3279,
      "step": 281
    },
    {
      "epoch": 0.04139905310676405,
      "grad_norm": 0.07291466742753983,
      "learning_rate": 2.0937533765518187e-05,
      "loss": 10.3244,
      "step": 282
    },
    {
      "epoch": 0.04154585825962491,
      "grad_norm": 0.07320375740528107,
      "learning_rate": 2.061073738537635e-05,
      "loss": 10.3233,
      "step": 283
    },
    {
      "epoch": 0.04169266341248578,
      "grad_norm": 0.07049214094877243,
      "learning_rate": 2.0285848032369137e-05,
      "loss": 10.3192,
      "step": 284
    },
    {
      "epoch": 0.04183946856534664,
      "grad_norm": 0.07985290139913559,
      "learning_rate": 1.996288678810105e-05,
      "loss": 10.3242,
      "step": 285
    },
    {
      "epoch": 0.04198627371820751,
      "grad_norm": 0.0817054882645607,
      "learning_rate": 1.9641874609064443e-05,
      "loss": 10.3251,
      "step": 286
    },
    {
      "epoch": 0.042133078871068375,
      "grad_norm": 0.08590872585773468,
      "learning_rate": 1.932283232527956e-05,
      "loss": 10.3222,
      "step": 287
    },
    {
      "epoch": 0.04227988402392924,
      "grad_norm": 0.0857609361410141,
      "learning_rate": 1.9005780638942982e-05,
      "loss": 10.3265,
      "step": 288
    },
    {
      "epoch": 0.04242668917679011,
      "grad_norm": 0.09034871309995651,
      "learning_rate": 1.8690740123084316e-05,
      "loss": 10.3264,
      "step": 289
    },
    {
      "epoch": 0.04257349432965097,
      "grad_norm": 0.09301265329122543,
      "learning_rate": 1.837773122023114e-05,
      "loss": 10.3256,
      "step": 290
    },
    {
      "epoch": 0.04272029948251184,
      "grad_norm": 0.10398516058921814,
      "learning_rate": 1.8066774241082612e-05,
      "loss": 10.3282,
      "step": 291
    },
    {
      "epoch": 0.0428671046353727,
      "grad_norm": 0.09852512180805206,
      "learning_rate": 1.7757889363191483e-05,
      "loss": 10.3238,
      "step": 292
    },
    {
      "epoch": 0.043013909788233565,
      "grad_norm": 0.09250245988368988,
      "learning_rate": 1.745109662965481e-05,
      "loss": 10.325,
      "step": 293
    },
    {
      "epoch": 0.043160714941094434,
      "grad_norm": 0.08831433951854706,
      "learning_rate": 1.714641594781347e-05,
      "loss": 10.3188,
      "step": 294
    },
    {
      "epoch": 0.0433075200939553,
      "grad_norm": 0.09916289895772934,
      "learning_rate": 1.684386708796025e-05,
      "loss": 10.3225,
      "step": 295
    },
    {
      "epoch": 0.04345432524681617,
      "grad_norm": 0.0858798399567604,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 10.325,
      "step": 296
    },
    {
      "epoch": 0.04360113039967703,
      "grad_norm": 0.10966868698596954,
      "learning_rate": 1.62452432224612e-05,
      "loss": 10.3223,
      "step": 297
    },
    {
      "epoch": 0.04374793555253789,
      "grad_norm": 0.10104182362556458,
      "learning_rate": 1.5949207060660138e-05,
      "loss": 10.3243,
      "step": 298
    },
    {
      "epoch": 0.04389474070539876,
      "grad_norm": 0.1346552073955536,
      "learning_rate": 1.5655380406016235e-05,
      "loss": 10.3324,
      "step": 299
    },
    {
      "epoch": 0.044041545858259624,
      "grad_norm": 0.23662683367729187,
      "learning_rate": 1.536378232452003e-05,
      "loss": 10.3352,
      "step": 300
    },
    {
      "epoch": 0.044041545858259624,
      "eval_loss": 10.326071739196777,
      "eval_runtime": 40.598,
      "eval_samples_per_second": 282.6,
      "eval_steps_per_second": 70.669,
      "step": 300
    },
    {
      "epoch": 0.04418835101112049,
