{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.999935917975008,
  "eval_steps": 500,
  "global_step": 7802,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001281640499839795,
      "grad_norm": 0.6719354391098022,
      "learning_rate": 2.560819462227913e-07,
      "loss": 0.1253,
      "step": 10
    },
    {
      "epoch": 0.00256328099967959,
      "grad_norm": 0.6422625184059143,
      "learning_rate": 5.121638924455826e-07,
      "loss": 0.1306,
      "step": 20
    },
    {
      "epoch": 0.003844921499519385,
      "grad_norm": 0.6302109956741333,
      "learning_rate": 7.682458386683739e-07,
      "loss": 0.118,
      "step": 30
    },
    {
      "epoch": 0.00512656199935918,
      "grad_norm": 0.6015211939811707,
      "learning_rate": 1.0243277848911651e-06,
      "loss": 0.1186,
      "step": 40
    },
    {
      "epoch": 0.006408202499198975,
      "grad_norm": 0.49298664927482605,
      "learning_rate": 1.2804097311139565e-06,
      "loss": 0.1123,
      "step": 50
    },
    {
      "epoch": 0.00768984299903877,
      "grad_norm": 0.4810634255409241,
      "learning_rate": 1.5364916773367477e-06,
      "loss": 0.1097,
      "step": 60
    },
    {
      "epoch": 0.008971483498878564,
      "grad_norm": 0.4184609353542328,
      "learning_rate": 1.7925736235595393e-06,
      "loss": 0.101,
      "step": 70
    },
    {
      "epoch": 0.01025312399871836,
      "grad_norm": 0.33549964427948,
      "learning_rate": 2.0486555697823303e-06,
      "loss": 0.0952,
      "step": 80
    },
    {
      "epoch": 0.011534764498558154,
      "grad_norm": 0.3348263204097748,
      "learning_rate": 2.304737516005122e-06,
      "loss": 0.0927,
      "step": 90
    },
    {
      "epoch": 0.01281640499839795,
      "grad_norm": 0.31015241146087646,
      "learning_rate": 2.560819462227913e-06,
      "loss": 0.089,
      "step": 100
    },
    {
      "epoch": 0.014098045498237744,
      "grad_norm": 0.2687341272830963,
      "learning_rate": 2.8169014084507046e-06,
      "loss": 0.085,
      "step": 110
    },
    {
      "epoch": 0.01537968599807754,
      "grad_norm": 0.24938784539699554,
      "learning_rate": 3.0729833546734954e-06,
      "loss": 0.0854,
      "step": 120
    },
    {
      "epoch": 0.016661326497917333,
      "grad_norm": 0.23371560871601105,
      "learning_rate": 3.329065300896287e-06,
      "loss": 0.0849,
      "step": 130
    },
    {
      "epoch": 0.017942966997757128,
      "grad_norm": 0.23148757219314575,
      "learning_rate": 3.5851472471190786e-06,
      "loss": 0.0822,
      "step": 140
    },
    {
      "epoch": 0.019224607497596923,
      "grad_norm": 0.24376137554645538,
      "learning_rate": 3.84122919334187e-06,
      "loss": 0.0832,
      "step": 150
    },
    {
      "epoch": 0.02050624799743672,
      "grad_norm": 0.24407723546028137,
      "learning_rate": 4.0973111395646605e-06,
      "loss": 0.0795,
      "step": 160
    },
    {
      "epoch": 0.021787888497276513,
      "grad_norm": 0.2162434309720993,
      "learning_rate": 4.3533930857874526e-06,
      "loss": 0.08,
      "step": 170
    },
    {
      "epoch": 0.02306952899711631,
      "grad_norm": 0.2380385547876358,
      "learning_rate": 4.609475032010244e-06,
      "loss": 0.077,
      "step": 180
    },
    {
      "epoch": 0.024351169496956104,
      "grad_norm": 0.21600289642810822,
      "learning_rate": 4.865556978233035e-06,
      "loss": 0.0801,
      "step": 190
    },
    {
      "epoch": 0.0256328099967959,
      "grad_norm": 0.21687458455562592,
      "learning_rate": 5.121638924455826e-06,
      "loss": 0.0777,
      "step": 200
    },
    {
      "epoch": 0.026914450496635694,
      "grad_norm": 0.20008929073810577,
      "learning_rate": 5.377720870678618e-06,
      "loss": 0.0767,
      "step": 210
    },
    {
      "epoch": 0.02819609099647549,
      "grad_norm": 0.2190534472465515,
      "learning_rate": 5.633802816901409e-06,
      "loss": 0.0792,
      "step": 220
    },
    {
      "epoch": 0.029477731496315284,
      "grad_norm": 0.2292010486125946,
      "learning_rate": 5.8898847631242005e-06,
      "loss": 0.0775,
      "step": 230
    },
    {
      "epoch": 0.03075937199615508,
      "grad_norm": 0.22605210542678833,
      "learning_rate": 6.145966709346991e-06,
      "loss": 0.0748,
      "step": 240
    },
    {
      "epoch": 0.03204101249599487,
      "grad_norm": 0.24900269508361816,
      "learning_rate": 6.402048655569783e-06,
      "loss": 0.0765,
      "step": 250
    },
    {
      "epoch": 0.033322652995834666,
      "grad_norm": 0.22502242028713226,
      "learning_rate": 6.658130601792574e-06,
      "loss": 0.0753,
      "step": 260
    },
    {
      "epoch": 0.03460429349567446,
      "grad_norm": 0.2257997840642929,
      "learning_rate": 6.914212548015365e-06,
      "loss": 0.0775,
      "step": 270
    },
    {
      "epoch": 0.035885933995514256,
      "grad_norm": 0.20458129048347473,
      "learning_rate": 7.170294494238157e-06,
      "loss": 0.074,
      "step": 280
    },
    {
      "epoch": 0.03716757449535405,
      "grad_norm": 0.19960536062717438,
      "learning_rate": 7.426376440460948e-06,
      "loss": 0.0753,
      "step": 290
    },
    {
      "epoch": 0.038449214995193846,
      "grad_norm": 0.20938412845134735,
      "learning_rate": 7.68245838668374e-06,
      "loss": 0.0731,
      "step": 300
    },
    {
      "epoch": 0.03973085549503364,
      "grad_norm": 0.215741828083992,
      "learning_rate": 7.93854033290653e-06,
      "loss": 0.0744,
      "step": 310
    },
    {
      "epoch": 0.04101249599487344,
      "grad_norm": 0.19086730480194092,
      "learning_rate": 8.194622279129321e-06,
