{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 5412,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.011092623405435386,
      "grad_norm": 2.009918689727783,
      "learning_rate": 1.998002219755827e-05,
      "loss": 0.5022,
      "step": 10
    },
    {
      "epoch": 0.022185246810870772,
      "grad_norm": 0.49915584921836853,
      "learning_rate": 1.995782463928968e-05,
      "loss": 0.1943,
      "step": 20
    },
    {
      "epoch": 0.033277870216306155,
      "grad_norm": 0.47061625123023987,
      "learning_rate": 1.993562708102109e-05,
      "loss": 0.1436,
      "step": 30
    },
    {
      "epoch": 0.044370493621741544,
      "grad_norm": 0.2260124832391739,
      "learning_rate": 1.99134295227525e-05,
      "loss": 0.1106,
      "step": 40
    },
    {
      "epoch": 0.05546311702717693,
      "grad_norm": 0.31766048073768616,
      "learning_rate": 1.989123196448391e-05,
      "loss": 0.0994,
      "step": 50
    },
    {
      "epoch": 0.06655574043261231,
      "grad_norm": 0.2788391709327698,
      "learning_rate": 1.9869034406215316e-05,
      "loss": 0.111,
      "step": 60
    },
    {
      "epoch": 0.0776483638380477,
      "grad_norm": 0.24376386404037476,
      "learning_rate": 1.9846836847946726e-05,
      "loss": 0.0849,
      "step": 70
    },
    {
      "epoch": 0.08874098724348309,
      "grad_norm": 0.1789788454771042,
      "learning_rate": 1.982463928967814e-05,
      "loss": 0.0971,
      "step": 80
    },
    {
      "epoch": 0.09983361064891846,
      "grad_norm": 0.29698437452316284,
      "learning_rate": 1.9802441731409546e-05,
      "loss": 0.0856,
      "step": 90
    },
    {
      "epoch": 0.11092623405435385,
      "grad_norm": 0.2553412616252899,
      "learning_rate": 1.9780244173140956e-05,
      "loss": 0.0976,
      "step": 100
    },
    {
      "epoch": 0.12201885745978924,
      "grad_norm": 0.20608791708946228,
      "learning_rate": 1.9758046614872365e-05,
      "loss": 0.0947,
      "step": 110
    },
    {
      "epoch": 0.13311148086522462,
      "grad_norm": 0.23551669716835022,
      "learning_rate": 1.9735849056603775e-05,
      "loss": 0.0965,
      "step": 120
    },
    {
      "epoch": 0.14420410427066002,
      "grad_norm": 0.3207148611545563,
      "learning_rate": 1.9713651498335185e-05,
      "loss": 0.0931,
      "step": 130
    },
    {
      "epoch": 0.1552967276760954,
      "grad_norm": 0.34813177585601807,
      "learning_rate": 1.9691453940066595e-05,
      "loss": 0.1,
      "step": 140
    },
    {
      "epoch": 0.16638935108153077,
      "grad_norm": 0.41501104831695557,
      "learning_rate": 1.9669256381798e-05,
      "loss": 0.0922,
      "step": 150
    },
    {
      "epoch": 0.17748197448696618,
      "grad_norm": 0.24550506472587585,
      "learning_rate": 1.964705882352941e-05,
      "loss": 0.0857,
      "step": 160
    },
    {
      "epoch": 0.18857459789240155,
      "grad_norm": 0.4205056130886078,
      "learning_rate": 1.9624861265260825e-05,
      "loss": 0.0831,
      "step": 170
    },
    {
      "epoch": 0.19966722129783693,
      "grad_norm": 0.6002993583679199,
      "learning_rate": 1.9602663706992235e-05,
      "loss": 0.0792,
      "step": 180
    },
    {
      "epoch": 0.21075984470327233,
      "grad_norm": 0.27535638213157654,
      "learning_rate": 1.958046614872364e-05,
      "loss": 0.0786,
      "step": 190
    },
    {
      "epoch": 0.2218524681087077,
      "grad_norm": 0.41602805256843567,
      "learning_rate": 1.955826859045505e-05,
      "loss": 0.0789,
      "step": 200
    },
    {
      "epoch": 0.23294509151414308,
      "grad_norm": 0.41827332973480225,
      "learning_rate": 1.953607103218646e-05,
      "loss": 0.0663,
      "step": 210
    },
    {
      "epoch": 0.24403771491957849,
      "grad_norm": 0.47976112365722656,
      "learning_rate": 1.951387347391787e-05,
      "loss": 0.0727,
      "step": 220
    },
    {
      "epoch": 0.25513033832501386,
      "grad_norm": 0.7771281599998474,
      "learning_rate": 1.949167591564928e-05,
      "loss": 0.0776,
      "step": 230
    },
    {
      "epoch": 0.26622296173044924,
      "grad_norm": 1.0573914051055908,
      "learning_rate": 1.946947835738069e-05,
      "loss": 0.0828,
      "step": 240
    },
    {
      "epoch": 0.2773155851358846,
      "grad_norm": 0.5186040997505188,
      "learning_rate": 1.9447280799112097e-05,
      "loss": 0.0767,
      "step": 250
    },
    {
      "epoch": 0.28840820854132004,
      "grad_norm": 0.5594862699508667,
      "learning_rate": 1.942508324084351e-05,
      "loss": 0.0699,
      "step": 260
    },
    {
      "epoch": 0.2995008319467554,
      "grad_norm": 0.5147818922996521,
      "learning_rate": 1.940288568257492e-05,
      "loss": 0.0591,
      "step": 270
    },
    {
      "epoch": 0.3105934553521908,
      "grad_norm": 0.5510437488555908,
      "learning_rate": 1.938068812430633e-05,
      "loss": 0.0761,
      "step": 280
    },
    {
      "epoch": 0.32168607875762617,
      "grad_norm": 0.4817625880241394,
      "learning_rate": 1.9358490566037736e-05,
      "loss": 0.0688,
      "step": 290
    },
    {
      "epoch": 0.33277870216306155,
      "grad_norm": 0.504362940788269,
      "learning_rate": 1.9336293007769146e-05,
      "loss": 0.0691,
      "step": 300
    },
    {
      "epoch": 0.343871325568497,
      "grad_norm": 0.5732384920120239,
      "learning_rate": 1.9314095449500556e-05,
      "loss": 0.057,
      "step": 310
    },
    {
      "epoch": 0.35496394897393235,
      "grad_norm": 0.5892298221588135,
      "learning_rate": 1.9291897891231966e-05,
      "loss": 0.059,
      "step": 320
    },
    {
      "epoch": 0.36605657237936773,
      "grad_norm": 0.3680512309074402,
      "learning_rate": 1.9269700332963376e-05,
      "loss": 0.0706,
      "step": 330
    },
    {
      "epoch": 0.3771491957848031,
      "grad_norm": 0.41971662640571594,
      "learning_rate": 1.9247502774694786e-05,
      "loss": 0.053,
      "step": 340
    },
    {
      "epoch": 0.3882418191902385,
      "grad_norm": 0.37261390686035156,
      "learning_rate": 1.9225305216426195e-05,
      "loss": 0.0632,
      "step": 350
    },
    {
      "epoch": 0.39933444259567386,
      "grad_norm": 0.48256734013557434,
      "learning_rate": 1.9203107658157605e-05,
      "loss": 0.0651,
      "step": 360
    },
    {
      "epoch": 0.4104270660011093,
      "grad_norm": 0.7914339303970337,
      "learning_rate": 1.9180910099889015e-05,
      "loss": 0.0659,
      "step": 370
    },
    {
      "epoch": 0.42151968940654466,
      "grad_norm": 0.6772429347038269,
      "learning_rate": 1.915871254162042e-05,
      "loss": 0.0655,
      "step": 380
    },
    {
      "epoch": 0.43261231281198004,
      "grad_norm": 0.5621687173843384,
      "learning_rate": 1.913651498335183e-05,
      "loss": 0.0707,
      "step": 390
    },
    {
      "epoch": 0.4437049362174154,
      "grad_norm": 0.30515748262405396,
      "learning_rate": 1.911431742508324e-05,
      "loss": 0.0548,
      "step": 400
    },
    {
      "epoch": 0.4547975596228508,
      "grad_norm": 0.5506859421730042,
      "learning_rate": 1.909211986681465e-05,
      "loss": 0.0554,
      "step": 410
    },
    {
      "epoch": 0.46589018302828616,
      "grad_norm": 0.6387749314308167,
      "learning_rate": 1.906992230854606e-05,
      "loss": 0.0554,
      "step": 420
    },
    {
      "epoch": 0.4769828064337216,
      "grad_norm": 0.18500734865665436,
      "learning_rate": 1.904772475027747e-05,
      "loss": 0.0558,
      "step": 430
    },
    {
      "epoch": 0.48807542983915697,
      "grad_norm": 0.6815407276153564,
      "learning_rate": 1.902552719200888e-05,
      "loss": 0.0537,
      "step": 440
    },
    {
      "epoch": 0.49916805324459235,
      "grad_norm": 0.4826994836330414,
      "learning_rate": 1.900332963374029e-05,
      "loss": 0.0567,
      "step": 450
    },
    {
      "epoch": 0.5102606766500277,
      "grad_norm": 0.24618124961853027,
      "learning_rate": 1.89811320754717e-05,
      "loss": 0.0614,
      "step": 460
    },
    {
      "epoch": 0.5213533000554631,
      "grad_norm": 1.0037415027618408,
      "learning_rate": 1.895893451720311e-05,
      "loss": 0.0604,
      "step": 470
    },
    {
      "epoch": 0.5324459234608985,
      "grad_norm": 0.6000948548316956,
      "learning_rate": 1.8936736958934517e-05,
      "loss": 0.0683,
      "step": 480
    },
    {
      "epoch": 0.5435385468663338,
      "grad_norm": 0.3302474319934845,
      "learning_rate": 1.8914539400665927e-05,
      "loss": 0.0543,
      "step": 490
    },
    {
      "epoch": 0.5546311702717692,
      "grad_norm": 0.5560783743858337,
      "learning_rate": 1.8892341842397337e-05,
      "loss": 0.0569,
      "step": 500
    },
    {
      "epoch": 0.5657237936772047,
      "grad_norm": 0.8041097521781921,
      "learning_rate": 1.8870144284128747e-05,
      "loss": 0.0544,
      "step": 510
    },
    {
      "epoch": 0.5768164170826401,
      "grad_norm": 0.6846103668212891,
      "learning_rate": 1.8847946725860156e-05,
      "loss": 0.058,
      "step": 520
    },
    {
      "epoch": 0.5879090404880755,
      "grad_norm": 0.50434809923172,
      "learning_rate": 1.8825749167591566e-05,
      "loss": 0.0654,
      "step": 530
    },
    {
      "epoch": 0.5990016638935108,
      "grad_norm": 0.54362553358078,
      "learning_rate": 1.8803551609322976e-05,
      "loss": 0.0582,
      "step": 540
    },
    {
      "epoch": 0.6100942872989462,
      "grad_norm": 0.6166839599609375,
      "learning_rate": 1.8781354051054386e-05,
      "loss": 0.0672,
      "step": 550
    },
    {
      "epoch": 0.6211869107043816,
      "grad_norm": 0.4353054165840149,
      "learning_rate": 1.8759156492785796e-05,
      "loss": 0.0532,
      "step": 560
    },
    {
      "epoch": 0.632279534109817,
      "grad_norm": 0.6025580167770386,
      "learning_rate": 1.8736958934517206e-05,
      "loss": 0.0746,
      "step": 570
    },
    {
      "epoch": 0.6433721575152523,
      "grad_norm": 0.7259892225265503,
      "learning_rate": 1.8714761376248612e-05,
      "loss": 0.0642,
      "step": 580
    },
    {
      "epoch": 0.6544647809206877,
      "grad_norm": 0.4940318465232849,
      "learning_rate": 1.8692563817980022e-05,
      "loss": 0.0547,
      "step": 590
    },
    {
      "epoch": 0.6655574043261231,
      "grad_norm": 0.7005699872970581,
      "learning_rate": 1.8670366259711435e-05,
      "loss": 0.0522,
      "step": 600
    },
    {
      "epoch": 0.6766500277315585,
      "grad_norm": 0.4530707895755768,
      "learning_rate": 1.8648168701442845e-05,
      "loss": 0.054,
      "step": 610
    },
    {
      "epoch": 0.687742651136994,
      "grad_norm": 0.9097110629081726,
      "learning_rate": 1.8625971143174252e-05,
      "loss": 0.0622,
      "step": 620
    },
    {
      "epoch": 0.6988352745424293,
      "grad_norm": 0.5374599695205688,
      "learning_rate": 1.860377358490566e-05,
      "loss": 0.0522,
      "step": 630
    },
    {
      "epoch": 0.7099278979478647,
      "grad_norm": 0.3849945664405823,
      "learning_rate": 1.858157602663707e-05,
      "loss": 0.0571,
      "step": 640
    },
    {
      "epoch": 0.7210205213533001,
      "grad_norm": 0.5918008685112,
      "learning_rate": 1.855937846836848e-05,
      "loss": 0.0576,
      "step": 650
    },
    {
      "epoch": 0.7321131447587355,
      "grad_norm": 0.3229956030845642,
      "learning_rate": 1.853718091009989e-05,
      "loss": 0.0537,
      "step": 660
    },
    {
      "epoch": 0.7432057681641708,
      "grad_norm": 0.5264039039611816,
      "learning_rate": 1.85149833518313e-05,
      "loss": 0.06,
      "step": 670
    },
    {
      "epoch": 0.7542983915696062,
      "grad_norm": 0.36795660853385925,
      "learning_rate": 1.8492785793562708e-05,
      "loss": 0.0511,
      "step": 680
    },
    {
      "epoch": 0.7653910149750416,
      "grad_norm": 0.5905130505561829,
      "learning_rate": 1.847058823529412e-05,
      "loss": 0.0636,