      "grad_norm": 0.15617330372333527,
      "learning_rate": 1.5074431737553157e-05,
      "loss": 10.3324,
      "step": 301
    },
    {
      "epoch": 0.04433515616398136,
      "grad_norm": 0.1019449383020401,
      "learning_rate": 1.4787347420660541e-05,
      "loss": 10.3353,
      "step": 302
    },
    {
      "epoch": 0.04448196131684222,
      "grad_norm": 0.07090485841035843,
      "learning_rate": 1.4502548002332088e-05,
      "loss": 10.3358,
      "step": 303
    },
    {
      "epoch": 0.04462876646970309,
      "grad_norm": 0.08353332430124283,
      "learning_rate": 1.422005196279395e-05,
      "loss": 10.3286,
      "step": 304
    },
    {
      "epoch": 0.04477557162256395,
      "grad_norm": 0.06909141689538956,
      "learning_rate": 1.3939877632809278e-05,
      "loss": 10.3307,
      "step": 305
    },
    {
      "epoch": 0.044922376775424815,
      "grad_norm": 0.07847987115383148,
      "learning_rate": 1.3662043192488849e-05,
      "loss": 10.3325,
      "step": 306
    },
    {
      "epoch": 0.045069181928285684,
      "grad_norm": 0.08882429450750351,
      "learning_rate": 1.338656667011134e-05,
      "loss": 10.3254,
      "step": 307
    },
    {
      "epoch": 0.04521598708114655,
      "grad_norm": 0.094759002327919,
      "learning_rate": 1.3113465940953495e-05,
      "loss": 10.3323,
      "step": 308
    },
    {
      "epoch": 0.04536279223400742,
      "grad_norm": 0.0710742175579071,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 10.328,
      "step": 309
    },
    {
      "epoch": 0.04550959738686828,
      "grad_norm": 0.07474274188280106,
      "learning_rate": 1.257446259144494e-05,
      "loss": 10.3261,
      "step": 310
    },
    {
      "epoch": 0.04565640253972914,
      "grad_norm": 0.092684805393219,
      "learning_rate": 1.2308594946249163e-05,
      "loss": 10.3258,
      "step": 311
    },
    {
      "epoch": 0.04580320769259001,
      "grad_norm": 0.07652658969163895,
      "learning_rate": 1.204517304231343e-05,
      "loss": 10.3218,
      "step": 312
    },
    {
      "epoch": 0.045950012845450874,
      "grad_norm": 0.0737638920545578,
      "learning_rate": 1.178421397270758e-05,
      "loss": 10.334,
      "step": 313
    },
    {
      "epoch": 0.046096817998311744,
      "grad_norm": 0.05947414040565491,
      "learning_rate": 1.1525734670691701e-05,
      "loss": 10.3312,
      "step": 314
    },
    {
      "epoch": 0.04624362315117261,
      "grad_norm": 0.08125612139701843,
      "learning_rate": 1.1269751908617277e-05,
      "loss": 10.3262,
      "step": 315
    },
    {
      "epoch": 0.04639042830403347,
      "grad_norm": 0.0674387738108635,
      "learning_rate": 1.1016282296838887e-05,
      "loss": 10.3267,
      "step": 316
    },
    {
      "epoch": 0.04653723345689434,
      "grad_norm": 0.06924287229776382,
      "learning_rate": 1.0765342282636416e-05,
      "loss": 10.3254,
      "step": 317
    },
    {
      "epoch": 0.0466840386097552,
      "grad_norm": 0.06078397482633591,
      "learning_rate": 1.0516948149147754e-05,
      "loss": 10.3342,
      "step": 318
    },
    {
      "epoch": 0.046830843762616065,
      "grad_norm": 0.06581950187683105,
      "learning_rate": 1.0271116014312293e-05,
      "loss": 10.3266,
      "step": 319
    },
    {
      "epoch": 0.046977648915476934,
      "grad_norm": 0.07352259755134583,
      "learning_rate": 1.0027861829824952e-05,
      "loss": 10.3265,
      "step": 320
    },
    {
      "epoch": 0.0471244540683378,
      "grad_norm": 0.06884633004665375,
      "learning_rate": 9.787201380101157e-06,
      "loss": 10.3221,