      "loss": 0.072,
      "step": 320
    },
    {
      "epoch": 0.04229413649471323,
      "grad_norm": 0.204061821103096,
      "learning_rate": 8.450704225352114e-06,
      "loss": 0.0728,
      "step": 330
    },
    {
      "epoch": 0.04357577699455303,
      "grad_norm": 0.20682527124881744,
      "learning_rate": 8.706786171574905e-06,
      "loss": 0.0742,
      "step": 340
    },
    {
      "epoch": 0.04485741749439282,
      "grad_norm": 0.20134615898132324,
      "learning_rate": 8.962868117797696e-06,
      "loss": 0.0751,
      "step": 350
    },
    {
      "epoch": 0.04613905799423262,
      "grad_norm": 0.19907858967781067,
      "learning_rate": 9.218950064020487e-06,
      "loss": 0.0728,
      "step": 360
    },
    {
      "epoch": 0.04742069849407241,
      "grad_norm": 0.21698465943336487,
      "learning_rate": 9.475032010243279e-06,
      "loss": 0.0748,
      "step": 370
    },
    {
      "epoch": 0.04870233899391221,
      "grad_norm": 0.21630790829658508,
      "learning_rate": 9.73111395646607e-06,
      "loss": 0.0737,
      "step": 380
    },
    {
      "epoch": 0.049983979493752,
      "grad_norm": 0.2053464651107788,
      "learning_rate": 9.987195902688861e-06,
      "loss": 0.0733,
      "step": 390
    },
    {
      "epoch": 0.0512656199935918,
      "grad_norm": 0.22032414376735687,
      "learning_rate": 1.0243277848911652e-05,
      "loss": 0.0727,
      "step": 400
    },
    {
      "epoch": 0.05254726049343159,
      "grad_norm": 0.2172485888004303,
      "learning_rate": 1.0499359795134443e-05,
      "loss": 0.071,
      "step": 410
    },
    {
      "epoch": 0.05382890099327139,
      "grad_norm": 0.19321809709072113,
      "learning_rate": 1.0755441741357236e-05,
      "loss": 0.0739,
      "step": 420
    },
    {
      "epoch": 0.05511054149311118,
      "grad_norm": 0.1873639076948166,
      "learning_rate": 1.1011523687580026e-05,
      "loss": 0.0685,
      "step": 430
    },
    {
      "epoch": 0.05639218199295098,
      "grad_norm": 0.21123793721199036,
      "learning_rate": 1.1267605633802819e-05,
      "loss": 0.0703,
      "step": 440
    },
    {
      "epoch": 0.05767382249279077,
      "grad_norm": 0.20596912503242493,
      "learning_rate": 1.1523687580025608e-05,
      "loss": 0.0723,
      "step": 450
    },
    {
      "epoch": 0.05895546299263057,
      "grad_norm": 0.20657142996788025,
      "learning_rate": 1.1779769526248401e-05,
      "loss": 0.0705,
      "step": 460
    },
    {
      "epoch": 0.06023710349247036,
      "grad_norm": 0.19684277474880219,
      "learning_rate": 1.2035851472471192e-05,
      "loss": 0.0711,
      "step": 470
    },
    {
      "epoch": 0.06151874399231016,
      "grad_norm": 0.19987432658672333,
      "learning_rate": 1.2291933418693982e-05,
      "loss": 0.0733,
      "step": 480
    },
    {
      "epoch": 0.06280038449214995,
      "grad_norm": 0.2000504434108734,
      "learning_rate": 1.2548015364916774e-05,
      "loss": 0.0708,
      "step": 490
    },
    {
      "epoch": 0.06408202499198974,
      "grad_norm": 0.1920643448829651,
      "learning_rate": 1.2804097311139566e-05,
      "loss": 0.0707,
      "step": 500
    },
    {
      "epoch": 0.06408202499198974,
      "eval_f1": 0.7841977645041103,
      "eval_loss": 0.10520466417074203,
      "eval_precision": 0.7908863175474173,
      "eval_recall": 0.7776213933849402,
      "eval_runtime": 2.9914,
      "eval_samples_per_second": 2164.9,
      "eval_steps_per_second": 8.692,
      "step": 500
    },
    {
      "epoch": 0.06536366549182954,
      "grad_norm": 0.18772949278354645,
      "learning_rate": 1.3060179257362357e-05,
      "loss": 0.0694,
      "step": 510
    },
    {
      "epoch": 0.06664530599166933,
      "grad_norm": 0.17525950074195862,
      "learning_rate": 1.3316261203585148e-05,
      "loss": 0.0709,
      "step": 520
    },
    {
      "epoch": 0.06792694649150913,
      "grad_norm": 0.20220516622066498,
      "learning_rate": 1.3572343149807941e-05,
      "loss": 0.0714,
      "step": 530
    },
    {
      "epoch": 0.06920858699134892,
      "grad_norm": 0.2033197432756424,
      "learning_rate": 1.382842509603073e-05,
      "loss": 0.0719,
      "step": 540
    },
    {
      "epoch": 0.07049022749118872,
      "grad_norm": 0.1880401372909546,
      "learning_rate": 1.4084507042253522e-05,
      "loss": 0.0717,
      "step": 550
    },
    {
      "epoch": 0.07177186799102851,
      "grad_norm": 0.20734059810638428,
      "learning_rate": 1.4340588988476314e-05,
      "loss": 0.0696,
      "step": 560
    },
    {
      "epoch": 0.07305350849086831,
      "grad_norm": 0.18614107370376587,
      "learning_rate": 1.4596670934699104e-05,
      "loss": 0.0728,
      "step": 570
    },
    {
      "epoch": 0.0743351489907081,
      "grad_norm": 0.19527311623096466,
      "learning_rate": 1.4852752880921897e-05,
      "loss": 0.0721,
      "step": 580
    },
    {
      "epoch": 0.0756167894905479,
      "grad_norm": 0.19726967811584473,
      "learning_rate": 1.5108834827144688e-05,
      "loss": 0.0715,
      "step": 590
    },
    {
      "epoch": 0.07689842999038769,
      "grad_norm": 0.2089812010526657,
      "learning_rate": 1.536491677336748e-05,
      "loss": 0.0702,
      "step": 600
    },
    {
      "epoch": 0.07818007049022749,
      "grad_norm": 0.186571404337883,
      "learning_rate": 1.562099871959027e-05,
      "loss": 0.071,
      "step": 610
    },
    {
      "epoch": 0.07946171099006728,
      "grad_norm": 0.20535047352313995,
      "learning_rate": 1.587708066581306e-05,
      "loss": 0.0718,
      "step": 620
    },
    {
      "epoch": 0.08074335148990708,