      "step": 690
    },
    {
      "epoch": 0.776483638380477,
      "grad_norm": 0.36266571283340454,
      "learning_rate": 1.844839067702553e-05,
      "loss": 0.0598,
      "step": 700
    },
    {
      "epoch": 0.7875762617859123,
      "grad_norm": 0.4978592097759247,
      "learning_rate": 1.8426193118756937e-05,
      "loss": 0.0607,
      "step": 710
    },
    {
      "epoch": 0.7986688851913477,
      "grad_norm": 0.4635021686553955,
      "learning_rate": 1.8403995560488347e-05,
      "loss": 0.0543,
      "step": 720
    },
    {
      "epoch": 0.8097615085967831,
      "grad_norm": 0.44571858644485474,
      "learning_rate": 1.8381798002219757e-05,
      "loss": 0.0678,
      "step": 730
    },
    {
      "epoch": 0.8208541320022186,
      "grad_norm": 0.8265877366065979,
      "learning_rate": 1.8359600443951167e-05,
      "loss": 0.0552,
      "step": 740
    },
    {
      "epoch": 0.831946755407654,
      "grad_norm": 0.5776472091674805,
      "learning_rate": 1.8337402885682577e-05,
      "loss": 0.0505,
      "step": 750
    },
    {
      "epoch": 0.8430393788130893,
      "grad_norm": 0.299274742603302,
      "learning_rate": 1.8315205327413986e-05,
      "loss": 0.0619,
      "step": 760
    },
    {
      "epoch": 0.8541320022185247,
      "grad_norm": 0.587645947933197,
      "learning_rate": 1.8293007769145393e-05,
      "loss": 0.0575,
      "step": 770
    },
    {
      "epoch": 0.8652246256239601,
      "grad_norm": 0.39164137840270996,
      "learning_rate": 1.8270810210876806e-05,
      "loss": 0.0458,
      "step": 780
    },
    {
      "epoch": 0.8763172490293955,
      "grad_norm": 0.4663292169570923,
      "learning_rate": 1.8248612652608216e-05,
      "loss": 0.048,
      "step": 790
    },
    {
      "epoch": 0.8874098724348308,
      "grad_norm": 0.5804581642150879,
      "learning_rate": 1.8226415094339626e-05,
      "loss": 0.0557,
      "step": 800
    },
    {
      "epoch": 0.8985024958402662,
      "grad_norm": 0.4279440939426422,
      "learning_rate": 1.8204217536071032e-05,
      "loss": 0.051,
      "step": 810
    },
    {
      "epoch": 0.9095951192457016,
      "grad_norm": 0.5384302735328674,
      "learning_rate": 1.8182019977802442e-05,
      "loss": 0.0558,
      "step": 820
    },
    {
      "epoch": 0.920687742651137,
      "grad_norm": 0.5049973726272583,
      "learning_rate": 1.8159822419533852e-05,
      "loss": 0.047,
      "step": 830
    },
    {
      "epoch": 0.9317803660565723,
      "grad_norm": 0.8061177134513855,
      "learning_rate": 1.8137624861265262e-05,
      "loss": 0.0572,
      "step": 840
    },
    {
      "epoch": 0.9428729894620078,
      "grad_norm": 0.6056540012359619,
      "learning_rate": 1.8115427302996672e-05,
      "loss": 0.0539,
      "step": 850
    },
    {
      "epoch": 0.9539656128674432,
      "grad_norm": 0.3001384139060974,
      "learning_rate": 1.8093229744728082e-05,
      "loss": 0.053,
      "step": 860
    },
    {
      "epoch": 0.9650582362728786,
      "grad_norm": 0.6709749102592468,
      "learning_rate": 1.807103218645949e-05,
      "loss": 0.0526,
      "step": 870
    },
    {
      "epoch": 0.9761508596783139,
      "grad_norm": 0.8233507871627808,
      "learning_rate": 1.80488346281909e-05,
      "loss": 0.0621,
      "step": 880
    },
    {
      "epoch": 0.9872434830837493,
      "grad_norm": 0.5757150650024414,
      "learning_rate": 1.802663706992231e-05,
      "loss": 0.0546,
      "step": 890
    },
    {
      "epoch": 0.9983361064891847,
      "grad_norm": 0.6834889054298401,
      "learning_rate": 1.800443951165372e-05,
      "loss": 0.0462,
      "step": 900
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8720027017899359,
      "eval_f1": 0.5678449258836944,
      "eval_loss": 0.34414270520210266,
      "eval_precision": 0.7929936305732485,
      "eval_recall": 0.4422735346358792,
      "eval_runtime": 2.727,
      "eval_samples_per_second": 361.937,
      "eval_steps_per_second": 11.368,
      "step": 902
    },
    {
      "epoch": 1.0088740987243483,
      "grad_norm": 0.4411003887653351,
      "learning_rate": 1.7982241953385128e-05,
      "loss": 0.0312,
      "step": 910
    },
    {
      "epoch": 1.0199667221297837,
      "grad_norm": 0.40500083565711975,
      "learning_rate": 1.7960044395116538e-05,
      "loss": 0.0345,
      "step": 920
    },
    {
      "epoch": 1.031059345535219,
      "grad_norm": 0.9287449717521667,
      "learning_rate": 1.7937846836847947e-05,
      "loss": 0.0419,
      "step": 930
    },
    {
      "epoch": 1.0421519689406544,
      "grad_norm": 0.7969145178794861,
      "learning_rate": 1.7915649278579357e-05,
      "loss": 0.042,
      "step": 940
    },
    {
      "epoch": 1.0532445923460898,
      "grad_norm": 0.49280259013175964,
      "learning_rate": 1.7893451720310767e-05,
      "loss": 0.0427,
      "step": 950
    },
    {
      "epoch": 1.0643372157515252,
      "grad_norm": 1.1647624969482422,
      "learning_rate": 1.7871254162042177e-05,
      "loss": 0.0341,
      "step": 960
    },
    {
      "epoch": 1.0754298391569606,
      "grad_norm": 0.2908968925476074,
      "learning_rate": 1.7849056603773587e-05,
      "loss": 0.0389,
      "step": 970
    },
    {
      "epoch": 1.086522462562396,
      "grad_norm": 0.521218478679657,
      "learning_rate": 1.7826859045504997e-05,
      "loss": 0.0396,
      "step": 980
    },
    {
      "epoch": 1.0976150859678313,
      "grad_norm": 0.4156598150730133,
      "learning_rate": 1.7804661487236407e-05,
      "loss": 0.0359,
      "step": 990
    },
    {
      "epoch": 1.1087077093732667,
      "grad_norm": 0.7260242104530334,
      "learning_rate": 1.7782463928967813e-05,
      "loss": 0.0384,
      "step": 1000
    },
    {
      "epoch": 1.119800332778702,
      "grad_norm": 0.43658459186553955,
      "learning_rate": 1.7760266370699223e-05,
      "loss": 0.0383,
      "step": 1010
    },
    {
      "epoch": 1.1308929561841374,
      "grad_norm": 0.3585101068019867,
      "learning_rate": 1.7738068812430633e-05,
      "loss": 0.0401,
      "step": 1020
    },
    {
      "epoch": 1.141985579589573,
      "grad_norm": 0.5010389089584351,
      "learning_rate": 1.7715871254162043e-05,
      "loss": 0.0307,
      "step": 1030
    },
    {
      "epoch": 1.1530782029950084,
      "grad_norm": 0.6415812969207764,
      "learning_rate": 1.7693673695893453e-05,
      "loss": 0.0423,
      "step": 1040
    },
    {
      "epoch": 1.1641708264004438,
      "grad_norm": 0.6393259763717651,
      "learning_rate": 1.7671476137624862e-05,
      "loss": 0.0392,
      "step": 1050
    },
    {
      "epoch": 1.1752634498058792,
      "grad_norm": 0.6339041590690613,
      "learning_rate": 1.7649278579356272e-05,
      "loss": 0.0506,
      "step": 1060
    },
    {
      "epoch": 1.1863560732113145,
      "grad_norm": 0.6086763143539429,
      "learning_rate": 1.7627081021087682e-05,
      "loss": 0.0416,
      "step": 1070
    },
    {
      "epoch": 1.19744869661675,
      "grad_norm": 0.6648682355880737,
      "learning_rate": 1.7604883462819092e-05,
      "loss": 0.0379,
      "step": 1080
    },
    {
      "epoch": 1.2085413200221853,
      "grad_norm": 0.6018221378326416,
      "learning_rate": 1.7582685904550502e-05,
      "loss": 0.0354,
      "step": 1090
    },
    {
      "epoch": 1.2196339434276207,
      "grad_norm": 1.2366654872894287,
      "learning_rate": 1.756048834628191e-05,
      "loss": 0.037,
      "step": 1100
    },
    {
      "epoch": 1.230726566833056,
      "grad_norm": 0.5976310968399048,
      "learning_rate": 1.7538290788013318e-05,
      "loss": 0.0361,
      "step": 1110
    },
    {
      "epoch": 1.2418191902384914,
      "grad_norm": 0.8290308117866516,
      "learning_rate": 1.751609322974473e-05,
      "loss": 0.0383,
      "step": 1120
    },
    {
      "epoch": 1.2529118136439268,
      "grad_norm": 0.382548987865448,
      "learning_rate": 1.749389567147614e-05,
      "loss": 0.0449,
      "step": 1130
    },
    {
      "epoch": 1.2640044370493622,
      "grad_norm": 0.5307976007461548,
      "learning_rate": 1.7471698113207548e-05,
      "loss": 0.0364,
      "step": 1140
    },
    {
      "epoch": 1.2750970604547975,
      "grad_norm": 0.5508521795272827,
      "learning_rate": 1.7449500554938958e-05,
      "loss": 0.0312,
      "step": 1150
    },
    {
      "epoch": 1.286189683860233,
      "grad_norm": 0.37057268619537354,
      "learning_rate": 1.7427302996670368e-05,
      "loss": 0.0397,
      "step": 1160
    },
    {
      "epoch": 1.2972823072656683,
      "grad_norm": 0.4806898236274719,
      "learning_rate": 1.7405105438401777e-05,
      "loss": 0.0391,
      "step": 1170
    },
    {
      "epoch": 1.3083749306711037,
      "grad_norm": 0.759772002696991,
      "learning_rate": 1.7382907880133187e-05,
      "loss": 0.0322,
      "step": 1180
    },
    {
      "epoch": 1.319467554076539,
      "grad_norm": 0.39819085597991943,
      "learning_rate": 1.7360710321864597e-05,
      "loss": 0.0345,
      "step": 1190
    },
    {
      "epoch": 1.3305601774819744,
      "grad_norm": 0.7733897566795349,
      "learning_rate": 1.7338512763596004e-05,
      "loss": 0.0424,
      "step": 1200
    },
    {
      "epoch": 1.3416528008874098,
      "grad_norm": 0.7979075908660889,
      "learning_rate": 1.7316315205327417e-05,
      "loss": 0.0403,
      "step": 1210
    },
    {
      "epoch": 1.3527454242928454,
      "grad_norm": 0.5658752918243408,
      "learning_rate": 1.7294117647058827e-05,
      "loss": 0.0276,
      "step": 1220
    },
    {
      "epoch": 1.3638380476982808,
      "grad_norm": 0.25207844376564026,
      "learning_rate": 1.7271920088790237e-05,
      "loss": 0.0347,
      "step": 1230
    },
    {
      "epoch": 1.3749306711037161,
      "grad_norm": 0.6338945031166077,
      "learning_rate": 1.7249722530521643e-05,
      "loss": 0.0416,
      "step": 1240
    },
    {
      "epoch": 1.3860232945091515,
      "grad_norm": 0.8125913143157959,
      "learning_rate": 1.7227524972253053e-05,
      "loss": 0.0442,
      "step": 1250
    },
    {
      "epoch": 1.397115917914587,
      "grad_norm": 0.5067325234413147,
      "learning_rate": 1.7205327413984463e-05,
      "loss": 0.0358,
      "step": 1260
    },
    {
      "epoch": 1.4082085413200223,
      "grad_norm": 0.40132638812065125,
      "learning_rate": 1.7183129855715873e-05,
      "loss": 0.0303,
      "step": 1270
    },
    {
      "epoch": 1.4193011647254576,
      "grad_norm": 0.6337258815765381,
      "learning_rate": 1.7160932297447283e-05,
      "loss": 0.0471,
      "step": 1280
    },
    {
      "epoch": 1.430393788130893,
      "grad_norm": 0.43456393480300903,
      "learning_rate": 1.713873473917869e-05,
      "loss": 0.0354,
      "step": 1290
    },
    {
      "epoch": 1.4414864115363284,
      "grad_norm": 0.3291069269180298,
      "learning_rate": 1.7116537180910102e-05,
      "loss": 0.0312,
      "step": 1300
    },
    {
      "epoch": 1.4525790349417638,
      "grad_norm": 0.2960388660430908,
      "learning_rate": 1.7094339622641512e-05,
      "loss": 0.0362,
      "step": 1310
    },
    {
      "epoch": 1.4636716583471991,
      "grad_norm": 0.820751428604126,
      "learning_rate": 1.7072142064372922e-05,
      "loss": 0.0396,
      "step": 1320
    },
    {
      "epoch": 1.4747642817526345,
      "grad_norm": 0.7335907816886902,
      "learning_rate": 1.704994450610433e-05,
      "loss": 0.0386,
      "step": 1330
    },
    {
      "epoch": 1.48585690515807,
      "grad_norm": 0.40592002868652344,
      "learning_rate": 1.702774694783574e-05,
      "loss": 0.0358,
      "step": 1340
    },
    {
      "epoch": 1.4969495285635053,
      "grad_norm": 0.6194770336151123,
      "learning_rate": 1.7005549389567148e-05,
      "loss": 0.0361,
      "step": 1350
    },
    {
      "epoch": 1.5080421519689406,
      "grad_norm": 0.44590774178504944,
      "learning_rate": 1.6983351831298558e-05,
      "loss": 0.033,
      "step": 1360
    },
    {