      "step": 321
    },
    {
      "epoch": 0.047271259221198667,
      "grad_norm": 0.07347755134105682,
      "learning_rate": 9.549150281252633e-06,
      "loss": 10.3293,
      "step": 322
    },
    {
      "epoch": 0.04741806437405953,
      "grad_norm": 0.07484126836061478,
      "learning_rate": 9.313723980074018e-06,
      "loss": 10.3207,
      "step": 323
    },
    {
      "epoch": 0.04756486952692039,
      "grad_norm": 0.08332342654466629,
      "learning_rate": 9.080937753040646e-06,
      "loss": 10.324,
      "step": 324
    },
    {
      "epoch": 0.04771167467978126,
      "grad_norm": 0.07034022361040115,
      "learning_rate": 8.850806705317183e-06,
      "loss": 10.3264,
      "step": 325
    },
    {
      "epoch": 0.047858479832642124,
      "grad_norm": 0.07176532596349716,
      "learning_rate": 8.623345769777514e-06,
      "loss": 10.325,
      "step": 326
    },
    {
      "epoch": 0.048005284985502994,
      "grad_norm": 0.1011543869972229,
      "learning_rate": 8.398569706035792e-06,
      "loss": 10.308,
      "step": 327
    },
    {
      "epoch": 0.04815209013836386,
      "grad_norm": 0.07753263413906097,
      "learning_rate": 8.176493099488663e-06,
      "loss": 10.3269,
      "step": 328
    },
    {
      "epoch": 0.04829889529122472,
      "grad_norm": 0.06734514981508255,
      "learning_rate": 7.957130360368898e-06,
      "loss": 10.3221,
      "step": 329
    },
    {
      "epoch": 0.04844570044408559,
      "grad_norm": 0.08725865185260773,
      "learning_rate": 7.740495722810271e-06,
      "loss": 10.3183,
      "step": 330
    },
    {
      "epoch": 0.04859250559694645,
      "grad_norm": 0.07323329895734787,
      "learning_rate": 7.526603243923957e-06,
      "loss": 10.3186,
      "step": 331
    },
    {
      "epoch": 0.04873931074980732,
      "grad_norm": 0.0848127231001854,
      "learning_rate": 7.315466802886401e-06,
      "loss": 10.3222,
      "step": 332
    },
    {
      "epoch": 0.048886115902668184,
      "grad_norm": 0.06692821532487869,
      "learning_rate": 7.107100100038671e-06,
      "loss": 10.3143,
      "step": 333
    },
    {
      "epoch": 0.04903292105552905,
      "grad_norm": 0.08171255886554718,
      "learning_rate": 6.901516655997536e-06,
      "loss": 10.3238,
      "step": 334
    },
    {
      "epoch": 0.049179726208389916,
      "grad_norm": 0.08207869529724121,
      "learning_rate": 6.698729810778065e-06,
      "loss": 10.3232,
      "step": 335
    },
    {
      "epoch": 0.04932653136125078,
      "grad_norm": 0.09229540079832077,
      "learning_rate": 6.498752722928042e-06,
      "loss": 10.3215,
      "step": 336
    },
    {
      "epoch": 0.04947333651411165,
      "grad_norm": 0.0909721627831459,
      "learning_rate": 6.301598368674105e-06,
      "loss": 10.3242,
      "step": 337
    },
    {
      "epoch": 0.04962014166697251,
      "grad_norm": 0.07625994086265564,
      "learning_rate": 6.107279541079769e-06,
      "loss": 10.3268,
      "step": 338
    },
    {
      "epoch": 0.049766946819833374,
      "grad_norm": 0.08614172041416168,
      "learning_rate": 5.915808849215304e-06,
      "loss": 10.3201,
      "step": 339
    },
    {
      "epoch": 0.049913751972694244,
      "grad_norm": 0.08884906768798828,
      "learning_rate": 5.727198717339511e-06,
      "loss": 10.3232,
      "step": 340
    },
    {
      "epoch": 0.05006055712555511,
      "grad_norm": 0.09030349552631378,
      "learning_rate": 5.54146138409355e-06,
      "loss": 10.3216,
      "step": 341
    },
    {
      "epoch": 0.05020736227841597,
      "grad_norm": 0.11042314767837524,