      "grad_norm": 0.20184259116649628,
      "learning_rate": 1.6133162612035853e-05,
      "loss": 0.0701,
      "step": 630
    },
    {
      "epoch": 0.08202499198974687,
      "grad_norm": 0.20490041375160217,
      "learning_rate": 1.6389244558258642e-05,
      "loss": 0.0713,
      "step": 640
    },
    {
      "epoch": 0.08330663248958667,
      "grad_norm": 0.2261057049036026,
      "learning_rate": 1.6645326504481435e-05,
      "loss": 0.0715,
      "step": 650
    },
    {
      "epoch": 0.08458827298942646,
      "grad_norm": 0.2002391666173935,
      "learning_rate": 1.6901408450704228e-05,
      "loss": 0.072,
      "step": 660
    },
    {
      "epoch": 0.08586991348926626,
      "grad_norm": 0.1953844130039215,
      "learning_rate": 1.7157490396927017e-05,
      "loss": 0.0691,
      "step": 670
    },
    {
      "epoch": 0.08715155398910605,
      "grad_norm": 0.19759021699428558,
      "learning_rate": 1.741357234314981e-05,
      "loss": 0.0713,
      "step": 680
    },
    {
      "epoch": 0.08843319448894585,
      "grad_norm": 0.19472911953926086,
      "learning_rate": 1.76696542893726e-05,
      "loss": 0.0687,
      "step": 690
    },
    {
      "epoch": 0.08971483498878564,
      "grad_norm": 0.2004728615283966,
      "learning_rate": 1.7925736235595393e-05,
      "loss": 0.0713,
      "step": 700
    },
    {
      "epoch": 0.09099647548862544,
      "grad_norm": 0.2075321078300476,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 0.0681,
      "step": 710
    },
    {
      "epoch": 0.09227811598846523,
      "grad_norm": 0.2087249606847763,
      "learning_rate": 1.8437900128040975e-05,
      "loss": 0.0682,
      "step": 720
    },
    {
      "epoch": 0.09355975648830503,
      "grad_norm": 0.19576896727085114,
      "learning_rate": 1.8693982074263764e-05,
      "loss": 0.0695,
      "step": 730
    },
    {
      "epoch": 0.09484139698814482,
      "grad_norm": 0.1800433099269867,
      "learning_rate": 1.8950064020486557e-05,
      "loss": 0.0686,
      "step": 740
    },
    {
      "epoch": 0.09612303748798462,
      "grad_norm": 0.2001904547214508,
      "learning_rate": 1.920614596670935e-05,
      "loss": 0.0686,
      "step": 750
    },
    {
      "epoch": 0.09740467798782441,
      "grad_norm": 0.20266686379909515,
      "learning_rate": 1.946222791293214e-05,
      "loss": 0.0697,
      "step": 760
    },
    {
      "epoch": 0.09868631848766421,
      "grad_norm": 0.1891346275806427,
      "learning_rate": 1.9718309859154933e-05,
      "loss": 0.0681,
      "step": 770
    },
    {
      "epoch": 0.099967958987504,
      "grad_norm": 0.18907149136066437,
      "learning_rate": 1.9974391805377722e-05,
      "loss": 0.07,
      "step": 780
    },
    {
      "epoch": 0.1012495994873438,
      "grad_norm": 0.19888722896575928,
      "learning_rate": 1.9999918912065962e-05,
      "loss": 0.0696,
      "step": 790
    },
    {
      "epoch": 0.1025312399871836,
      "grad_norm": 0.19157752394676208,
      "learning_rate": 1.9999638609784763e-05,
      "loss": 0.0694,
      "step": 800
    },
    {
      "epoch": 0.10381288048702339,
      "grad_norm": 0.21350421011447906,
      "learning_rate": 1.99991580976816e-05,
      "loss": 0.0703,
      "step": 810
    },
    {
      "epoch": 0.10509452098686319,
      "grad_norm": 0.20590972900390625,
      "learning_rate": 1.9998477385377137e-05,
      "loss": 0.0708,
      "step": 820
    },
    {
      "epoch": 0.10637616148670298,
      "grad_norm": 0.22003325819969177,
      "learning_rate": 1.9997596486500402e-05,
      "loss": 0.0694,
      "step": 830
    },
    {
      "epoch": 0.10765780198654278,
      "grad_norm": 0.2006986439228058,
      "learning_rate": 1.9996515418688493e-05,
      "loss": 0.0707,
      "step": 840
    },
    {
      "epoch": 0.10893944248638257,
      "grad_norm": 0.1951693445444107,
      "learning_rate": 1.9995234203586223e-05,
      "loss": 0.069,
      "step": 850
    },
    {
      "epoch": 0.11022108298622237,
      "grad_norm": 0.20470306277275085,
      "learning_rate": 1.999375286684571e-05,
      "loss": 0.0685,
      "step": 860
    },
    {
      "epoch": 0.11150272348606216,
      "grad_norm": 0.21397650241851807,
      "learning_rate": 1.9992071438125846e-05,
      "loss": 0.0691,
      "step": 870
    },
    {
      "epoch": 0.11278436398590196,
      "grad_norm": 0.19602704048156738,
      "learning_rate": 1.9990189951091697e-05,
      "loss": 0.0689,
      "step": 880
    },
    {
      "epoch": 0.11406600448574175,
      "grad_norm": 0.20522291958332062,
      "learning_rate": 1.998810844341384e-05,
      "loss": 0.0714,
      "step": 890
    },
    {
      "epoch": 0.11534764498558155,
      "grad_norm": 0.2263880968093872,
      "learning_rate": 1.998582695676762e-05,
      "loss": 0.0702,
      "step": 900
    },
    {
      "epoch": 0.11662928548542134,
      "grad_norm": 0.20244184136390686,
      "learning_rate": 1.9983345536832282e-05,
      "loss": 0.0701,
      "step": 910
    },
    {
      "epoch": 0.11791092598526114,
      "grad_norm": 0.17990508675575256,
      "learning_rate": 1.998066423329009e-05,
      "loss": 0.0695,
      "step": 920
    },
    {
      "epoch": 0.11919256648510093,
      "grad_norm": 0.18233992159366608,
      "learning_rate": 1.9977783099825314e-05,
      "loss": 0.067,
      "step": 930
    },
    {
      "epoch": 0.12047420698494073,
      "grad_norm": 0.20347543060779572,
      "learning_rate": 1.9974702194123156e-05,
      "loss": 0.071,
      "step": 940
    },
    {
      "epoch": 0.12175584748478052,
      "grad_norm": 0.19996987283229828,
      "learning_rate": 1.9971421577868605e-05,
      "loss": 0.069,