      "epoch": 1.519134775374376,
      "grad_norm": 0.5857370495796204,
      "learning_rate": 1.6961154273029968e-05,
      "loss": 0.0423,
      "step": 1370
    },
    {
      "epoch": 1.5302273987798114,
      "grad_norm": 0.6809953451156616,
      "learning_rate": 1.6938956714761378e-05,
      "loss": 0.0345,
      "step": 1380
    },
    {
      "epoch": 1.5413200221852468,
      "grad_norm": 0.7102778553962708,
      "learning_rate": 1.6916759156492788e-05,
      "loss": 0.0372,
      "step": 1390
    },
    {
      "epoch": 1.5524126455906821,
      "grad_norm": 0.31781554222106934,
      "learning_rate": 1.6894561598224198e-05,
      "loss": 0.032,
      "step": 1400
    },
    {
      "epoch": 1.5635052689961175,
      "grad_norm": 0.4304943382740021,
      "learning_rate": 1.6872364039955607e-05,
      "loss": 0.0375,
      "step": 1410
    },
    {
      "epoch": 1.574597892401553,
      "grad_norm": 0.918550968170166,
      "learning_rate": 1.6850166481687017e-05,
      "loss": 0.0416,
      "step": 1420
    },
    {
      "epoch": 1.5856905158069883,
      "grad_norm": 0.722892701625824,
      "learning_rate": 1.6827968923418424e-05,
      "loss": 0.0359,
      "step": 1430
    },
    {
      "epoch": 1.5967831392124237,
      "grad_norm": 0.508703351020813,
      "learning_rate": 1.6805771365149834e-05,
      "loss": 0.0402,
      "step": 1440
    },
    {
      "epoch": 1.607875762617859,
      "grad_norm": 0.4725389778614044,
      "learning_rate": 1.6783573806881244e-05,
      "loss": 0.0335,
      "step": 1450
    },
    {
      "epoch": 1.6189683860232944,
      "grad_norm": 0.4699971079826355,
      "learning_rate": 1.6761376248612653e-05,
      "loss": 0.0277,
      "step": 1460
    },
    {
      "epoch": 1.6300610094287298,
      "grad_norm": 0.775764524936676,
      "learning_rate": 1.6739178690344063e-05,
      "loss": 0.0333,
      "step": 1470
    },
    {
      "epoch": 1.6411536328341652,
      "grad_norm": 0.6115106344223022,
      "learning_rate": 1.6716981132075473e-05,
      "loss": 0.0301,
      "step": 1480
    },
    {
      "epoch": 1.6522462562396005,
      "grad_norm": 0.743077278137207,
      "learning_rate": 1.6694783573806883e-05,
      "loss": 0.035,
      "step": 1490
    },
    {
      "epoch": 1.663338879645036,
      "grad_norm": 0.5189201235771179,
      "learning_rate": 1.6672586015538293e-05,
      "loss": 0.0475,
      "step": 1500
    },
    {
      "epoch": 1.6744315030504713,
      "grad_norm": 0.5089607238769531,
      "learning_rate": 1.6650388457269703e-05,
      "loss": 0.0377,
      "step": 1510
    },
    {
      "epoch": 1.6855241264559067,
      "grad_norm": 0.46067437529563904,
      "learning_rate": 1.6628190899001113e-05,
      "loss": 0.0297,
      "step": 1520
    },
    {
      "epoch": 1.6966167498613423,
      "grad_norm": 0.5661717653274536,
      "learning_rate": 1.660599334073252e-05,
      "loss": 0.0372,
      "step": 1530
    },
    {
      "epoch": 1.7077093732667776,
      "grad_norm": 0.45938414335250854,
      "learning_rate": 1.658379578246393e-05,
      "loss": 0.0371,
      "step": 1540
    },
    {
      "epoch": 1.718801996672213,
      "grad_norm": 0.7390128970146179,
      "learning_rate": 1.656159822419534e-05,
      "loss": 0.0332,
      "step": 1550
    },
    {
      "epoch": 1.7298946200776484,
      "grad_norm": 0.6475571990013123,
      "learning_rate": 1.653940066592675e-05,
      "loss": 0.0352,
      "step": 1560
    },
    {
      "epoch": 1.7409872434830838,
      "grad_norm": 0.8529049754142761,
      "learning_rate": 1.651720310765816e-05,
      "loss": 0.0386,
      "step": 1570
    },
    {
      "epoch": 1.7520798668885191,
      "grad_norm": 0.6980950236320496,
      "learning_rate": 1.649500554938957e-05,
      "loss": 0.0358,
      "step": 1580
    },
    {
      "epoch": 1.7631724902939545,
      "grad_norm": 0.5858293175697327,
      "learning_rate": 1.6472807991120978e-05,
      "loss": 0.0418,
      "step": 1590
    },
    {
      "epoch": 1.7742651136993899,
      "grad_norm": 0.49686577916145325,
      "learning_rate": 1.6450610432852388e-05,
      "loss": 0.0361,
      "step": 1600
    },
    {
      "epoch": 1.7853577371048253,
      "grad_norm": 0.3631349205970764,
      "learning_rate": 1.6428412874583798e-05,
      "loss": 0.0352,
      "step": 1610
    },
    {
      "epoch": 1.7964503605102606,
      "grad_norm": 0.5329940915107727,
      "learning_rate": 1.6406215316315204e-05,
      "loss": 0.04,
      "step": 1620
    },
    {
      "epoch": 1.807542983915696,
      "grad_norm": 0.5995050668716431,
      "learning_rate": 1.6384017758046614e-05,
      "loss": 0.0368,
      "step": 1630
    },
    {
      "epoch": 1.8186356073211316,
      "grad_norm": 1.3379504680633545,
      "learning_rate": 1.6361820199778028e-05,
      "loss": 0.0397,
      "step": 1640
    },
    {
      "epoch": 1.829728230726567,
      "grad_norm": 0.8026002645492554,
      "learning_rate": 1.6339622641509437e-05,
      "loss": 0.0318,
      "step": 1650
    },
    {
      "epoch": 1.8408208541320024,
      "grad_norm": 0.917950451374054,
      "learning_rate": 1.6317425083240844e-05,
      "loss": 0.0352,
      "step": 1660
    },
    {
      "epoch": 1.8519134775374377,
      "grad_norm": 0.6763226389884949,
      "learning_rate": 1.6295227524972254e-05,
      "loss": 0.0413,
      "step": 1670
    },
    {
      "epoch": 1.8630061009428731,
      "grad_norm": 0.5728912949562073,
      "learning_rate": 1.6273029966703664e-05,
      "loss": 0.0291,
      "step": 1680
    },
    {
      "epoch": 1.8740987243483085,
      "grad_norm": 0.6339443325996399,
      "learning_rate": 1.6250832408435074e-05,
      "loss": 0.0409,
      "step": 1690
    },
    {
      "epoch": 1.8851913477537439,
      "grad_norm": 0.6930853128433228,
      "learning_rate": 1.6228634850166483e-05,
      "loss": 0.0371,
      "step": 1700
    },
    {
      "epoch": 1.8962839711591792,
      "grad_norm": 0.37715452909469604,
      "learning_rate": 1.6206437291897893e-05,
      "loss": 0.031,
      "step": 1710
    },
    {
      "epoch": 1.9073765945646146,
      "grad_norm": 0.5656572580337524,
      "learning_rate": 1.61842397336293e-05,
      "loss": 0.0414,
      "step": 1720
    },
    {
      "epoch": 1.91846921797005,
      "grad_norm": 0.3563915491104126,
      "learning_rate": 1.6162042175360713e-05,
      "loss": 0.0276,
      "step": 1730
    },
    {
      "epoch": 1.9295618413754854,
      "grad_norm": 0.8260300159454346,
      "learning_rate": 1.6139844617092123e-05,
      "loss": 0.0337,
      "step": 1740
    },
    {
      "epoch": 1.9406544647809207,
      "grad_norm": 0.6907551884651184,
      "learning_rate": 1.6117647058823533e-05,
      "loss": 0.031,
      "step": 1750
    },
    {
      "epoch": 1.9517470881863561,
      "grad_norm": 0.5782826542854309,
      "learning_rate": 1.609544950055494e-05,
      "loss": 0.0419,
      "step": 1760
    },
    {
      "epoch": 1.9628397115917915,
      "grad_norm": 0.36368125677108765,
      "learning_rate": 1.607325194228635e-05,
      "loss": 0.0356,
      "step": 1770
    },
    {
      "epoch": 1.9739323349972269,
      "grad_norm": 0.369911789894104,
      "learning_rate": 1.605105438401776e-05,
      "loss": 0.0264,
      "step": 1780
    },
    {
      "epoch": 1.9850249584026622,
      "grad_norm": 0.6072190999984741,
      "learning_rate": 1.602885682574917e-05,
      "loss": 0.0366,
      "step": 1790
    },
    {
      "epoch": 1.9961175818080976,
      "grad_norm": 0.9733797311782837,
      "learning_rate": 1.600665926748058e-05,
      "loss": 0.0371,
      "step": 1800
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8973319824383654,
      "eval_f1": 0.6872427983539094,
      "eval_loss": 0.31070804595947266,
      "eval_precision": 0.8166259168704156,
      "eval_recall": 0.5932504440497336,
      "eval_runtime": 2.7114,
      "eval_samples_per_second": 364.014,
      "eval_steps_per_second": 11.433,
      "step": 1804
    },
    {
      "epoch": 2.0066555740432612,
      "grad_norm": 0.2766992449760437,
      "learning_rate": 1.598446170921199e-05,
      "loss": 0.0218,
      "step": 1810
    },
    {
      "epoch": 2.0177481974486966,
      "grad_norm": 0.6556938290596008,
      "learning_rate": 1.59622641509434e-05,
      "loss": 0.0187,
      "step": 1820
    },
    {
      "epoch": 2.028840820854132,
      "grad_norm": 0.24634911119937897,
      "learning_rate": 1.5940066592674808e-05,
      "loss": 0.0132,
      "step": 1830
    },
    {
      "epoch": 2.0399334442595674,
      "grad_norm": 0.683587372303009,
      "learning_rate": 1.5917869034406218e-05,
      "loss": 0.0158,
      "step": 1840
    },
    {
      "epoch": 2.0510260676650027,
      "grad_norm": 0.47117751836776733,
      "learning_rate": 1.5895671476137625e-05,
      "loss": 0.018,
      "step": 1850
    },
    {
      "epoch": 2.062118691070438,
      "grad_norm": 0.4102018475532532,
      "learning_rate": 1.5873473917869034e-05,
      "loss": 0.0153,
      "step": 1860
    },
    {
      "epoch": 2.0732113144758735,
      "grad_norm": 0.4639064371585846,
      "learning_rate": 1.5851276359600444e-05,
      "loss": 0.0171,
      "step": 1870
    },
    {
      "epoch": 2.084303937881309,
      "grad_norm": 0.49497199058532715,
      "learning_rate": 1.5829078801331854e-05,
      "loss": 0.0168,
      "step": 1880
    },
    {
      "epoch": 2.0953965612867442,
      "grad_norm": 0.6302080154418945,
      "learning_rate": 1.5806881243063264e-05,
      "loss": 0.0191,
      "step": 1890
    },
    {
      "epoch": 2.1064891846921796,
      "grad_norm": 0.3464473783969879,
      "learning_rate": 1.5784683684794674e-05,
      "loss": 0.0184,
      "step": 1900
    },
    {
      "epoch": 2.117581808097615,
      "grad_norm": 0.6507964730262756,
      "learning_rate": 1.5762486126526084e-05,
      "loss": 0.0135,
      "step": 1910
    },
    {
      "epoch": 2.1286744315030504,
      "grad_norm": 0.1759006232023239,
      "learning_rate": 1.5740288568257494e-05,
      "loss": 0.0118,
      "step": 1920
    },
    {
      "epoch": 2.1397670549084857,
      "grad_norm": 0.5952832698822021,
      "learning_rate": 1.5718091009988904e-05,
      "loss": 0.0157,
      "step": 1930
    },
    {
      "epoch": 2.150859678313921,
      "grad_norm": 0.8094580769538879,
      "learning_rate": 1.5695893451720313e-05,
      "loss": 0.0125,
      "step": 1940
    },
    {
      "epoch": 2.1619523017193565,
      "grad_norm": 0.6284286379814148,
      "learning_rate": 1.567369589345172e-05,
      "loss": 0.0178,
      "step": 1950
    },
    {
      "epoch": 2.173044925124792,
      "grad_norm": 0.6017957925796509,
      "learning_rate": 1.565149833518313e-05,
      "loss": 0.0136,
      "step": 1960
    },
    {
      "epoch": 2.1841375485302272,
      "grad_norm": 1.2836827039718628,
      "learning_rate": 1.562930077691454e-05,
      "loss": 0.014,
      "step": 1970
    },
    {
      "epoch": 2.1952301719356626,
      "grad_norm": 0.6213756203651428,
      "learning_rate": 1.560710321864595e-05,
      "loss": 0.0136,
      "step": 1980
    },
    {
      "epoch": 2.206322795341098,
      "grad_norm": 0.48122018575668335,
      "learning_rate": 1.558490566037736e-05,
      "loss": 0.0147,
      "step": 1990
    },
    {
      "epoch": 2.2174154187465334,
      "grad_norm": 0.7161231637001038,
      "learning_rate": 1.556270810210877e-05,
      "loss": 0.0105,
      "step": 2000
    },
    {
      "epoch": 2.2285080421519687,
      "grad_norm": 0.6598195433616638,
      "learning_rate": 1.554051054384018e-05,
      "loss": 0.0172,
      "step": 2010
    },
    {
      "epoch": 2.239600665557404,
      "grad_norm": 0.6077088117599487,
      "learning_rate": 1.551831298557159e-05,
      "loss": 0.0144,
      "step": 2020
    },
    {
      "epoch": 2.2506932889628395,
      "grad_norm": 0.5622262954711914,
      "learning_rate": 1.5496115427303e-05,
      "loss": 0.0175,
      "step": 2030
    },
    {
      "epoch": 2.261785912368275,