      "learning_rate": 5.358608901706802e-06,
      "loss": 10.3223,
      "step": 342
    },
    {
      "epoch": 0.05035416743127684,
      "grad_norm": 0.0956040620803833,
      "learning_rate": 5.178653135214812e-06,
      "loss": 10.3204,
      "step": 343
    },
    {
      "epoch": 0.0505009725841377,
      "grad_norm": 0.10560595989227295,
      "learning_rate": 5.001605761689398e-06,
      "loss": 10.333,
      "step": 344
    },
    {
      "epoch": 0.05064777773699857,
      "grad_norm": 0.10146863758563995,
      "learning_rate": 4.827478269480895e-06,
      "loss": 10.3263,
      "step": 345
    },
    {
      "epoch": 0.050794582889859434,
      "grad_norm": 0.10761269927024841,
      "learning_rate": 4.65628195747273e-06,
      "loss": 10.3272,
      "step": 346
    },
    {
      "epoch": 0.0509413880427203,
      "grad_norm": 0.11111512780189514,
      "learning_rate": 4.488027934348271e-06,
      "loss": 10.3298,
      "step": 347
    },
    {
      "epoch": 0.051088193195581166,
      "grad_norm": 0.16782677173614502,
      "learning_rate": 4.322727117869951e-06,
      "loss": 10.3297,
      "step": 348
    },
    {
      "epoch": 0.05123499834844203,
      "grad_norm": 0.13419117033481598,
      "learning_rate": 4.16039023417088e-06,
      "loss": 10.3217,
      "step": 349
    },
    {
      "epoch": 0.0513818035013029,
      "grad_norm": 0.30833181738853455,
      "learning_rate": 4.001027817058789e-06,
      "loss": 10.3408,
      "step": 350
    },
    {
      "epoch": 0.0513818035013029,
      "eval_loss": 10.325580596923828,
      "eval_runtime": 40.8534,
      "eval_samples_per_second": 280.834,
      "eval_steps_per_second": 70.227,
      "step": 350
    },
    {
      "epoch": 0.05152860865416376,
      "grad_norm": 0.153910294175148,
      "learning_rate": 3.844650207332562e-06,
      "loss": 10.3357,
      "step": 351
    },
    {
      "epoch": 0.051675413807024624,
      "grad_norm": 0.1252238154411316,
      "learning_rate": 3.691267552111183e-06,
      "loss": 10.3391,
      "step": 352
    },
    {
      "epoch": 0.051822218959885494,
      "grad_norm": 0.07995588332414627,
      "learning_rate": 3.54088980417534e-06,
      "loss": 10.3333,
      "step": 353
    },
    {
      "epoch": 0.051969024112746356,
      "grad_norm": 0.07696188986301422,
      "learning_rate": 3.393526721321616e-06,
      "loss": 10.3322,
      "step": 354
    },
    {
      "epoch": 0.052115829265607226,
      "grad_norm": 0.06434796750545502,
      "learning_rate": 3.249187865729264e-06,
      "loss": 10.3319,
      "step": 355
    },
    {
      "epoch": 0.05226263441846809,
      "grad_norm": 0.06256424635648727,
      "learning_rate": 3.1078826033397843e-06,
      "loss": 10.325,
      "step": 356
    },
    {
      "epoch": 0.05240943957132895,
      "grad_norm": 0.06362050026655197,
      "learning_rate": 2.9696201032491434e-06,
      "loss": 10.3244,
      "step": 357
    },
    {
      "epoch": 0.05255624472418982,
      "grad_norm": 0.06841494143009186,
      "learning_rate": 2.8344093371128424e-06,
      "loss": 10.3259,
      "step": 358
    },
    {
      "epoch": 0.052703049877050684,
      "grad_norm": 0.06482333689928055,
      "learning_rate": 2.70225907856374e-06,
      "loss": 10.3296,
      "step": 359
    },
    {
      "epoch": 0.05284985502991155,
      "grad_norm": 0.06362397223711014,
      "learning_rate": 2.573177902642726e-06,
      "loss": 10.3246,
      "step": 360
    },
    {
      "epoch": 0.052996660182772416,
      "grad_norm": 0.10203403979539871,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 10.3307,