      "step": 950
    },
    {
      "epoch": 0.12303748798462032,
      "grad_norm": 0.17577455937862396,
      "learning_rate": 1.9967941316745197e-05,
      "loss": 0.0648,
      "step": 960
    },
    {
      "epoch": 0.12431912848446011,
      "grad_norm": 0.20562253892421722,
      "learning_rate": 1.99642614804337e-05,
      "loss": 0.0708,
      "step": 970
    },
    {
      "epoch": 0.1256007689842999,
      "grad_norm": 0.20121215283870697,
      "learning_rate": 1.996038214261071e-05,
      "loss": 0.0692,
      "step": 980
    },
    {
      "epoch": 0.1268824094841397,
      "grad_norm": 0.17846404016017914,
      "learning_rate": 1.9956303380947182e-05,
      "loss": 0.0663,
      "step": 990
    },
    {
      "epoch": 0.12816404998397948,
      "grad_norm": 0.2094384878873825,
      "learning_rate": 1.995202527710689e-05,
      "loss": 0.0704,
      "step": 1000
    },
    {
      "epoch": 0.12816404998397948,
      "eval_f1": 0.8013096465167041,
      "eval_loss": 0.09550511837005615,
      "eval_precision": 0.8294213631228818,
      "eval_recall": 0.7750410509031199,
      "eval_runtime": 2.9934,
      "eval_samples_per_second": 2163.405,
      "eval_steps_per_second": 8.686,
      "step": 1000
    },
    {
      "epoch": 0.1294456904838193,
      "grad_norm": 0.2003297060728073,
      "learning_rate": 1.9947547916744762e-05,
      "loss": 0.0668,
      "step": 1010
    },
    {
      "epoch": 0.13072733098365907,
      "grad_norm": 0.18323542177677155,
      "learning_rate": 1.99428713895052e-05,
      "loss": 0.0675,
      "step": 1020
    },
    {
      "epoch": 0.13200897148349888,
      "grad_norm": 0.1899520307779312,
      "learning_rate": 1.993799578902025e-05,
      "loss": 0.0694,
      "step": 1030
    },
    {
      "epoch": 0.13329061198333866,
      "grad_norm": 0.19993416965007782,
      "learning_rate": 1.9932921212907753e-05,
      "loss": 0.0702,
      "step": 1040
    },
    {
      "epoch": 0.13457225248317847,
      "grad_norm": 0.2051459103822708,
      "learning_rate": 1.9927647762769372e-05,
      "loss": 0.0712,
      "step": 1050
    },
    {
      "epoch": 0.13585389298301825,
      "grad_norm": 0.19359096884727478,
      "learning_rate": 1.9922175544188587e-05,
      "loss": 0.0707,
      "step": 1060
    },
    {
      "epoch": 0.13713553348285806,
      "grad_norm": 0.18876825273036957,
      "learning_rate": 1.9916504666728533e-05,
      "loss": 0.0683,
      "step": 1070
    },
    {
      "epoch": 0.13841717398269784,
      "grad_norm": 0.18740342557430267,
      "learning_rate": 1.9910635243929864e-05,
      "loss": 0.069,
      "step": 1080
    },
    {
      "epoch": 0.13969881448253765,
      "grad_norm": 0.18416374921798706,
      "learning_rate": 1.9904567393308425e-05,
      "loss": 0.07,
      "step": 1090
    },
    {
      "epoch": 0.14098045498237743,
      "grad_norm": 0.17465655505657196,
      "learning_rate": 1.9898301236352945e-05,
      "loss": 0.0681,
      "step": 1100
    },
    {
      "epoch": 0.14226209548221724,
      "grad_norm": 0.19771921634674072,
      "learning_rate": 1.9891836898522566e-05,
      "loss": 0.0677,
      "step": 1110
    },
    {
      "epoch": 0.14354373598205702,
      "grad_norm": 0.19512903690338135,
      "learning_rate": 1.9885174509244366e-05,
      "loss": 0.0702,
      "step": 1120
    },
    {
      "epoch": 0.14482537648189683,
      "grad_norm": 0.18367761373519897,
      "learning_rate": 1.9878314201910734e-05,
      "loss": 0.068,
      "step": 1130
    },
    {
      "epoch": 0.14610701698173661,
      "grad_norm": 0.19701771438121796,
      "learning_rate": 1.9871256113876736e-05,
      "loss": 0.0711,
      "step": 1140
    },
    {
      "epoch": 0.14738865748157642,
      "grad_norm": 0.18831095099449158,
      "learning_rate": 1.9864000386457323e-05,
      "loss": 0.0703,
      "step": 1150
    },
    {
      "epoch": 0.1486702979814162,
      "grad_norm": 0.1935907006263733,
      "learning_rate": 1.9856547164924542e-05,
      "loss": 0.0698,
      "step": 1160
    },
    {
      "epoch": 0.14995193848125601,
      "grad_norm": 0.2067263126373291,
      "learning_rate": 1.98488965985046e-05,
      "loss": 0.0716,
      "step": 1170
    },
    {
      "epoch": 0.1512335789810958,
      "grad_norm": 0.20484298467636108,
      "learning_rate": 1.9841048840374885e-05,
      "loss": 0.0672,
      "step": 1180
    },
    {
      "epoch": 0.1525152194809356,
      "grad_norm": 0.19146059453487396,
      "learning_rate": 1.98330040476609e-05,
      "loss": 0.0699,
      "step": 1190
    },
    {
      "epoch": 0.15379685998077539,
      "grad_norm": 0.1922774314880371,
      "learning_rate": 1.982476238143312e-05,
      "loss": 0.0689,
      "step": 1200
    },
    {
      "epoch": 0.1550785004806152,
      "grad_norm": 0.17350053787231445,
      "learning_rate": 1.981632400670376e-05,
      "loss": 0.0698,
      "step": 1210
    },
    {
      "epoch": 0.15636014098045498,
      "grad_norm": 0.20516818761825562,
      "learning_rate": 1.9807689092423478e-05,
      "loss": 0.0687,
      "step": 1220
    },
    {
      "epoch": 0.15764178148029478,
      "grad_norm": 0.18992391228675842,
      "learning_rate": 1.9798857811477984e-05,
      "loss": 0.0646,
      "step": 1230
    },
    {
      "epoch": 0.15892342198013457,
      "grad_norm": 0.1936413198709488,
      "learning_rate": 1.9789830340684593e-05,
      "loss": 0.0693,
      "step": 1240
    },
    {
      "epoch": 0.16020506247997437,
      "grad_norm": 0.19793152809143066,
      "learning_rate": 1.978060686078866e-05,
      "loss": 0.0685,
      "step": 1250
    },
    {
      "epoch": 0.16148670297981416,