      "grad_norm": 0.1281885802745819,
      "learning_rate": 1.547391786903441e-05,
      "loss": 0.018,
      "step": 2040
    },
    {
      "epoch": 2.2728785357737102,
      "grad_norm": 0.408607542514801,
      "learning_rate": 1.5451720310765815e-05,
      "loss": 0.0095,
      "step": 2050
    },
    {
      "epoch": 2.283971159179146,
      "grad_norm": 0.7786557674407959,
      "learning_rate": 1.5429522752497225e-05,
      "loss": 0.0163,
      "step": 2060
    },
    {
      "epoch": 2.2950637825845814,
      "grad_norm": 0.22975876927375793,
      "learning_rate": 1.5407325194228635e-05,
      "loss": 0.0193,
      "step": 2070
    },
    {
      "epoch": 2.306156405990017,
      "grad_norm": 0.4313502311706543,
      "learning_rate": 1.5385127635960048e-05,
      "loss": 0.0186,
      "step": 2080
    },
    {
      "epoch": 2.317249029395452,
      "grad_norm": 0.794448733329773,
      "learning_rate": 1.5362930077691455e-05,
      "loss": 0.0159,
      "step": 2090
    },
    {
      "epoch": 2.3283416528008876,
      "grad_norm": 0.4191996157169342,
      "learning_rate": 1.5340732519422865e-05,
      "loss": 0.0208,
      "step": 2100
    },
    {
      "epoch": 2.339434276206323,
      "grad_norm": 0.5219516754150391,
      "learning_rate": 1.5318534961154274e-05,
      "loss": 0.0136,
      "step": 2110
    },
    {
      "epoch": 2.3505268996117583,
      "grad_norm": 1.1869897842407227,
      "learning_rate": 1.5296337402885684e-05,
      "loss": 0.0177,
      "step": 2120
    },
    {
      "epoch": 2.3616195230171937,
      "grad_norm": 0.12048923969268799,
      "learning_rate": 1.5274139844617094e-05,
      "loss": 0.0129,
      "step": 2130
    },
    {
      "epoch": 2.372712146422629,
      "grad_norm": 0.5213373899459839,
      "learning_rate": 1.5251942286348502e-05,
      "loss": 0.0127,
      "step": 2140
    },
    {
      "epoch": 2.3838047698280644,
      "grad_norm": 0.43704670667648315,
      "learning_rate": 1.5229744728079912e-05,
      "loss": 0.0154,
      "step": 2150
    },
    {
      "epoch": 2.3948973932335,
      "grad_norm": 0.567150890827179,
      "learning_rate": 1.5207547169811324e-05,
      "loss": 0.0117,
      "step": 2160
    },
    {
      "epoch": 2.405990016638935,
      "grad_norm": 0.571408212184906,
      "learning_rate": 1.5185349611542732e-05,
      "loss": 0.0152,
      "step": 2170
    },
    {
      "epoch": 2.4170826400443706,
      "grad_norm": 0.524834394454956,
      "learning_rate": 1.5163152053274142e-05,
      "loss": 0.017,
      "step": 2180
    },
    {
      "epoch": 2.428175263449806,
      "grad_norm": 0.5026165246963501,
      "learning_rate": 1.5140954495005552e-05,
      "loss": 0.0193,
      "step": 2190
    },
    {
      "epoch": 2.4392678868552413,
      "grad_norm": 0.3944782614707947,
      "learning_rate": 1.511875693673696e-05,
      "loss": 0.0159,
      "step": 2200
    },
    {
      "epoch": 2.4503605102606767,
      "grad_norm": 0.6460635662078857,
      "learning_rate": 1.509655937846837e-05,
      "loss": 0.0139,
      "step": 2210
    },
    {
      "epoch": 2.461453133666112,
      "grad_norm": 0.5573295950889587,
      "learning_rate": 1.507436182019978e-05,
      "loss": 0.0183,
      "step": 2220
    },
    {
      "epoch": 2.4725457570715474,
      "grad_norm": 0.5001458525657654,
      "learning_rate": 1.5052164261931188e-05,
      "loss": 0.0167,
      "step": 2230
    },
    {
      "epoch": 2.483638380476983,
      "grad_norm": 1.0214160680770874,
      "learning_rate": 1.5029966703662598e-05,
      "loss": 0.0182,
      "step": 2240
    },
    {
      "epoch": 2.494731003882418,
      "grad_norm": 0.5756349563598633,
      "learning_rate": 1.5007769145394009e-05,
      "loss": 0.0164,
      "step": 2250
    },
    {
      "epoch": 2.5058236272878536,
      "grad_norm": 0.771986722946167,
      "learning_rate": 1.4985571587125419e-05,
      "loss": 0.0183,
      "step": 2260
    },
    {
      "epoch": 2.516916250693289,
      "grad_norm": 0.5560303926467896,
      "learning_rate": 1.4963374028856827e-05,
      "loss": 0.0178,
      "step": 2270
    },
    {
      "epoch": 2.5280088740987243,
      "grad_norm": 0.8550804853439331,
      "learning_rate": 1.4941176470588237e-05,
      "loss": 0.0118,
      "step": 2280
    },
    {
      "epoch": 2.5391014975041597,
      "grad_norm": 0.8129355907440186,
      "learning_rate": 1.4918978912319645e-05,
      "loss": 0.0126,
      "step": 2290
    },
    {
      "epoch": 2.550194120909595,
      "grad_norm": 1.176377773284912,
      "learning_rate": 1.4896781354051055e-05,
      "loss": 0.0126,
      "step": 2300
    },
    {
      "epoch": 2.5612867443150305,
      "grad_norm": 0.1419341266155243,
      "learning_rate": 1.4874583795782465e-05,
      "loss": 0.0196,
      "step": 2310
    },
    {
      "epoch": 2.572379367720466,
      "grad_norm": 1.4311761856079102,
      "learning_rate": 1.4852386237513873e-05,
      "loss": 0.0143,
      "step": 2320
    },
    {
      "epoch": 2.583471991125901,
      "grad_norm": 0.9924690127372742,
      "learning_rate": 1.4830188679245283e-05,
      "loss": 0.0198,
      "step": 2330
    },
    {
      "epoch": 2.5945646145313366,
      "grad_norm": 0.6496040225028992,
      "learning_rate": 1.4807991120976695e-05,
      "loss": 0.014,
      "step": 2340
    },
    {
      "epoch": 2.605657237936772,
      "grad_norm": 0.2335842400789261,
      "learning_rate": 1.4785793562708104e-05,
      "loss": 0.0128,
      "step": 2350
    },
    {
      "epoch": 2.6167498613422073,
      "grad_norm": 0.6549142003059387,
      "learning_rate": 1.4763596004439513e-05,
      "loss": 0.019,
      "step": 2360
    },
    {
      "epoch": 2.6278424847476427,
      "grad_norm": 0.3264187276363373,
      "learning_rate": 1.4741398446170922e-05,
      "loss": 0.0168,
      "step": 2370
    },
    {
      "epoch": 2.638935108153078,
      "grad_norm": 0.4175679683685303,
      "learning_rate": 1.4719200887902332e-05,
      "loss": 0.0122,
      "step": 2380
    },
    {
      "epoch": 2.6500277315585135,
      "grad_norm": 0.2968366742134094,
      "learning_rate": 1.469700332963374e-05,
      "loss": 0.0189,
      "step": 2390
    },
    {
      "epoch": 2.661120354963949,
      "grad_norm": 1.5489327907562256,
      "learning_rate": 1.467480577136515e-05,
      "loss": 0.0158,
      "step": 2400
    },
    {
      "epoch": 2.672212978369384,
      "grad_norm": 0.8017039895057678,
      "learning_rate": 1.465260821309656e-05,
      "loss": 0.0178,
      "step": 2410
    },
    {
      "epoch": 2.6833056017748196,
      "grad_norm": 0.6136099100112915,
      "learning_rate": 1.4630410654827972e-05,
      "loss": 0.0164,
      "step": 2420
    },
    {
      "epoch": 2.6943982251802554,
      "grad_norm": 0.5461011528968811,
      "learning_rate": 1.460821309655938e-05,
      "loss": 0.0168,
      "step": 2430
    },
    {
      "epoch": 2.7054908485856908,
      "grad_norm": 0.2680525779724121,
      "learning_rate": 1.458601553829079e-05,
      "loss": 0.0138,
      "step": 2440
    },
    {
      "epoch": 2.716583471991126,
      "grad_norm": 0.6884289383888245,
      "learning_rate": 1.45638179800222e-05,
      "loss": 0.0145,
      "step": 2450
    },
    {
      "epoch": 2.7276760953965615,
      "grad_norm": 1.226928472518921,
      "learning_rate": 1.4541620421753608e-05,
      "loss": 0.0145,
      "step": 2460
    },
    {
      "epoch": 2.738768718801997,
      "grad_norm": 0.447221040725708,
      "learning_rate": 1.4519422863485018e-05,
      "loss": 0.0168,
      "step": 2470
    },
    {
      "epoch": 2.7498613422074323,
      "grad_norm": 0.47167718410491943,
      "learning_rate": 1.4497225305216428e-05,
      "loss": 0.0124,
      "step": 2480
    },
    {
      "epoch": 2.7609539656128677,
      "grad_norm": 1.326282262802124,
      "learning_rate": 1.4475027746947836e-05,
      "loss": 0.0183,
      "step": 2490
    },
    {
      "epoch": 2.772046589018303,
      "grad_norm": 0.9036744236946106,
      "learning_rate": 1.4452830188679246e-05,
      "loss": 0.023,
      "step": 2500
    },
    {
      "epoch": 2.7831392124237384,
      "grad_norm": 0.30263999104499817,
      "learning_rate": 1.4430632630410657e-05,
      "loss": 0.0163,
      "step": 2510
    },
    {
      "epoch": 2.794231835829174,
      "grad_norm": 0.7316049933433533,
      "learning_rate": 1.4408435072142067e-05,
      "loss": 0.0156,
      "step": 2520
    },
    {
      "epoch": 2.805324459234609,
      "grad_norm": 0.8268038630485535,
      "learning_rate": 1.4386237513873475e-05,
      "loss": 0.0169,
      "step": 2530
    },
    {
      "epoch": 2.8164170826400445,
      "grad_norm": 0.633816659450531,
      "learning_rate": 1.4364039955604885e-05,
      "loss": 0.0147,
      "step": 2540
    },
    {
      "epoch": 2.82750970604548,
      "grad_norm": 0.43913397192955017,
      "learning_rate": 1.4341842397336295e-05,
      "loss": 0.0122,
      "step": 2550
    },
    {
      "epoch": 2.8386023294509153,
      "grad_norm": 0.37594836950302124,
      "learning_rate": 1.4319644839067703e-05,
      "loss": 0.0149,
      "step": 2560
    },
    {
      "epoch": 2.8496949528563507,
      "grad_norm": 0.5090307593345642,
      "learning_rate": 1.4297447280799113e-05,
      "loss": 0.015,
      "step": 2570
    },
    {
      "epoch": 2.860787576261786,
      "grad_norm": 0.47723984718322754,
      "learning_rate": 1.4275249722530521e-05,
      "loss": 0.0166,
      "step": 2580
    },
    {
      "epoch": 2.8718801996672214,
      "grad_norm": 1.2512151002883911,
      "learning_rate": 1.4253052164261931e-05,
      "loss": 0.0136,
      "step": 2590
    },
    {
      "epoch": 2.882972823072657,
      "grad_norm": 0.676337718963623,
      "learning_rate": 1.4230854605993343e-05,
      "loss": 0.0152,
      "step": 2600
    },
    {
      "epoch": 2.894065446478092,
      "grad_norm": 0.7379328012466431,
      "learning_rate": 1.4208657047724752e-05,
      "loss": 0.0182,
      "step": 2610
    },
    {
      "epoch": 2.9051580698835275,
      "grad_norm": 0.8266171813011169,
      "learning_rate": 1.418645948945616e-05,
      "loss": 0.0207,
      "step": 2620
    },
    {
      "epoch": 2.916250693288963,
      "grad_norm": 0.9925495982170105,
      "learning_rate": 1.416426193118757e-05,
      "loss": 0.0157,
      "step": 2630
    },
    {
      "epoch": 2.9273433166943983,
      "grad_norm": 1.1120530366897583,
      "learning_rate": 1.414206437291898e-05,
      "loss": 0.0164,
      "step": 2640
    },
    {
      "epoch": 2.9384359400998337,
      "grad_norm": 0.9687446355819702,
      "learning_rate": 1.4119866814650389e-05,
      "loss": 0.0211,
      "step": 2650
    },
    {
      "epoch": 2.949528563505269,
      "grad_norm": 0.19811566174030304,
      "learning_rate": 1.4097669256381798e-05,
      "loss": 0.0181,
      "step": 2660
    },
    {
      "epoch": 2.9606211869107044,
      "grad_norm": 0.9253482818603516,
      "learning_rate": 1.4075471698113208e-05,
      "loss": 0.0158,
      "step": 2670
    },
    {
      "epoch": 2.97171381031614,
      "grad_norm": 0.677994430065155,
      "learning_rate": 1.405327413984462e-05,
      "loss": 0.024,
      "step": 2680
    },
    {
      "epoch": 2.982806433721575,
      "grad_norm": 0.3959580361843109,
      "learning_rate": 1.4031076581576028e-05,
      "loss": 0.009,
      "step": 2690
    },
    {
      "epoch": 2.9938990571270105,
      "grad_norm": 1.0569937229156494,
      "learning_rate": 1.4008879023307438e-05,
      "loss": 0.0187,
      "step": 2700
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_f1": 0.7171109200343938,
      "eval_loss": 0.3638182580471039,
      "eval_precision": 0.695,
      "eval_recall": 0.7406749555950266,
      "eval_runtime": 2.7252,
      "eval_samples_per_second": 362.177,
      "eval_steps_per_second": 11.375,
      "step": 2706
    },
    {
      "epoch": 3.004437049362174,
      "grad_norm": 0.22475391626358032,
      "learning_rate": 1.3986681465038848e-05,