      "step": 361
    },
    {
      "epoch": 0.05314346533563328,
      "grad_norm": 0.05716697499155998,
      "learning_rate": 2.324256102563188e-06,
      "loss": 10.333,
      "step": 362
    },
    {
      "epoch": 0.05329027048849415,
      "grad_norm": 0.07640895247459412,
      "learning_rate": 2.204431630583548e-06,
      "loss": 10.3311,
      "step": 363
    },
    {
      "epoch": 0.05343707564135501,
      "grad_norm": 0.05863022431731224,
      "learning_rate": 2.087708544541689e-06,
      "loss": 10.3189,
      "step": 364
    },
    {
      "epoch": 0.053583880794215874,
      "grad_norm": 0.06322285532951355,
      "learning_rate": 1.974094418431388e-06,
      "loss": 10.3253,
      "step": 365
    },
    {
      "epoch": 0.053730685947076744,
      "grad_norm": 0.06476978957653046,
      "learning_rate": 1.8635966245104664e-06,
      "loss": 10.3276,
      "step": 366
    },
    {
      "epoch": 0.053877491099937606,
      "grad_norm": 0.07806657999753952,
      "learning_rate": 1.7562223328224325e-06,
      "loss": 10.3272,
      "step": 367
    },
    {
      "epoch": 0.054024296252798476,
      "grad_norm": 0.06654366850852966,
      "learning_rate": 1.6519785107311891e-06,
      "loss": 10.3252,
      "step": 368
    },
    {
      "epoch": 0.05417110140565934,
      "grad_norm": 0.06841534376144409,
      "learning_rate": 1.5508719224689717e-06,
      "loss": 10.3211,
      "step": 369
    },
    {
      "epoch": 0.0543179065585202,
      "grad_norm": 0.05707764998078346,
      "learning_rate": 1.4529091286973995e-06,
      "loss": 10.3314,
      "step": 370
    },
    {
      "epoch": 0.05446471171138107,
      "grad_norm": 0.08282582461833954,
      "learning_rate": 1.358096486081778e-06,
      "loss": 10.3208,
      "step": 371
    },
    {
      "epoch": 0.054611516864241934,
      "grad_norm": 0.07956176996231079,
      "learning_rate": 1.2664401468786114e-06,
      "loss": 10.3265,
      "step": 372
    },
    {
      "epoch": 0.0547583220171028,
      "grad_norm": 0.06514423340559006,
      "learning_rate": 1.1779460585363944e-06,
      "loss": 10.3231,
      "step": 373
    },
    {
      "epoch": 0.054905127169963666,
      "grad_norm": 0.06955509632825851,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 10.3287,
      "step": 374
    },
    {
      "epoch": 0.05505193232282453,
      "grad_norm": 0.06904707849025726,
      "learning_rate": 1.0104673978866164e-06,
      "loss": 10.3216,
      "step": 375
    },
    {
      "epoch": 0.0551987374756854,
      "grad_norm": 0.06475381553173065,
      "learning_rate": 9.314936930293283e-07,
      "loss": 10.3313,
      "step": 376
    },
    {
      "epoch": 0.05534554262854626,
      "grad_norm": 0.06641801446676254,
      "learning_rate": 8.557039732283944e-07,
      "loss": 10.3227,
      "step": 377
    },
    {
      "epoch": 0.055492347781407124,
      "grad_norm": 0.07730254530906677,
      "learning_rate": 7.83103156370113e-07,
      "loss": 10.3248,
      "step": 378
    },
    {
      "epoch": 0.055639152934267994,
      "grad_norm": 0.07971889525651932,
      "learning_rate": 7.136959534174592e-07,
      "loss": 10.3144,
      "step": 379
    },
    {
      "epoch": 0.055785958087128856,
      "grad_norm": 0.07381610572338104,
      "learning_rate": 6.474868681043578e-07,
      "loss": 10.3267,
      "step": 380
    },
    {
      "epoch": 0.055932763239989726,
      "grad_norm": 0.08179368078708649,
      "learning_rate": 5.844801966434832e-07,
      "loss": 10.3195,
      "step": 381
    },
    {
      "epoch": 0.05607956839285059,
      "grad_norm": 0.08641988039016724,