      "grad_norm": 0.19546934962272644,
      "learning_rate": 1.9771187556459988e-05,
      "loss": 0.0698,
      "step": 1260
    },
    {
      "epoch": 0.16276834347965397,
      "grad_norm": 0.21572382748126984,
      "learning_rate": 1.976157261628912e-05,
      "loss": 0.07,
      "step": 1270
    },
    {
      "epoch": 0.16404998397949375,
      "grad_norm": 0.2009066939353943,
      "learning_rate": 1.975176223278356e-05,
      "loss": 0.0668,
      "step": 1280
    },
    {
      "epoch": 0.16533162447933356,
      "grad_norm": 0.19803877174854279,
      "learning_rate": 1.9741756602363914e-05,
      "loss": 0.0672,
      "step": 1290
    },
    {
      "epoch": 0.16661326497917334,
      "grad_norm": 0.1853305697441101,
      "learning_rate": 1.9731555925359974e-05,
      "loss": 0.0676,
      "step": 1300
    },
    {
      "epoch": 0.16789490547901315,
      "grad_norm": 0.1944134682416916,
      "learning_rate": 1.9721160406006697e-05,
      "loss": 0.0685,
      "step": 1310
    },
    {
      "epoch": 0.16917654597885293,
      "grad_norm": 0.20351552963256836,
      "learning_rate": 1.9710570252440106e-05,
      "loss": 0.0671,
      "step": 1320
    },
    {
      "epoch": 0.17045818647869274,
      "grad_norm": 0.18567000329494476,
      "learning_rate": 1.969978567669314e-05,
      "loss": 0.0676,
      "step": 1330
    },
    {
      "epoch": 0.17173982697853252,
      "grad_norm": 0.17873169481754303,
      "learning_rate": 1.968880689469141e-05,
      "loss": 0.0682,
      "step": 1340
    },
    {
      "epoch": 0.17302146747837233,
      "grad_norm": 0.2007400393486023,
      "learning_rate": 1.9677634126248847e-05,
      "loss": 0.067,
      "step": 1350
    },
    {
      "epoch": 0.1743031079782121,
      "grad_norm": 0.1938178986310959,
      "learning_rate": 1.9666267595063344e-05,
      "loss": 0.0683,
      "step": 1360
    },
    {
      "epoch": 0.17558474847805192,
      "grad_norm": 0.21833261847496033,
      "learning_rate": 1.9654707528712236e-05,
      "loss": 0.0682,
      "step": 1370
    },
    {
      "epoch": 0.1768663889778917,
      "grad_norm": 0.1728629469871521,
      "learning_rate": 1.9642954158647776e-05,
      "loss": 0.0686,
      "step": 1380
    },
    {
      "epoch": 0.1781480294777315,
      "grad_norm": 0.19660884141921997,
      "learning_rate": 1.9631007720192475e-05,
      "loss": 0.0663,
      "step": 1390
    },
    {
      "epoch": 0.1794296699775713,
      "grad_norm": 0.19190044701099396,
      "learning_rate": 1.9618868452534415e-05,
      "loss": 0.0684,
      "step": 1400
    },
    {
      "epoch": 0.1807113104774111,
      "grad_norm": 0.20303724706172943,
      "learning_rate": 1.9606536598722435e-05,
      "loss": 0.0658,
      "step": 1410
    },
    {
      "epoch": 0.18199295097725088,
      "grad_norm": 0.18915805220603943,
      "learning_rate": 1.959401240566129e-05,
      "loss": 0.0696,
      "step": 1420
    },
    {
      "epoch": 0.1832745914770907,
      "grad_norm": 0.19406339526176453,
      "learning_rate": 1.9581296124106682e-05,
      "loss": 0.0671,
      "step": 1430
    },
    {
      "epoch": 0.18455623197693047,
      "grad_norm": 0.18201453983783722,
      "learning_rate": 1.9568388008660262e-05,
      "loss": 0.0684,
      "step": 1440
    },
    {
      "epoch": 0.18583787247677028,
      "grad_norm": 0.18637534976005554,
      "learning_rate": 1.955528831776452e-05,
      "loss": 0.0653,
      "step": 1450
    },
    {
      "epoch": 0.18711951297661006,
      "grad_norm": 0.1933770328760147,
      "learning_rate": 1.9541997313697614e-05,
      "loss": 0.0698,
      "step": 1460
    },
    {
      "epoch": 0.18840115347644987,
      "grad_norm": 0.18085801601409912,
      "learning_rate": 1.952851526256811e-05,
      "loss": 0.0664,
      "step": 1470
    },
    {
      "epoch": 0.18968279397628965,
      "grad_norm": 0.20404258370399475,
      "learning_rate": 1.951484243430968e-05,
      "loss": 0.0675,
      "step": 1480
    },
    {
      "epoch": 0.19096443447612946,
      "grad_norm": 0.1916743814945221,
      "learning_rate": 1.9500979102675654e-05,
      "loss": 0.0669,
      "step": 1490
    },
    {
      "epoch": 0.19224607497596924,
      "grad_norm": 0.20056353509426117,
      "learning_rate": 1.9486925545233587e-05,
      "loss": 0.066,
      "step": 1500
    },
    {
      "epoch": 0.19224607497596924,
      "eval_f1": 0.825402613187481,
      "eval_loss": 0.08602051436901093,
      "eval_precision": 0.8564762265102788,
      "eval_recall": 0.7965048088200798,
      "eval_runtime": 2.9899,
      "eval_samples_per_second": 2165.931,
      "eval_steps_per_second": 8.696,
      "step": 1500
    },
    {
      "epoch": 0.19352771547580905,
      "grad_norm": 0.21483555436134338,
      "learning_rate": 1.9472682043359676e-05,
      "loss": 0.068,
      "step": 1510
    },
    {
      "epoch": 0.19480935597564883,
      "grad_norm": 0.19758599996566772,
      "learning_rate": 1.9458248882233117e-05,
      "loss": 0.0683,
      "step": 1520
    },
    {
      "epoch": 0.19609099647548864,
      "grad_norm": 0.18787485361099243,
      "learning_rate": 1.9443626350830417e-05,
      "loss": 0.0663,
      "step": 1530
    },
    {
      "epoch": 0.19737263697532842,
      "grad_norm": 0.20583724975585938,
      "learning_rate": 1.9428814741919595e-05,
      "loss": 0.0692,
      "step": 1540
    },
    {
      "epoch": 0.19865427747516823,
      "grad_norm": 0.1839466392993927,
      "learning_rate": 1.941381435205433e-05,
      "loss": 0.0684,
      "step": 1550
    },
    {
      "epoch": 0.199935917975008,
      "grad_norm": 0.2200762778520584,
      "learning_rate": 1.9398625481568015e-05,