      "loss": 0.0127,
      "step": 2710
    },
    {
      "epoch": 3.0155296727676095,
      "grad_norm": 0.42929062247276306,
      "learning_rate": 1.3964483906770256e-05,
      "loss": 0.0131,
      "step": 2720
    },
    {
      "epoch": 3.026622296173045,
      "grad_norm": 2.398026466369629,
      "learning_rate": 1.3942286348501666e-05,
      "loss": 0.0072,
      "step": 2730
    },
    {
      "epoch": 3.0377149195784803,
      "grad_norm": 0.14908376336097717,
      "learning_rate": 1.3920088790233076e-05,
      "loss": 0.0082,
      "step": 2740
    },
    {
      "epoch": 3.0488075429839157,
      "grad_norm": 0.8177769184112549,
      "learning_rate": 1.3897891231964484e-05,
      "loss": 0.0088,
      "step": 2750
    },
    {
      "epoch": 3.059900166389351,
      "grad_norm": 0.13737396895885468,
      "learning_rate": 1.3875693673695894e-05,
      "loss": 0.008,
      "step": 2760
    },
    {
      "epoch": 3.0709927897947864,
      "grad_norm": 0.7825105786323547,
      "learning_rate": 1.3853496115427305e-05,
      "loss": 0.0034,
      "step": 2770
    },
    {
      "epoch": 3.082085413200222,
      "grad_norm": 0.42062872648239136,
      "learning_rate": 1.3831298557158715e-05,
      "loss": 0.0062,
      "step": 2780
    },
    {
      "epoch": 3.093178036605657,
      "grad_norm": 0.7224815487861633,
      "learning_rate": 1.3809100998890123e-05,
      "loss": 0.0055,
      "step": 2790
    },
    {
      "epoch": 3.1042706600110925,
      "grad_norm": 0.04094177484512329,
      "learning_rate": 1.3786903440621533e-05,
      "loss": 0.002,
      "step": 2800
    },
    {
      "epoch": 3.115363283416528,
      "grad_norm": 0.03310002386569977,
      "learning_rate": 1.3764705882352943e-05,
      "loss": 0.0055,
      "step": 2810
    },
    {
      "epoch": 3.1264559068219633,
      "grad_norm": 0.5118809938430786,
      "learning_rate": 1.3742508324084351e-05,
      "loss": 0.0054,
      "step": 2820
    },
    {
      "epoch": 3.1375485302273987,
      "grad_norm": 0.5759711861610413,
      "learning_rate": 1.3720310765815761e-05,
      "loss": 0.0086,
      "step": 2830
    },
    {
      "epoch": 3.148641153632834,
      "grad_norm": 1.1310100555419922,
      "learning_rate": 1.3698113207547171e-05,
      "loss": 0.0091,
      "step": 2840
    },
    {
      "epoch": 3.1597337770382694,
      "grad_norm": 0.04265379533171654,
      "learning_rate": 1.3675915649278579e-05,
      "loss": 0.0061,
      "step": 2850
    },
    {
      "epoch": 3.170826400443705,
      "grad_norm": 0.6648514270782471,
      "learning_rate": 1.365371809100999e-05,
      "loss": 0.0035,
      "step": 2860
    },
    {
      "epoch": 3.18191902384914,
      "grad_norm": 0.02486938051879406,
      "learning_rate": 1.36315205327414e-05,
      "loss": 0.0055,
      "step": 2870
    },
    {
      "epoch": 3.1930116472545755,
      "grad_norm": 0.5384965538978577,
      "learning_rate": 1.3609322974472809e-05,
      "loss": 0.0029,
      "step": 2880
    },
    {
      "epoch": 3.204104270660011,
      "grad_norm": 0.6231627464294434,
      "learning_rate": 1.3587125416204219e-05,
      "loss": 0.005,
      "step": 2890
    },
    {
      "epoch": 3.2151968940654463,
      "grad_norm": 0.5190199613571167,
      "learning_rate": 1.3564927857935628e-05,
      "loss": 0.0083,
      "step": 2900
    },
    {
      "epoch": 3.2262895174708817,
      "grad_norm": 0.39337947964668274,
      "learning_rate": 1.3542730299667037e-05,
      "loss": 0.0081,
      "step": 2910
    },
    {
      "epoch": 3.237382140876317,
      "grad_norm": 0.6862775683403015,
      "learning_rate": 1.3520532741398446e-05,
      "loss": 0.0074,
      "step": 2920
    },
    {
      "epoch": 3.2484747642817524,
      "grad_norm": 1.0862469673156738,
      "learning_rate": 1.3498335183129856e-05,
      "loss": 0.0074,
      "step": 2930
    },
    {
      "epoch": 3.259567387687188,
      "grad_norm": 1.1846354007720947,
      "learning_rate": 1.3476137624861268e-05,
      "loss": 0.0075,
      "step": 2940
    },
    {
      "epoch": 3.270660011092623,
      "grad_norm": 1.2481039762496948,
      "learning_rate": 1.3453940066592676e-05,
      "loss": 0.003,
      "step": 2950
    },
    {
      "epoch": 3.281752634498059,
      "grad_norm": 0.3161996006965637,
      "learning_rate": 1.3431742508324086e-05,
      "loss": 0.0034,
      "step": 2960
    },
    {
      "epoch": 3.2928452579034944,
      "grad_norm": 0.6262934803962708,
      "learning_rate": 1.3409544950055496e-05,
      "loss": 0.0107,
      "step": 2970
    },
    {
      "epoch": 3.3039378813089297,
      "grad_norm": 0.9278308153152466,
      "learning_rate": 1.3387347391786904e-05,
      "loss": 0.0057,
      "step": 2980
    },
    {
      "epoch": 3.315030504714365,
      "grad_norm": 0.40893518924713135,
      "learning_rate": 1.3365149833518314e-05,
      "loss": 0.0041,
      "step": 2990
    },
    {
      "epoch": 3.3261231281198005,
      "grad_norm": 0.6295328736305237,
      "learning_rate": 1.3342952275249724e-05,
      "loss": 0.0032,
      "step": 3000
    },
    {
      "epoch": 3.337215751525236,
      "grad_norm": 0.039073534309864044,
      "learning_rate": 1.3320754716981132e-05,
      "loss": 0.0067,
      "step": 3010
    },
    {
      "epoch": 3.3483083749306712,
      "grad_norm": 0.5157305002212524,
      "learning_rate": 1.3298557158712542e-05,
      "loss": 0.0093,
      "step": 3020
    },
    {
      "epoch": 3.3594009983361066,
      "grad_norm": 0.9451714158058167,
      "learning_rate": 1.3276359600443953e-05,
      "loss": 0.0064,
      "step": 3030
    },
    {
      "epoch": 3.370493621741542,
      "grad_norm": 0.043243326246738434,
      "learning_rate": 1.3254162042175363e-05,
      "loss": 0.0062,
      "step": 3040
    },
    {
      "epoch": 3.3815862451469774,
      "grad_norm": 1.2751350402832031,
      "learning_rate": 1.3231964483906771e-05,
      "loss": 0.011,
      "step": 3050
    },
    {
      "epoch": 3.3926788685524127,
      "grad_norm": 1.1212610006332397,
      "learning_rate": 1.3209766925638181e-05,
      "loss": 0.008,
      "step": 3060
    },
    {
      "epoch": 3.403771491957848,
      "grad_norm": 0.3985590934753418,
      "learning_rate": 1.3187569367369591e-05,
      "loss": 0.0066,
      "step": 3070
    },
    {
      "epoch": 3.4148641153632835,
      "grad_norm": 0.6546558737754822,
      "learning_rate": 1.3165371809101e-05,
      "loss": 0.0115,
      "step": 3080
    },
    {
      "epoch": 3.425956738768719,
      "grad_norm": 1.0568732023239136,
      "learning_rate": 1.3143174250832409e-05,
      "loss": 0.0104,
      "step": 3090
    },
    {
      "epoch": 3.4370493621741542,
      "grad_norm": 0.9182056188583374,
      "learning_rate": 1.3120976692563819e-05,
      "loss": 0.0079,
      "step": 3100
    },
    {
      "epoch": 3.4481419855795896,
      "grad_norm": 1.700160026550293,
      "learning_rate": 1.3098779134295227e-05,
      "loss": 0.0061,
      "step": 3110
    },
    {
      "epoch": 3.459234608985025,
      "grad_norm": 0.09296636283397675,
      "learning_rate": 1.3076581576026639e-05,
      "loss": 0.0106,
      "step": 3120
    },
    {
      "epoch": 3.4703272323904604,
      "grad_norm": 0.8721600770950317,
      "learning_rate": 1.3054384017758049e-05,
      "loss": 0.0058,
      "step": 3130
    },
    {
      "epoch": 3.4814198557958957,
      "grad_norm": 0.5778414607048035,
      "learning_rate": 1.3032186459489457e-05,
      "loss": 0.0075,
      "step": 3140
    },
    {
      "epoch": 3.492512479201331,
      "grad_norm": 0.632172167301178,
      "learning_rate": 1.3009988901220867e-05,
      "loss": 0.0057,
      "step": 3150
    },
    {
      "epoch": 3.5036051026067665,
      "grad_norm": 0.49320322275161743,
      "learning_rate": 1.2987791342952276e-05,
      "loss": 0.0065,
      "step": 3160
    },
    {
      "epoch": 3.514697726012202,
      "grad_norm": 1.2454118728637695,
      "learning_rate": 1.2965593784683685e-05,
      "loss": 0.0049,
      "step": 3170
    },
    {
      "epoch": 3.5257903494176372,
      "grad_norm": 0.16420379281044006,
      "learning_rate": 1.2943396226415095e-05,
      "loss": 0.0042,
      "step": 3180
    },
    {
      "epoch": 3.5368829728230726,
      "grad_norm": 0.7556213736534119,
      "learning_rate": 1.2921198668146504e-05,
      "loss": 0.0076,
      "step": 3190
    },
    {
      "epoch": 3.547975596228508,
      "grad_norm": 0.7741811871528625,
      "learning_rate": 1.2899001109877916e-05,
      "loss": 0.0053,
      "step": 3200
    },
    {
      "epoch": 3.5590682196339434,
      "grad_norm": 0.3981403708457947,
      "learning_rate": 1.2876803551609324e-05,
      "loss": 0.0111,
      "step": 3210
    },
    {
      "epoch": 3.5701608430393788,
      "grad_norm": 1.556243658065796,
      "learning_rate": 1.2854605993340734e-05,
      "loss": 0.0082,
      "step": 3220
    },
    {
      "epoch": 3.581253466444814,
      "grad_norm": 0.24557125568389893,
      "learning_rate": 1.2832408435072144e-05,
      "loss": 0.0048,
      "step": 3230
    },
    {
      "epoch": 3.5923460898502495,
      "grad_norm": 1.2477893829345703,
      "learning_rate": 1.2810210876803552e-05,
      "loss": 0.0023,
      "step": 3240
    },
    {
      "epoch": 3.603438713255685,
      "grad_norm": 0.04675845056772232,
      "learning_rate": 1.2788013318534962e-05,
      "loss": 0.0041,
      "step": 3250
    },
    {
      "epoch": 3.6145313366611203,
      "grad_norm": 0.571173906326294,
      "learning_rate": 1.2765815760266372e-05,
      "loss": 0.0055,
      "step": 3260
    },
    {
      "epoch": 3.6256239600665556,
      "grad_norm": 0.5565969944000244,
      "learning_rate": 1.274361820199778e-05,
      "loss": 0.0092,
      "step": 3270
    },
    {
      "epoch": 3.636716583471991,
      "grad_norm": 0.013311430811882019,
      "learning_rate": 1.272142064372919e-05,
      "loss": 0.0071,
      "step": 3280
    },
    {
      "epoch": 3.6478092068774264,
      "grad_norm": 1.6714210510253906,
      "learning_rate": 1.2699223085460601e-05,
      "loss": 0.0069,
      "step": 3290
    },
    {
      "epoch": 3.6589018302828618,
      "grad_norm": 0.7213656306266785,
      "learning_rate": 1.2677025527192011e-05,
      "loss": 0.0053,
      "step": 3300
    },
    {
      "epoch": 3.669994453688297,
      "grad_norm": 0.9073006510734558,
      "learning_rate": 1.265482796892342e-05,
      "loss": 0.0101,
      "step": 3310
    },
    {
      "epoch": 3.6810870770937325,
      "grad_norm": 0.29895493388175964,
      "learning_rate": 1.263263041065483e-05,
      "loss": 0.0052,
      "step": 3320
    },
    {
      "epoch": 3.6921797004991683,
      "grad_norm": 1.0842756032943726,
      "learning_rate": 1.2610432852386239e-05,
      "loss": 0.008,
      "step": 3330
    },
    {
      "epoch": 3.7032723239046037,
      "grad_norm": 0.7172666192054749,
      "learning_rate": 1.2588235294117647e-05,
      "loss": 0.01,
      "step": 3340
    },
    {
      "epoch": 3.714364947310039,
      "grad_norm": 0.10475369542837143,
      "learning_rate": 1.2566037735849057e-05,
      "loss": 0.0063,
      "step": 3350
    },
    {
      "epoch": 3.7254575707154745,
      "grad_norm": 0.9639983177185059,
      "learning_rate": 1.2543840177580467e-05,
      "loss": 0.0059,
      "step": 3360
    },
    {
      "epoch": 3.73655019412091,
      "grad_norm": 0.9579765796661377,
      "learning_rate": 1.2521642619311875e-05,
      "loss": 0.0053,
      "step": 3370
    },
    {
      "epoch": 3.747642817526345,
      "grad_norm": 0.14046674966812134,
      "learning_rate": 1.2499445061043287e-05,
      "loss": 0.0081,
      "step": 3380
    },
    {
      "epoch": 3.7587354409317806,
      "grad_norm": 0.4136715233325958,
      "learning_rate": 1.2477247502774697e-05,
      "loss": 0.0101,
      "step": 3390
    },
    {
      "epoch": 3.769828064337216,
      "grad_norm": 0.30935949087142944,
      "learning_rate": 1.2455049944506107e-05,
      "loss": 0.0046,
      "step": 3400