      "learning_rate": 5.246800274474439e-07,
      "loss": 10.3268,
      "step": 382
    },
    {
      "epoch": 0.05622637354571145,
      "grad_norm": 0.07887589931488037,
      "learning_rate": 4.680902408635335e-07,
      "loss": 10.3292,
      "step": 383
    },
    {
      "epoch": 0.05637317869857232,
      "grad_norm": 0.09002108871936798,
      "learning_rate": 4.1471450892189846e-07,
      "loss": 10.3262,
      "step": 384
    },
    {
      "epoch": 0.056519983851433184,
      "grad_norm": 0.07206162810325623,
      "learning_rate": 3.6455629509730136e-07,
      "loss": 10.3272,
      "step": 385
    },
    {
      "epoch": 0.05666678900429405,
      "grad_norm": 0.07797471433877945,
      "learning_rate": 3.1761885408435054e-07,
      "loss": 10.3229,
      "step": 386
    },
    {
      "epoch": 0.056813594157154916,
      "grad_norm": 0.11121383309364319,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 10.3183,
      "step": 387
    },
    {
      "epoch": 0.05696039931001578,
      "grad_norm": 0.08970195055007935,
      "learning_rate": 2.334182641175686e-07,
      "loss": 10.328,
      "step": 388
    },
    {
      "epoch": 0.05710720446287665,
      "grad_norm": 0.07631402462720871,
      "learning_rate": 1.9616057881935436e-07,
      "loss": 10.3236,
      "step": 389
    },
    {
      "epoch": 0.05725400961573751,
      "grad_norm": 0.0931524783372879,
      "learning_rate": 1.6213459328950352e-07,
      "loss": 10.3283,
      "step": 390
    },
    {
      "epoch": 0.05740081476859838,
      "grad_norm": 0.09834590554237366,
      "learning_rate": 1.3134251542544774e-07,
      "loss": 10.32,
      "step": 391
    },
    {
      "epoch": 0.057547619921459243,
      "grad_norm": 0.09804739058017731,
      "learning_rate": 1.0378634328099269e-07,
      "loss": 10.319,
      "step": 392
    },
    {
      "epoch": 0.057694425074320106,
      "grad_norm": 0.1139400452375412,
      "learning_rate": 7.946786493666647e-08,
      "loss": 10.3223,
      "step": 393
    },
    {
      "epoch": 0.057841230227180976,
      "grad_norm": 0.08956670761108398,
      "learning_rate": 5.838865838366792e-08,
      "loss": 10.3264,
      "step": 394
    },
    {
      "epoch": 0.05798803538004184,
      "grad_norm": 0.10495942085981369,
      "learning_rate": 4.055009142152067e-08,
      "loss": 10.3278,
      "step": 395
    },
    {
      "epoch": 0.05813484053290271,
      "grad_norm": 0.09951158612966537,
      "learning_rate": 2.595332156925534e-08,
      "loss": 10.3286,
      "step": 396
    },
    {
      "epoch": 0.05828164568576357,
      "grad_norm": 0.15127702057361603,
      "learning_rate": 1.4599295990352924e-08,
      "loss": 10.3261,
      "step": 397
    },
    {
      "epoch": 0.058428450838624434,
      "grad_norm": 0.13188035786151886,
      "learning_rate": 6.488751431266149e-09,
      "loss": 10.3251,
      "step": 398
    },
    {
      "epoch": 0.0585752559914853,
      "grad_norm": 0.15595850348472595,
      "learning_rate": 1.622214173602199e-09,
      "loss": 10.3293,
      "step": 399
    },
    {
      "epoch": 0.058722061144346166,
      "grad_norm": 0.48936566710472107,
      "learning_rate": 0.0,
      "loss": 10.337,
      "step": 400
    },
    {
      "epoch": 0.058722061144346166,
      "eval_loss": 10.32548999786377,
      "eval_runtime": 40.5941,
      "eval_samples_per_second": 282.627,
      "eval_steps_per_second": 70.675,
      "step": 400
    }
  ],
  "logging_steps": 1,
  "max_steps": 400,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 44843365564416.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}