      "loss": 0.0694,
      "step": 1560
    },
    {
      "epoch": 0.20121755847484782,
      "grad_norm": 0.1883755922317505,
      "learning_rate": 1.9383248434567745e-05,
      "loss": 0.0671,
      "step": 1570
    },
    {
      "epoch": 0.2024991989746876,
      "grad_norm": 0.19850397109985352,
      "learning_rate": 1.9367683518928228e-05,
      "loss": 0.0645,
      "step": 1580
    },
    {
      "epoch": 0.2037808394745274,
      "grad_norm": 0.19924448430538177,
      "learning_rate": 1.9351931046285622e-05,
      "loss": 0.0704,
      "step": 1590
    },
    {
      "epoch": 0.2050624799743672,
      "grad_norm": 0.1877722144126892,
      "learning_rate": 1.933599133203131e-05,
      "loss": 0.0686,
      "step": 1600
    },
    {
      "epoch": 0.20634412047420697,
      "grad_norm": 0.20414628088474274,
      "learning_rate": 1.931986469530555e-05,
      "loss": 0.067,
      "step": 1610
    },
    {
      "epoch": 0.20762576097404678,
      "grad_norm": 0.18630723655223846,
      "learning_rate": 1.9303551458991127e-05,
      "loss": 0.0667,
      "step": 1620
    },
    {
      "epoch": 0.20890740147388656,
      "grad_norm": 0.18654927611351013,
      "learning_rate": 1.928705194970685e-05,
      "loss": 0.066,
      "step": 1630
    },
    {
      "epoch": 0.21018904197372637,
      "grad_norm": 0.1941138505935669,
      "learning_rate": 1.9270366497801044e-05,
      "loss": 0.0659,
      "step": 1640
    },
    {
      "epoch": 0.21147068247356615,
      "grad_norm": 0.23397959768772125,
      "learning_rate": 1.9253495437344914e-05,
      "loss": 0.0684,
      "step": 1650
    },
    {
      "epoch": 0.21275232297340596,
      "grad_norm": 0.19757859408855438,
      "learning_rate": 1.9236439106125874e-05,
      "loss": 0.0675,
      "step": 1660
    },
    {
      "epoch": 0.21403396347324574,
      "grad_norm": 0.20172664523124695,
      "learning_rate": 1.9219197845640766e-05,
      "loss": 0.0664,
      "step": 1670
    },
    {
      "epoch": 0.21531560397308555,
      "grad_norm": 0.20976141095161438,
      "learning_rate": 1.9201772001089033e-05,
      "loss": 0.0665,
      "step": 1680
    },
    {
      "epoch": 0.21659724447292533,
      "grad_norm": 0.19794906675815582,
      "learning_rate": 1.9184161921365806e-05,
      "loss": 0.0657,
      "step": 1690
    },
    {
      "epoch": 0.21787888497276514,
      "grad_norm": 0.22197921574115753,
      "learning_rate": 1.9166367959054924e-05,
      "loss": 0.0651,
      "step": 1700
    },
    {
      "epoch": 0.21916052547260492,
      "grad_norm": 0.2003927230834961,
      "learning_rate": 1.9148390470421862e-05,
      "loss": 0.0664,
      "step": 1710
    },
    {
      "epoch": 0.22044216597244473,
      "grad_norm": 0.1864669919013977,
      "learning_rate": 1.9130229815406605e-05,
      "loss": 0.0665,
      "step": 1720
    },
    {
      "epoch": 0.2217238064722845,
      "grad_norm": 0.19202478229999542,
      "learning_rate": 1.911188635761645e-05,
      "loss": 0.0692,
      "step": 1730
    },
    {
      "epoch": 0.22300544697212432,
      "grad_norm": 0.19340120255947113,
      "learning_rate": 1.909336046431871e-05,
      "loss": 0.065,
      "step": 1740
    },
    {
      "epoch": 0.2242870874719641,
      "grad_norm": 0.19901137053966522,
      "learning_rate": 1.907465250643337e-05,
      "loss": 0.0666,
      "step": 1750
    },
    {
      "epoch": 0.2255687279718039,
      "grad_norm": 0.19657251238822937,
      "learning_rate": 1.9055762858525654e-05,
      "loss": 0.0674,
      "step": 1760
    },
    {
      "epoch": 0.2268503684716437,
      "grad_norm": 0.19742804765701294,
      "learning_rate": 1.9036691898798535e-05,
      "loss": 0.0672,
      "step": 1770
    },
    {
      "epoch": 0.2281320089714835,
      "grad_norm": 0.2104116529226303,
      "learning_rate": 1.9017440009085165e-05,
      "loss": 0.0704,
      "step": 1780
    },
    {
      "epoch": 0.22941364947132328,
      "grad_norm": 0.18256083130836487,
      "learning_rate": 1.8998007574841206e-05,
      "loss": 0.0692,
      "step": 1790
    },
    {
      "epoch": 0.2306952899711631,
      "grad_norm": 0.18848790228366852,
      "learning_rate": 1.8978394985137144e-05,
      "loss": 0.0666,
      "step": 1800
    },
    {
      "epoch": 0.23197693047100287,
      "grad_norm": 0.1659405380487442,
      "learning_rate": 1.8958602632650474e-05,
      "loss": 0.0641,
      "step": 1810
    },
    {
      "epoch": 0.23325857097084268,
      "grad_norm": 0.18415990471839905,
      "learning_rate": 1.8938630913657863e-05,
      "loss": 0.0684,
      "step": 1820
    },
    {
      "epoch": 0.23454021147068246,
      "grad_norm": 0.18959344923496246,
      "learning_rate": 1.8918480228027182e-05,
      "loss": 0.0676,
      "step": 1830
    },
    {
      "epoch": 0.23582185197052227,
      "grad_norm": 0.20328116416931152,
      "learning_rate": 1.8898150979209536e-05,
      "loss": 0.0649,
      "step": 1840
    },
    {
      "epoch": 0.23710349247036205,
      "grad_norm": 0.21128340065479279,
      "learning_rate": 1.8877643574231165e-05,
      "loss": 0.0654,
      "step": 1850
    },
    {
      "epoch": 0.23838513297020186,
      "grad_norm": 0.18140177428722382,
      "learning_rate": 1.8856958423685288e-05,
      "loss": 0.0688,
      "step": 1860
    },
    {
      "epoch": 0.23966677347004164,
      "grad_norm": 0.18850302696228027,
      "learning_rate": 1.883609594172391e-05,
      "loss": 0.0667,
      "step": 1870
    },
    {
      "epoch": 0.24094841396988145,
      "grad_norm": 0.20093129575252533,
      "learning_rate": 1.8815056546049505e-05,
      "loss": 0.0678,
      "step": 1880