    },
    {
      "epoch": 3.7809206877426513,
      "grad_norm": 0.25942277908325195,
      "learning_rate": 1.2432852386237515e-05,
      "loss": 0.006,
      "step": 3410
    },
    {
      "epoch": 3.7920133111480867,
      "grad_norm": 0.08538083732128143,
      "learning_rate": 1.2410654827968925e-05,
      "loss": 0.0038,
      "step": 3420
    },
    {
      "epoch": 3.803105934553522,
      "grad_norm": 0.11854628473520279,
      "learning_rate": 1.2388457269700334e-05,
      "loss": 0.0038,
      "step": 3430
    },
    {
      "epoch": 3.8141985579589575,
      "grad_norm": 0.2808975279331207,
      "learning_rate": 1.2366259711431743e-05,
      "loss": 0.0093,
      "step": 3440
    },
    {
      "epoch": 3.825291181364393,
      "grad_norm": 1.0285228490829468,
      "learning_rate": 1.2344062153163152e-05,
      "loss": 0.0101,
      "step": 3450
    },
    {
      "epoch": 3.836383804769828,
      "grad_norm": 0.43529564142227173,
      "learning_rate": 1.2321864594894564e-05,
      "loss": 0.0057,
      "step": 3460
    },
    {
      "epoch": 3.8474764281752636,
      "grad_norm": 2.0706746578216553,
      "learning_rate": 1.2299667036625972e-05,
      "loss": 0.0079,
      "step": 3470
    },
    {
      "epoch": 3.858569051580699,
      "grad_norm": 0.9805829524993896,
      "learning_rate": 1.2277469478357382e-05,
      "loss": 0.0116,
      "step": 3480
    },
    {
      "epoch": 3.8696616749861343,
      "grad_norm": 0.754189133644104,
      "learning_rate": 1.2255271920088792e-05,
      "loss": 0.0056,
      "step": 3490
    },
    {
      "epoch": 3.8807542983915697,
      "grad_norm": 0.9179038405418396,
      "learning_rate": 1.22330743618202e-05,
      "loss": 0.0079,
      "step": 3500
    },
    {
      "epoch": 3.891846921797005,
      "grad_norm": 1.5467790365219116,
      "learning_rate": 1.221087680355161e-05,
      "loss": 0.009,
      "step": 3510
    },
    {
      "epoch": 3.9029395452024405,
      "grad_norm": 0.44232311844825745,
      "learning_rate": 1.218867924528302e-05,
      "loss": 0.0072,
      "step": 3520
    },
    {
      "epoch": 3.914032168607876,
      "grad_norm": 0.05650203302502632,
      "learning_rate": 1.2166481687014428e-05,
      "loss": 0.0053,
      "step": 3530
    },
    {
      "epoch": 3.925124792013311,
      "grad_norm": 0.6348217129707336,
      "learning_rate": 1.2144284128745838e-05,
      "loss": 0.009,
      "step": 3540
    },
    {
      "epoch": 3.9362174154187466,
      "grad_norm": 1.0749729871749878,
      "learning_rate": 1.212208657047725e-05,
      "loss": 0.0086,
      "step": 3550
    },
    {
      "epoch": 3.947310038824182,
      "grad_norm": 2.320844888687134,
      "learning_rate": 1.209988901220866e-05,
      "loss": 0.0096,
      "step": 3560
    },
    {
      "epoch": 3.9584026622296173,
      "grad_norm": 0.5694814324378967,
      "learning_rate": 1.2077691453940067e-05,
      "loss": 0.008,
      "step": 3570
    },
    {
      "epoch": 3.9694952856350527,
      "grad_norm": 0.9794883728027344,
      "learning_rate": 1.2055493895671477e-05,
      "loss": 0.0045,
      "step": 3580
    },
    {
      "epoch": 3.980587909040488,
      "grad_norm": 0.2133321315050125,
      "learning_rate": 1.2033296337402887e-05,
      "loss": 0.0054,
      "step": 3590
    },
    {
      "epoch": 3.9916805324459235,
      "grad_norm": 1.423915147781372,
      "learning_rate": 1.2011098779134295e-05,
      "loss": 0.0127,
      "step": 3600
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8699763593380615,
      "eval_f1": 0.6728971962616822,
      "eval_loss": 0.4916483163833618,
      "eval_precision": 0.6449511400651465,
      "eval_recall": 0.7033747779751333,
      "eval_runtime": 2.7388,
      "eval_samples_per_second": 360.381,
      "eval_steps_per_second": 11.319,
      "step": 3608
    },
    {
      "epoch": 4.002218524681087,
      "grad_norm": 0.7580460906028748,
      "learning_rate": 1.1988901220865705e-05,
      "loss": 0.0052,
      "step": 3610
    },
    {
      "epoch": 4.0133111480865225,
      "grad_norm": 0.13793574273586273,
      "learning_rate": 1.1966703662597115e-05,
      "loss": 0.0024,
      "step": 3620
    },
    {
      "epoch": 4.024403771491958,
      "grad_norm": 0.41575559973716736,
      "learning_rate": 1.1944506104328523e-05,
      "loss": 0.0028,
      "step": 3630
    },
    {
      "epoch": 4.035496394897393,
      "grad_norm": 0.268417626619339,
      "learning_rate": 1.1922308546059935e-05,
      "loss": 0.0043,
      "step": 3640
    },
    {
      "epoch": 4.046589018302829,
      "grad_norm": 0.227146178483963,
      "learning_rate": 1.1900110987791345e-05,
      "loss": 0.0014,
      "step": 3650
    },
    {
      "epoch": 4.057681641708264,
      "grad_norm": 0.1434301882982254,
      "learning_rate": 1.1877913429522755e-05,
      "loss": 0.0012,
      "step": 3660
    },
    {
      "epoch": 4.068774265113699,
      "grad_norm": 0.3861040472984314,
      "learning_rate": 1.1855715871254163e-05,
      "loss": 0.0027,
      "step": 3670
    },
    {
      "epoch": 4.079866888519135,
      "grad_norm": 0.2243143618106842,
      "learning_rate": 1.1833518312985573e-05,
      "loss": 0.0062,
      "step": 3680
    },
    {
      "epoch": 4.09095951192457,
      "grad_norm": 0.5330334901809692,
      "learning_rate": 1.1811320754716982e-05,
      "loss": 0.0074,
      "step": 3690
    },
    {
      "epoch": 4.1020521353300055,
      "grad_norm": 0.3948166072368622,
      "learning_rate": 1.178912319644839e-05,
      "loss": 0.0047,
      "step": 3700
    },
    {
      "epoch": 4.113144758735441,
      "grad_norm": 0.8226463198661804,
      "learning_rate": 1.17669256381798e-05,
      "loss": 0.0047,
      "step": 3710
    },
    {
      "epoch": 4.124237382140876,
      "grad_norm": 0.5453665852546692,
      "learning_rate": 1.1744728079911212e-05,
      "loss": 0.003,
      "step": 3720
    },
    {
      "epoch": 4.135330005546312,
      "grad_norm": 0.1433408111333847,
      "learning_rate": 1.172253052164262e-05,
      "loss": 0.0029,
      "step": 3730
    },
    {
      "epoch": 4.146422628951747,
      "grad_norm": 0.14077675342559814,
      "learning_rate": 1.170033296337403e-05,
      "loss": 0.004,
      "step": 3740
    },
    {
      "epoch": 4.157515252357182,
      "grad_norm": 0.08359857648611069,
      "learning_rate": 1.167813540510544e-05,
      "loss": 0.0034,
      "step": 3750
    },
    {
      "epoch": 4.168607875762618,
      "grad_norm": 0.10568249225616455,
      "learning_rate": 1.1655937846836848e-05,
      "loss": 0.0019,
      "step": 3760
    },
    {
      "epoch": 4.179700499168053,
      "grad_norm": 0.010569937527179718,
      "learning_rate": 1.1633740288568258e-05,
      "loss": 0.0039,
      "step": 3770
    },
    {
      "epoch": 4.1907931225734885,
      "grad_norm": 0.38508471846580505,
      "learning_rate": 1.1611542730299668e-05,
      "loss": 0.0035,
      "step": 3780
    },
    {
      "epoch": 4.201885745978924,
      "grad_norm": 0.4583463966846466,
      "learning_rate": 1.1589345172031076e-05,
      "loss": 0.0035,
      "step": 3790
    },
    {
      "epoch": 4.212978369384359,
      "grad_norm": 0.26573556661605835,
      "learning_rate": 1.1567147613762486e-05,
      "loss": 0.0014,
      "step": 3800
    },
    {
      "epoch": 4.224070992789795,
      "grad_norm": 0.699878454208374,
      "learning_rate": 1.1544950055493897e-05,
      "loss": 0.0034,
      "step": 3810
    },
    {
      "epoch": 4.23516361619523,
      "grad_norm": 0.02065327949821949,
      "learning_rate": 1.1522752497225307e-05,
      "loss": 0.0028,
      "step": 3820
    },
    {
      "epoch": 4.246256239600665,
      "grad_norm": 1.4631462097167969,
      "learning_rate": 1.1500554938956716e-05,
      "loss": 0.003,
      "step": 3830
    },
    {
      "epoch": 4.257348863006101,
      "grad_norm": 1.0686756372451782,
      "learning_rate": 1.1478357380688125e-05,
      "loss": 0.0026,
      "step": 3840
    },
    {
      "epoch": 4.268441486411536,
      "grad_norm": 0.15860460698604584,
      "learning_rate": 1.1456159822419535e-05,
      "loss": 0.0064,
      "step": 3850
    },
    {
      "epoch": 4.2795341098169715,
      "grad_norm": 0.05696827918291092,
      "learning_rate": 1.1433962264150943e-05,
      "loss": 0.0033,
      "step": 3860
    },
    {
      "epoch": 4.290626733222407,
      "grad_norm": 0.01276635192334652,
      "learning_rate": 1.1411764705882353e-05,
      "loss": 0.0059,
      "step": 3870
    },
    {
      "epoch": 4.301719356627842,
      "grad_norm": 0.3438052535057068,
      "learning_rate": 1.1389567147613763e-05,
      "loss": 0.0022,
      "step": 3880
    },
    {
      "epoch": 4.312811980033278,
      "grad_norm": 0.020309099927544594,
      "learning_rate": 1.1367369589345171e-05,
      "loss": 0.0029,
      "step": 3890
    },
    {
      "epoch": 4.323904603438713,
      "grad_norm": 1.0980569124221802,
      "learning_rate": 1.1345172031076583e-05,
      "loss": 0.0012,
      "step": 3900
    },
    {
      "epoch": 4.334997226844148,
      "grad_norm": 0.723771333694458,
      "learning_rate": 1.1322974472807993e-05,
      "loss": 0.0027,
      "step": 3910
    },
    {
      "epoch": 4.346089850249584,
      "grad_norm": 0.009631435386836529,
      "learning_rate": 1.1300776914539403e-05,
      "loss": 0.0023,
      "step": 3920
    },
    {
      "epoch": 4.357182473655019,
      "grad_norm": 0.29609665274620056,
      "learning_rate": 1.127857935627081e-05,
      "loss": 0.0044,
      "step": 3930
    },
    {
      "epoch": 4.3682750970604545,
      "grad_norm": 0.06304040551185608,
      "learning_rate": 1.125638179800222e-05,
      "loss": 0.0024,
      "step": 3940
    },
    {
      "epoch": 4.37936772046589,
      "grad_norm": 0.2765149176120758,
      "learning_rate": 1.123418423973363e-05,
      "loss": 0.003,
      "step": 3950
    },
    {
      "epoch": 4.390460343871325,
      "grad_norm": 0.628939688205719,
      "learning_rate": 1.1211986681465039e-05,
      "loss": 0.0019,
      "step": 3960
    },
    {
      "epoch": 4.401552967276761,
      "grad_norm": 0.5144364833831787,
      "learning_rate": 1.1189789123196449e-05,
      "loss": 0.0038,
      "step": 3970
    },
    {
      "epoch": 4.412645590682196,
      "grad_norm": 0.02215411141514778,
      "learning_rate": 1.116759156492786e-05,
      "loss": 0.0052,
      "step": 3980
    },
    {
      "epoch": 4.423738214087631,
      "grad_norm": 0.8358492851257324,
      "learning_rate": 1.114539400665927e-05,
      "loss": 0.0053,
      "step": 3990
    },
    {
      "epoch": 4.434830837493067,
      "grad_norm": 0.49006780982017517,
      "learning_rate": 1.1123196448390678e-05,
      "loss": 0.0029,
      "step": 4000
    },
    {
      "epoch": 4.445923460898502,
      "grad_norm": 0.08062740415334702,
      "learning_rate": 1.1100998890122088e-05,
      "loss": 0.0095,
      "step": 4010
    },
    {
      "epoch": 4.4570160843039375,
      "grad_norm": 0.931599497795105,
      "learning_rate": 1.1078801331853496e-05,
      "loss": 0.009,
      "step": 4020
    },
    {
      "epoch": 4.468108707709373,
      "grad_norm": 1.4779237508773804,
      "learning_rate": 1.1056603773584906e-05,
      "loss": 0.0075,
      "step": 4030
    },
    {
      "epoch": 4.479201331114808,
      "grad_norm": 0.0675542801618576,
      "learning_rate": 1.1034406215316316e-05,
      "loss": 0.0029,
      "step": 4040
    },
    {
      "epoch": 4.490293954520244,
      "grad_norm": 1.2957643270492554,
      "learning_rate": 1.1012208657047724e-05,
      "loss": 0.0079,
      "step": 4050
    },
    {
      "epoch": 4.501386577925679,
      "grad_norm": 1.259100079536438,
      "learning_rate": 1.0990011098779134e-05,
      "loss": 0.0045,
      "step": 4060
    },
    {
      "epoch": 4.512479201331114,
      "grad_norm": 0.4716441333293915,
      "learning_rate": 1.0967813540510546e-05,
      "loss": 0.0022,
      "step": 4070
    },
    {
      "epoch": 4.52357182473655,