    },
    {
      "epoch": 0.24223005446972123,
      "grad_norm": 0.1854446977376938,
      "learning_rate": 1.8793840657906664e-05,
      "loss": 0.0666,
      "step": 1890
    },
    {
      "epoch": 0.24351169496956104,
      "grad_norm": 0.20104651153087616,
      "learning_rate": 1.8772448702073654e-05,
      "loss": 0.0687,
      "step": 1900
    },
    {
      "epoch": 0.24479333546940082,
      "grad_norm": 0.19386902451515198,
      "learning_rate": 1.8750881106853923e-05,
      "loss": 0.0672,
      "step": 1910
    },
    {
      "epoch": 0.24607497596924063,
      "grad_norm": 0.19654834270477295,
      "learning_rate": 1.8729138304067512e-05,
      "loss": 0.0672,
      "step": 1920
    },
    {
      "epoch": 0.24735661646908041,
      "grad_norm": 0.17995163798332214,
      "learning_rate": 1.8707220729042427e-05,
      "loss": 0.0674,
      "step": 1930
    },
    {
      "epoch": 0.24863825696892022,
      "grad_norm": 0.2079264372587204,
      "learning_rate": 1.8685128820605904e-05,
      "loss": 0.0658,
      "step": 1940
    },
    {
      "epoch": 0.24991989746876,
      "grad_norm": 0.20509222149848938,
      "learning_rate": 1.8662863021075632e-05,
      "loss": 0.068,
      "step": 1950
    },
    {
      "epoch": 0.2512015379685998,
      "grad_norm": 0.20490337908267975,
      "learning_rate": 1.8640423776250895e-05,
      "loss": 0.066,
      "step": 1960
    },
    {
      "epoch": 0.2524831784684396,
      "grad_norm": 0.19565828144550323,
      "learning_rate": 1.8617811535403658e-05,
      "loss": 0.0686,
      "step": 1970
    },
    {
      "epoch": 0.2537648189682794,
      "grad_norm": 0.20575258135795593,
      "learning_rate": 1.859502675126955e-05,
      "loss": 0.0664,
      "step": 1980
    },
    {
      "epoch": 0.2550464594681192,
      "grad_norm": 0.2111874371767044,
      "learning_rate": 1.857206988003881e-05,
      "loss": 0.0667,
      "step": 1990
    },
    {
      "epoch": 0.25632809996795897,
      "grad_norm": 0.19794400036334991,
      "learning_rate": 1.8548941381347165e-05,
      "loss": 0.0677,
      "step": 2000
    },
    {
      "epoch": 0.25632809996795897,
      "eval_f1": 0.8409049773755656,
      "eval_loss": 0.08060676604509354,
      "eval_precision": 0.8658218412225122,
      "eval_recall": 0.8173821252638986,
      "eval_runtime": 2.9949,
      "eval_samples_per_second": 2162.361,
      "eval_steps_per_second": 8.681,
      "step": 2000
    },
    {
      "epoch": 0.2576097404677988,
      "grad_norm": 0.18614304065704346,
      "learning_rate": 1.8525641718266612e-05,
      "loss": 0.0645,
      "step": 2010
    },
    {
      "epoch": 0.2588913809676386,
      "grad_norm": 0.19621381163597107,
      "learning_rate": 1.8502171357296144e-05,
      "loss": 0.0673,
      "step": 2020
    },
    {
      "epoch": 0.2601730214674784,
      "grad_norm": 0.20377185940742493,
      "learning_rate": 1.8478530768352424e-05,
      "loss": 0.0677,
      "step": 2030
    },
    {
      "epoch": 0.26145466196731815,
      "grad_norm": 0.2045605629682541,
      "learning_rate": 1.845472042476037e-05,
      "loss": 0.0679,
      "step": 2040
    },
    {
      "epoch": 0.26273630246715796,
      "grad_norm": 0.16738948225975037,
      "learning_rate": 1.8430740803243674e-05,
      "loss": 0.0639,
      "step": 2050
    },
    {
      "epoch": 0.26401794296699777,
      "grad_norm": 0.17515943944454193,
      "learning_rate": 1.8406592383915263e-05,
      "loss": 0.0627,
      "step": 2060
    },
    {
      "epoch": 0.2652995834668376,
      "grad_norm": 0.18184049427509308,
      "learning_rate": 1.838227565026768e-05,
      "loss": 0.0675,
      "step": 2070
    },
    {
      "epoch": 0.2665812239666773,
      "grad_norm": 0.1939259171485901,
      "learning_rate": 1.835779108916342e-05,
      "loss": 0.0665,
      "step": 2080
    },
    {
      "epoch": 0.26786286446651714,
      "grad_norm": 0.18487761914730072,
      "learning_rate": 1.833313919082515e-05,
      "loss": 0.0685,
      "step": 2090
    },
    {
      "epoch": 0.26914450496635695,
      "grad_norm": 0.20124389231204987,
      "learning_rate": 1.8308320448825934e-05,
      "loss": 0.0661,
      "step": 2100
    },
    {
      "epoch": 0.27042614546619675,
      "grad_norm": 0.19602565467357635,
      "learning_rate": 1.8283335360079317e-05,
      "loss": 0.0679,
      "step": 2110
    },
    {
      "epoch": 0.2717077859660365,
      "grad_norm": 0.19158747792243958,
      "learning_rate": 1.8258184424829392e-05,
      "loss": 0.065,
      "step": 2120
    },
    {
      "epoch": 0.2729894264658763,
      "grad_norm": 0.21096321940422058,
      "learning_rate": 1.8232868146640792e-05,
      "loss": 0.0633,
      "step": 2130
    },
    {
      "epoch": 0.2742710669657161,
      "grad_norm": 0.18472693860530853,
      "learning_rate": 1.8207387032388584e-05,
      "loss": 0.0659,
      "step": 2140
    },
    {
      "epoch": 0.27555270746555593,
      "grad_norm": 0.2008291631937027,
      "learning_rate": 1.818174159224814e-05,
      "loss": 0.067,
      "step": 2150
    },
    {
      "epoch": 0.2768343479653957,
      "grad_norm": 0.20867876708507538,
      "learning_rate": 1.815593233968492e-05,
      "loss": 0.0643,
      "step": 2160
    },
    {
      "epoch": 0.2781159884652355,
      "grad_norm": 0.191690132021904,
      "learning_rate": 1.812995979144418e-05,
      "loss": 0.0657,
      "step": 2170
    },
    {
      "epoch": 0.2793976289650753,
      "grad_norm": 0.18234828114509583,
      "learning_rate": 1.8103824467540642e-05,
      "loss": 0.0667,
      "step": 2180
    },
    {
      "epoch": 0.2806792694649151,
      "grad_norm": 0.1962231993675232,