      "grad_norm": 0.12484975159168243,
      "learning_rate": 1.0945615982241955e-05,
      "loss": 0.0031,
      "step": 4080
    },
    {
      "epoch": 4.534664448141985,
      "grad_norm": 0.07594279199838638,
      "learning_rate": 1.0923418423973364e-05,
      "loss": 0.0027,
      "step": 4090
    },
    {
      "epoch": 4.5457570715474205,
      "grad_norm": 0.3258889317512512,
      "learning_rate": 1.0901220865704773e-05,
      "loss": 0.0051,
      "step": 4100
    },
    {
      "epoch": 4.556849694952856,
      "grad_norm": 0.4499701261520386,
      "learning_rate": 1.0879023307436183e-05,
      "loss": 0.0016,
      "step": 4110
    },
    {
      "epoch": 4.567942318358292,
      "grad_norm": 0.15266552567481995,
      "learning_rate": 1.0856825749167591e-05,
      "loss": 0.0026,
      "step": 4120
    },
    {
      "epoch": 4.5790349417637275,
      "grad_norm": 0.07247844338417053,
      "learning_rate": 1.0834628190899001e-05,
      "loss": 0.0021,
      "step": 4130
    },
    {
      "epoch": 4.590127565169163,
      "grad_norm": 1.395110845565796,
      "learning_rate": 1.0812430632630411e-05,
      "loss": 0.0033,
      "step": 4140
    },
    {
      "epoch": 4.601220188574598,
      "grad_norm": 1.36636221408844,
      "learning_rate": 1.079023307436182e-05,
      "loss": 0.003,
      "step": 4150
    },
    {
      "epoch": 4.612312811980034,
      "grad_norm": 1.0960825681686401,
      "learning_rate": 1.0768035516093231e-05,
      "loss": 0.0011,
      "step": 4160
    },
    {
      "epoch": 4.623405435385469,
      "grad_norm": 1.0867245197296143,
      "learning_rate": 1.074583795782464e-05,
      "loss": 0.0045,
      "step": 4170
    },
    {
      "epoch": 4.634498058790904,
      "grad_norm": 0.01242793258279562,
      "learning_rate": 1.072364039955605e-05,
      "loss": 0.0034,
      "step": 4180
    },
    {
      "epoch": 4.64559068219634,
      "grad_norm": 0.0266682468354702,
      "learning_rate": 1.0701442841287459e-05,
      "loss": 0.0026,
      "step": 4190
    },
    {
      "epoch": 4.656683305601775,
      "grad_norm": 1.711166262626648,
      "learning_rate": 1.0679245283018869e-05,
      "loss": 0.0032,
      "step": 4200
    },
    {
      "epoch": 4.6677759290072105,
      "grad_norm": 1.225320816040039,
      "learning_rate": 1.0657047724750279e-05,
      "loss": 0.0049,
      "step": 4210
    },
    {
      "epoch": 4.678868552412646,
      "grad_norm": 0.05706857517361641,
      "learning_rate": 1.0634850166481687e-05,
      "loss": 0.0024,
      "step": 4220
    },
    {
      "epoch": 4.689961175818081,
      "grad_norm": 0.08420650660991669,
      "learning_rate": 1.0612652608213097e-05,
      "loss": 0.006,
      "step": 4230
    },
    {
      "epoch": 4.701053799223517,
      "grad_norm": 0.014957012608647346,
      "learning_rate": 1.0590455049944508e-05,
      "loss": 0.0033,
      "step": 4240
    },
    {
      "epoch": 4.712146422628952,
      "grad_norm": 0.2216530293226242,
      "learning_rate": 1.0568257491675918e-05,
      "loss": 0.0039,
      "step": 4250
    },
    {
      "epoch": 4.723239046034387,
      "grad_norm": 0.1360142081975937,
      "learning_rate": 1.0546059933407326e-05,
      "loss": 0.0018,
      "step": 4260
    },
    {
      "epoch": 4.734331669439823,
      "grad_norm": 0.004492219537496567,
      "learning_rate": 1.0523862375138736e-05,
      "loss": 0.0016,
      "step": 4270
    },
    {
      "epoch": 4.745424292845258,
      "grad_norm": 0.0539228618144989,
      "learning_rate": 1.0501664816870146e-05,
      "loss": 0.0022,
      "step": 4280
    },
    {
      "epoch": 4.7565169162506935,
      "grad_norm": 0.35766854882240295,
      "learning_rate": 1.0479467258601554e-05,
      "loss": 0.0022,
      "step": 4290
    },
    {
      "epoch": 4.767609539656129,
      "grad_norm": 0.3328166902065277,
      "learning_rate": 1.0457269700332964e-05,
      "loss": 0.0062,
      "step": 4300
    },
    {
      "epoch": 4.778702163061564,
      "grad_norm": 0.08985334634780884,
      "learning_rate": 1.0435072142064372e-05,
      "loss": 0.0013,
      "step": 4310
    },
    {
      "epoch": 4.789794786467,
      "grad_norm": 0.010276420041918755,
      "learning_rate": 1.0412874583795782e-05,
      "loss": 0.0031,
      "step": 4320
    },
    {
      "epoch": 4.800887409872435,
      "grad_norm": 0.5527135133743286,
      "learning_rate": 1.0390677025527194e-05,
      "loss": 0.004,
      "step": 4330
    },
    {
      "epoch": 4.81198003327787,
      "grad_norm": 0.21018469333648682,
      "learning_rate": 1.0368479467258603e-05,
      "loss": 0.0021,
      "step": 4340
    },
    {
      "epoch": 4.823072656683306,
      "grad_norm": 0.10334997624158859,
      "learning_rate": 1.0346281908990012e-05,
      "loss": 0.0045,
      "step": 4350
    },
    {
      "epoch": 4.834165280088741,
      "grad_norm": 0.09865798056125641,
      "learning_rate": 1.0324084350721422e-05,
      "loss": 0.0008,
      "step": 4360
    },
    {
      "epoch": 4.8452579034941765,
      "grad_norm": 0.24930772185325623,
      "learning_rate": 1.0301886792452831e-05,
      "loss": 0.0021,
      "step": 4370
    },
    {
      "epoch": 4.856350526899612,
      "grad_norm": 0.7967151999473572,
      "learning_rate": 1.027968923418424e-05,
      "loss": 0.0027,
      "step": 4380
    },
    {
      "epoch": 4.867443150305047,
      "grad_norm": 0.2117662876844406,
      "learning_rate": 1.025749167591565e-05,
      "loss": 0.0061,
      "step": 4390
    },
    {
      "epoch": 4.878535773710483,
      "grad_norm": 0.38198602199554443,
      "learning_rate": 1.023529411764706e-05,
      "loss": 0.0032,
      "step": 4400
    },
    {
      "epoch": 4.889628397115918,
      "grad_norm": 0.8646931648254395,
      "learning_rate": 1.0213096559378467e-05,
      "loss": 0.0074,
      "step": 4410
    },
    {
      "epoch": 4.900721020521353,
      "grad_norm": 0.5245551466941833,
      "learning_rate": 1.0190899001109879e-05,
      "loss": 0.0029,
      "step": 4420
    },
    {
      "epoch": 4.911813643926789,
      "grad_norm": 0.368184894323349,
      "learning_rate": 1.0168701442841289e-05,
      "loss": 0.0056,
      "step": 4430
    },
    {
      "epoch": 4.922906267332224,
      "grad_norm": 1.3652911186218262,
      "learning_rate": 1.0146503884572699e-05,
      "loss": 0.0022,
      "step": 4440
    },
    {
      "epoch": 4.9339988907376595,
      "grad_norm": 2.6624059677124023,
      "learning_rate": 1.0124306326304107e-05,
      "loss": 0.0059,
      "step": 4450
    },
    {
      "epoch": 4.945091514143095,
      "grad_norm": 0.10531987994909286,
      "learning_rate": 1.0102108768035517e-05,
      "loss": 0.0017,
      "step": 4460
    },
    {
      "epoch": 4.95618413754853,
      "grad_norm": 0.7769778966903687,
      "learning_rate": 1.0079911209766927e-05,
      "loss": 0.0054,
      "step": 4470
    },
    {
      "epoch": 4.967276760953966,
      "grad_norm": 0.016158653423190117,
      "learning_rate": 1.0057713651498335e-05,
      "loss": 0.002,
      "step": 4480
    },
    {
      "epoch": 4.978369384359401,
      "grad_norm": 0.09255735576152802,
      "learning_rate": 1.0035516093229745e-05,
      "loss": 0.005,
      "step": 4490
    },
    {
      "epoch": 4.989462007764836,
      "grad_norm": 0.7796350717544556,
      "learning_rate": 1.0013318534961156e-05,
      "loss": 0.0025,
      "step": 4500
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.36661309003829956,
      "learning_rate": 9.991120976692564e-06,
      "loss": 0.0011,
      "step": 4510
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8743667679837892,
      "eval_f1": 0.7009646302250804,
      "eval_loss": 0.5751733183860779,
      "eval_precision": 0.6402349486049926,
      "eval_recall": 0.7744227353463587,
      "eval_runtime": 2.6897,
      "eval_samples_per_second": 366.959,
      "eval_steps_per_second": 11.526,
      "step": 4510
    },
    {
      "epoch": 5.011092623405435,
      "grad_norm": 0.05989912524819374,
      "learning_rate": 9.968923418423974e-06,
      "loss": 0.0007,
      "step": 4520
    },
    {
      "epoch": 5.022185246810871,
      "grad_norm": 0.05131436884403229,
      "learning_rate": 9.946725860155384e-06,
      "loss": 0.0012,
      "step": 4530
    },
    {
      "epoch": 5.033277870216306,
      "grad_norm": 0.3775066137313843,
      "learning_rate": 9.924528301886794e-06,
      "loss": 0.001,
      "step": 4540
    },
    {
      "epoch": 5.0443704936217415,
      "grad_norm": 0.025283826515078545,
      "learning_rate": 9.902330743618202e-06,
      "loss": 0.0008,
      "step": 4550
    },
    {
      "epoch": 5.055463117027177,
      "grad_norm": 0.19627916812896729,
      "learning_rate": 9.880133185349612e-06,
      "loss": 0.0007,
      "step": 4560
    },
    {
      "epoch": 5.066555740432612,
      "grad_norm": 0.009000764228403568,
      "learning_rate": 9.857935627081022e-06,
      "loss": 0.001,
      "step": 4570
    },
    {
      "epoch": 5.077648363838048,
      "grad_norm": 0.031089652329683304,
      "learning_rate": 9.835738068812432e-06,
      "loss": 0.0011,
      "step": 4580
    },
    {
      "epoch": 5.088740987243483,
      "grad_norm": 1.2362189292907715,
      "learning_rate": 9.813540510543842e-06,
      "loss": 0.0015,
      "step": 4590
    },
    {
      "epoch": 5.099833610648918,
      "grad_norm": 0.5052623748779297,
      "learning_rate": 9.79134295227525e-06,
      "loss": 0.0042,
      "step": 4600
    },
    {
      "epoch": 5.110926234054354,
      "grad_norm": 1.2959685325622559,
      "learning_rate": 9.76914539400666e-06,
      "loss": 0.004,
      "step": 4610
    },
    {
      "epoch": 5.122018857459789,
      "grad_norm": 0.010257720947265625,
      "learning_rate": 9.74694783573807e-06,
      "loss": 0.0018,
      "step": 4620
    },
    {
      "epoch": 5.1331114808652245,
      "grad_norm": 0.8927929401397705,
      "learning_rate": 9.72475027746948e-06,
      "loss": 0.0032,
      "step": 4630
    },
    {
      "epoch": 5.14420410427066,
      "grad_norm": 0.6881095767021179,
      "learning_rate": 9.702552719200888e-06,
      "loss": 0.0018,
      "step": 4640
    },
    {
      "epoch": 5.155296727676095,
      "grad_norm": 0.6798466444015503,
      "learning_rate": 9.6803551609323e-06,
      "loss": 0.0024,
      "step": 4650
    },
    {
      "epoch": 5.166389351081531,
      "grad_norm": 0.09100968390703201,
      "learning_rate": 9.658157602663707e-06,
      "loss": 0.0009,
      "step": 4660
    },
    {
      "epoch": 5.177481974486966,
      "grad_norm": 0.38566479086875916,
      "learning_rate": 9.635960044395117e-06,
      "loss": 0.0011,
      "step": 4670
    },
    {
      "epoch": 5.188574597892401,
      "grad_norm": 0.5521411895751953,
      "learning_rate": 9.613762486126527e-06,
      "loss": 0.0017,
      "step": 4680
    },
    {
      "epoch": 5.199667221297837,
      "grad_norm": 0.20503102242946625,
      "learning_rate": 9.591564927857935e-06,
      "loss": 0.0004,
      "step": 4690
    },
    {
      "epoch": 5.210759844703272,
      "grad_norm": 0.02050948329269886,
      "learning_rate": 9.569367369589347e-06,
      "loss": 0.0012,
      "step": 4700
    },
    {
      "epoch": 5.2218524681087075,
      "grad_norm": 0.013379433192312717,
      "learning_rate": 9.547169811320755e-06,
      "loss": 0.0027,
      "step": 4710
    },
    {
      "epoch": 5.232945091514143,
      "grad_norm": 0.004559786058962345,
      "learning_rate": 9.524972253052165e-06,
      "loss": 0.0016,
      "step": 4720
    },
    {
      "epoch": 5.244037714919578,
      "grad_norm": 0.3056652843952179,
      "learning_rate": 9.502774694783575e-06,
      "loss": 0.0009,
      "step": 4730
    },
    {
      "epoch": 5.255130338325014,
      "grad_norm": 0.024444641545414925,
      "learning_rate": 9.480577136514985e-06,
      "loss": 0.0027,
      "step": 4740
    },
    {
      "epoch": 5.266222961730449,
      "grad_norm": 0.01696205884218216,
      "learning_rate": 9.458379578246394e-06,
      "loss": 0.002,
      "step": 4750
    },
    {
      "epoch": 5.277315585135884,
      "grad_norm": 0.009448543190956116,
      "learning_rate": 9.436182019977803e-06,
      "loss": 0.0003,
      "step": 4760
    },
    {
      "epoch": 5.28840820854132,
      "grad_norm": 0.6901306509971619,
      "learning_rate": 9.413984461709212e-06,
      "loss": 0.0008,
      "step": 4770
    },
    {
      "epoch": 5.299500831946755,
      "grad_norm": 0.9955122470855713,
      "learning_rate": 9.391786903440622e-06,
      "loss": 0.0014,
      "step": 4780
    },
    {
      "epoch": 5.3105934553521905,
      "grad_norm": 0.011066406965255737,
      "learning_rate": 9.369589345172032e-06,
      "loss": 0.0037,
      "step": 4790
    },
    {
      "epoch": 5.321686078757626,
      "grad_norm": 1.1432924270629883,
      "learning_rate": 9.347391786903442e-06,
      "loss": 0.0013,
      "step": 4800
    },
    {
      "epoch": 5.332778702163061,
      "grad_norm": 0.2927857041358948,
      "learning_rate": 9.32519422863485e-06,
      "loss": 0.001,
      "step": 4810
    },
    {
      "epoch": 5.343871325568497,
      "grad_norm": 0.05256081372499466,
      "learning_rate": 9.30299667036626e-06,
      "loss": 0.0042,
      "step": 4820
    },
    {
      "epoch": 5.354963948973932,
      "grad_norm": 0.10562248528003693,
      "learning_rate": 9.28079911209767e-06,
      "loss": 0.002,
      "step": 4830
    },
    {
      "epoch": 5.366056572379367,
      "grad_norm": 0.12000148743391037,
      "learning_rate": 9.25860155382908e-06,
      "loss": 0.0008,
      "step": 4840
    },
    {
      "epoch": 5.377149195784803,
      "grad_norm": 0.6670930981636047,
      "learning_rate": 9.23640399556049e-06,
      "loss": 0.0029,
      "step": 4850
    },
    {
      "epoch": 5.388241819190238,
      "grad_norm": 0.11181971430778503,
      "learning_rate": 9.214206437291898e-06,
      "loss": 0.0059,
      "step": 4860
    },
    {
      "epoch": 5.3993344425956735,
      "grad_norm": 1.371401309967041,
      "learning_rate": 9.19200887902331e-06,
      "loss": 0.0033,
      "step": 4870
    },
    {
      "epoch": 5.410427066001109,
      "grad_norm": 0.047688670456409454,
      "learning_rate": 9.169811320754718e-06,
      "loss": 0.001,
      "step": 4880
    },
    {
      "epoch": 5.421519689406544,
      "grad_norm": 0.5221829414367676,
      "learning_rate": 9.147613762486127e-06,
      "loss": 0.0028,
      "step": 4890
    },
    {
      "epoch": 5.43261231281198,
      "grad_norm": 0.009564414620399475,
      "learning_rate": 9.125416204217536e-06,
      "loss": 0.0032,
      "step": 4900
    },
    {
      "epoch": 5.443704936217415,
      "grad_norm": 0.8203772902488708,
      "learning_rate": 9.103218645948947e-06,
      "loss": 0.0015,
      "step": 4910
    },
    {
      "epoch": 5.45479755962285,
      "grad_norm": 0.756874680519104,
      "learning_rate": 9.081021087680355e-06,
      "loss": 0.0017,
      "step": 4920
    },
    {
      "epoch": 5.465890183028286,
      "grad_norm": 0.04516858235001564,
      "learning_rate": 9.058823529411765e-06,
      "loss": 0.0017,
      "step": 4930
    },
    {
      "epoch": 5.476982806433721,
      "grad_norm": 0.05763641372323036,
      "learning_rate": 9.036625971143175e-06,
      "loss": 0.0023,
      "step": 4940
    },
    {
      "epoch": 5.4880754298391565,
      "grad_norm": 0.06448545306921005,
      "learning_rate": 9.014428412874583e-06,
      "loss": 0.0027,
      "step": 4950
    },
    {
      "epoch": 5.499168053244592,
      "grad_norm": 0.032656289637088776,
      "learning_rate": 8.992230854605995e-06,
      "loss": 0.0003,
      "step": 4960
    },
    {
      "epoch": 5.510260676650027,
      "grad_norm": 0.48547258973121643,
      "learning_rate": 8.970033296337403e-06,
      "loss": 0.004,
      "step": 4970
    },
    {
      "epoch": 5.521353300055463,
      "grad_norm": 0.018192799761891365,
      "learning_rate": 8.947835738068813e-06,
      "loss": 0.0009,
      "step": 4980
    },
    {
      "epoch": 5.532445923460898,
      "grad_norm": 0.8040679097175598,
      "learning_rate": 8.925638179800223e-06,
      "loss": 0.0015,
      "step": 4990
    },
    {
      "epoch": 5.543538546866333,
      "grad_norm": 0.035649724304676056,
      "learning_rate": 8.903440621531633e-06,
      "loss": 0.0006,
      "step": 5000
    },
    {
      "epoch": 5.554631170271769,
      "grad_norm": 0.04822482913732529,
      "learning_rate": 8.881243063263043e-06,
      "loss": 0.0018,
      "step": 5010
    },
    {
      "epoch": 5.565723793677205,
      "grad_norm": 0.20339788496494293,
      "learning_rate": 8.85904550499445e-06,
      "loss": 0.0037,
      "step": 5020
    },
    {
      "epoch": 5.57681641708264,
      "grad_norm": 0.10009117424488068,
      "learning_rate": 8.83684794672586e-06,
      "loss": 0.0023,
      "step": 5030
    },
    {
      "epoch": 5.587909040488076,
      "grad_norm": 0.0017753434367477894,
      "learning_rate": 8.81465038845727e-06,
      "loss": 0.0012,
      "step": 5040
    },
    {
      "epoch": 5.599001663893511,
      "grad_norm": 0.04721298813819885,
      "learning_rate": 8.79245283018868e-06,
      "loss": 0.003,
      "step": 5050
    },
    {
      "epoch": 5.6100942872989465,
      "grad_norm": 0.32094866037368774,
      "learning_rate": 8.77025527192009e-06,
      "loss": 0.0058,
      "step": 5060
    },
    {
      "epoch": 5.621186910704382,
      "grad_norm": 0.009650022722780704,
      "learning_rate": 8.748057713651498e-06,
      "loss": 0.002,
      "step": 5070
    },
    {
      "epoch": 5.632279534109817,
      "grad_norm": 1.6382402181625366,
      "learning_rate": 8.725860155382908e-06,
      "loss": 0.0042,
      "step": 5080
    },
    {
      "epoch": 5.643372157515253,
      "grad_norm": 0.011900817044079304,
      "learning_rate": 8.703662597114318e-06,
      "loss": 0.0021,
      "step": 5090
    },
    {
      "epoch": 5.654464780920688,
      "grad_norm": 0.009083210490643978,
      "learning_rate": 8.681465038845728e-06,
      "loss": 0.0023,
      "step": 5100
    },
    {
      "epoch": 5.665557404326123,
      "grad_norm": 0.9369150400161743,
      "learning_rate": 8.659267480577138e-06,
      "loss": 0.0026,
      "step": 5110
    },
    {
      "epoch": 5.676650027731559,
      "grad_norm": 0.8348691463470459,
      "learning_rate": 8.637069922308546e-06,
      "loss": 0.0016,
      "step": 5120
    },
    {
      "epoch": 5.687742651136994,
      "grad_norm": 0.0950978547334671,
      "learning_rate": 8.614872364039958e-06,
      "loss": 0.0044,
      "step": 5130
    },
    {
      "epoch": 5.6988352745424296,
      "grad_norm": 0.5261467099189758,
      "learning_rate": 8.592674805771366e-06,
      "loss": 0.0015,
      "step": 5140
    },
    {
      "epoch": 5.709927897947865,
      "grad_norm": 0.01738225668668747,
      "learning_rate": 8.570477247502776e-06,
      "loss": 0.0057,
      "step": 5150
    },
    {
      "epoch": 5.7210205213533,
      "grad_norm": 0.40831008553504944,
      "learning_rate": 8.548279689234185e-06,
      "loss": 0.0056,
      "step": 5160
    },
    {
      "epoch": 5.732113144758736,
      "grad_norm": 0.3201698064804077,
      "learning_rate": 8.526082130965595e-06,
      "loss": 0.0023,
      "step": 5170
    },
    {
      "epoch": 5.743205768164171,
      "grad_norm": 0.01956314779818058,
      "learning_rate": 8.503884572697003e-06,
      "loss": 0.0011,
      "step": 5180
    },
    {
      "epoch": 5.754298391569606,
      "grad_norm": 0.091837577521801,
      "learning_rate": 8.481687014428413e-06,
      "loss": 0.0025,
      "step": 5190
    },
    {
      "epoch": 5.765391014975042,
      "grad_norm": 0.09794759005308151,
      "learning_rate": 8.459489456159823e-06,
      "loss": 0.0063,
      "step": 5200
    },
    {
      "epoch": 5.776483638380477,
      "grad_norm": 0.6827769875526428,
      "learning_rate": 8.437291897891231e-06,
      "loss": 0.0027,
      "step": 5210
    },
    {
      "epoch": 5.787576261785913,
      "grad_norm": 0.7685621976852417,
      "learning_rate": 8.415094339622643e-06,
      "loss": 0.0012,
      "step": 5220
    },
    {
      "epoch": 5.798668885191348,
      "grad_norm": 0.20335263013839722,
      "learning_rate": 8.392896781354051e-06,
      "loss": 0.0016,
      "step": 5230
    },
    {
      "epoch": 5.809761508596783,
      "grad_norm": 0.007598708849400282,
      "learning_rate": 8.370699223085461e-06,
      "loss": 0.0028,
      "step": 5240
    },
    {
      "epoch": 5.820854132002219,
      "grad_norm": 0.46612560749053955,
      "learning_rate": 8.34850166481687e-06,
      "loss": 0.0022,
      "step": 5250
    },
    {
      "epoch": 5.831946755407654,
      "grad_norm": 0.08132003247737885,
      "learning_rate": 8.32630410654828e-06,
      "loss": 0.001,
      "step": 5260
    },
    {
      "epoch": 5.843039378813089,
      "grad_norm": 1.4980528354644775,
      "learning_rate": 8.30410654827969e-06,
      "loss": 0.0028,
      "step": 5270
    },
    {
      "epoch": 5.854132002218525,
      "grad_norm": 0.4937552213668823,
      "learning_rate": 8.281908990011099e-06,
      "loss": 0.0008,
      "step": 5280
    },
    {
      "epoch": 5.86522462562396,
      "grad_norm": 0.02265891060233116,
      "learning_rate": 8.259711431742509e-06,
      "loss": 0.0018,
      "step": 5290
    },
    {
      "epoch": 5.876317249029396,
      "grad_norm": 0.02029411308467388,
      "learning_rate": 8.237513873473918e-06,
      "loss": 0.0019,
      "step": 5300
    },
    {
      "epoch": 5.887409872434831,
      "grad_norm": 0.4853488504886627,
      "learning_rate": 8.215316315205328e-06,
      "loss": 0.0007,
      "step": 5310
    },
    {
      "epoch": 5.898502495840266,
      "grad_norm": 0.6986839175224304,
      "learning_rate": 8.193118756936738e-06,
      "loss": 0.0021,
      "step": 5320
    },
    {
      "epoch": 5.909595119245702,
      "grad_norm": 0.15741075575351715,
      "learning_rate": 8.170921198668146e-06,
      "loss": 0.0023,
      "step": 5330
    },
    {
      "epoch": 5.920687742651137,
      "grad_norm": 2.578124523162842,
      "learning_rate": 8.148723640399556e-06,
      "loss": 0.0029,
      "step": 5340
    },
    {
      "epoch": 5.931780366056572,
      "grad_norm": 0.048262905329465866,
      "learning_rate": 8.126526082130966e-06,
      "loss": 0.0007,
      "step": 5350
    },
    {
      "epoch": 5.942872989462008,
      "grad_norm": 0.027366768568754196,
      "learning_rate": 8.104328523862376e-06,
      "loss": 0.005,
      "step": 5360
    },
    {
      "epoch": 5.953965612867443,
      "grad_norm": 0.009290866553783417,
      "learning_rate": 8.082130965593786e-06,
      "loss": 0.0025,
      "step": 5370
    },
    {
      "epoch": 5.965058236272879,
      "grad_norm": 0.24025924503803253,
      "learning_rate": 8.059933407325194e-06,
      "loss": 0.0024,
      "step": 5380
    },
    {
      "epoch": 5.976150859678314,
      "grad_norm": 0.03495923802256584,
      "learning_rate": 8.037735849056606e-06,
      "loss": 0.0019,
      "step": 5390
    },
    {
      "epoch": 5.987243483083749,
      "grad_norm": 0.32967111468315125,
      "learning_rate": 8.015538290788014e-06,
      "loss": 0.0018,
      "step": 5400
    },
    {
      "epoch": 5.998336106489185,
      "grad_norm": 0.041376374661922455,
      "learning_rate": 7.993340732519424e-06,
      "loss": 0.0035,
      "step": 5410
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8861870989530564,
      "eval_f1": 0.705161854768154,
      "eval_loss": 0.5837549567222595,
      "eval_precision": 0.6948275862068966,
      "eval_recall": 0.7158081705150977,
      "eval_runtime": 2.7378,
      "eval_samples_per_second": 360.51,
      "eval_steps_per_second": 11.323,
      "step": 5412
    }
  ],
  "logging_steps": 10,
  "max_steps": 9010,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5.768243781369717e+16,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}