      "learning_rate": 1.8077526891248073e-05,
      "loss": 0.0671,
      "step": 2190
    },
    {
      "epoch": 0.28196090996475487,
      "grad_norm": 0.1902410089969635,
      "learning_rate": 1.8051067589088803e-05,
      "loss": 0.066,
      "step": 2200
    },
    {
      "epoch": 0.2832425504645947,
      "grad_norm": 0.2025291472673416,
      "learning_rate": 1.8024447090823198e-05,
      "loss": 0.0669,
      "step": 2210
    },
    {
      "epoch": 0.2845241909644345,
      "grad_norm": 0.1987154334783554,
      "learning_rate": 1.7997665929439038e-05,
      "loss": 0.0675,
      "step": 2220
    },
    {
      "epoch": 0.2858058314642743,
      "grad_norm": 0.17381438612937927,
      "learning_rate": 1.7970724641140864e-05,
      "loss": 0.0663,
      "step": 2230
    },
    {
      "epoch": 0.28708747196411405,
      "grad_norm": 0.1927717924118042,
      "learning_rate": 1.7943623765339217e-05,
      "loss": 0.0675,
      "step": 2240
    },
    {
      "epoch": 0.28836911246395386,
      "grad_norm": 0.18249468505382538,
      "learning_rate": 1.7916363844639854e-05,
      "loss": 0.0672,
      "step": 2250
    },
    {
      "epoch": 0.28965075296379367,
      "grad_norm": 0.2089211344718933,
      "learning_rate": 1.7888945424832896e-05,
      "loss": 0.0671,
      "step": 2260
    },
    {
      "epoch": 0.2909323934636335,
      "grad_norm": 0.20524665713310242,
      "learning_rate": 1.7861369054881867e-05,
      "loss": 0.0685,
      "step": 2270
    },
    {
      "epoch": 0.29221403396347323,
      "grad_norm": 0.2076677680015564,
      "learning_rate": 1.7833635286912743e-05,
      "loss": 0.0655,
      "step": 2280
    },
    {
      "epoch": 0.29349567446331304,
      "grad_norm": 0.20620182156562805,
      "learning_rate": 1.780574467620286e-05,
      "loss": 0.0649,
      "step": 2290
    },
    {
      "epoch": 0.29477731496315285,
      "grad_norm": 0.188080295920372,
      "learning_rate": 1.7777697781169813e-05,
      "loss": 0.0649,
      "step": 2300
    },
    {
      "epoch": 0.29605895546299266,
      "grad_norm": 0.18712954223155975,
      "learning_rate": 1.7749495163360292e-05,
      "loss": 0.0662,
      "step": 2310
    },
    {
      "epoch": 0.2973405959628324,
      "grad_norm": 0.19281476736068726,
      "learning_rate": 1.772113738743881e-05,
      "loss": 0.0655,
      "step": 2320
    },
    {
      "epoch": 0.2986222364626722,
      "grad_norm": 0.18981251120567322,
      "learning_rate": 1.7692625021176412e-05,
      "loss": 0.0642,
      "step": 2330
    },
    {
      "epoch": 0.29990387696251203,
      "grad_norm": 0.1883469671010971,
      "learning_rate": 1.7663958635439303e-05,
      "loss": 0.0664,
      "step": 2340
    },
    {
      "epoch": 0.30118551746235184,
      "grad_norm": 0.1924140453338623,
      "learning_rate": 1.7635138804177424e-05,
      "loss": 0.0659,
      "step": 2350
    },
    {
      "epoch": 0.3024671579621916,
      "grad_norm": 0.18589209020137787,
      "learning_rate": 1.760616610441296e-05,
      "loss": 0.0677,
      "step": 2360
    },
    {
      "epoch": 0.3037487984620314,
      "grad_norm": 0.18945348262786865,
      "learning_rate": 1.757704111622878e-05,
      "loss": 0.0662,
      "step": 2370
    },
    {
      "epoch": 0.3050304389618712,
      "grad_norm": 0.1850900948047638,
      "learning_rate": 1.7547764422756827e-05,
      "loss": 0.0639,
      "step": 2380
    },
    {
      "epoch": 0.306312079461711,
      "grad_norm": 0.18490533530712128,
      "learning_rate": 1.751833661016644e-05,
      "loss": 0.065,
      "step": 2390
    },
    {
      "epoch": 0.30759371996155077,
      "grad_norm": 0.21076135337352753,
      "learning_rate": 1.748875826765263e-05,
      "loss": 0.0659,
      "step": 2400
    },
    {
      "epoch": 0.3088753604613906,
      "grad_norm": 0.18776057660579681,
      "learning_rate": 1.7459029987424267e-05,
      "loss": 0.0652,
      "step": 2410
    },
    {
      "epoch": 0.3101570009612304,
      "grad_norm": 0.18914464116096497,
      "learning_rate": 1.7429152364692225e-05,
      "loss": 0.0671,
      "step": 2420
    },
    {
      "epoch": 0.31143864146107014,
      "grad_norm": 0.16848713159561157,
      "learning_rate": 1.7399125997657478e-05,
      "loss": 0.0655,
      "step": 2430
    },
    {
      "epoch": 0.31272028196090995,
      "grad_norm": 0.18632112443447113,
      "learning_rate": 1.736895148749911e-05,
      "loss": 0.0692,
      "step": 2440
    },
    {
      "epoch": 0.31400192246074976,
      "grad_norm": 0.2024405598640442,
      "learning_rate": 1.7338629438362294e-05,
      "loss": 0.0669,
      "step": 2450
    },
    {
      "epoch": 0.31528356296058957,
      "grad_norm": 0.18051929771900177,
      "learning_rate": 1.7308160457346165e-05,
      "loss": 0.0647,
      "step": 2460
    },
    {
      "epoch": 0.3165652034604293,
      "grad_norm": 0.19417473673820496,
      "learning_rate": 1.7277545154491703e-05,
      "loss": 0.0653,
      "step": 2470
    },
    {
      "epoch": 0.31784684396026913,
      "grad_norm": 0.2000834196805954,
      "learning_rate": 1.724678414276949e-05,
      "loss": 0.0648,
      "step": 2480
    },
    {
      "epoch": 0.31912848446010894,
      "grad_norm": 0.19750289618968964,
      "learning_rate": 1.721587803806745e-05,
      "loss": 0.0681,
      "step": 2490
    },
    {
      "epoch": 0.32041012495994875,
      "grad_norm": 0.20135164260864258,
      "learning_rate": 1.7184827459178518e-05,
      "loss": 0.0673,
      "step": 2500
    },
    {
      "epoch": 0.32041012495994875,
      "eval_f1": 0.8547925608011445,
      "eval_loss": 0.07640008628368378,
      "eval_precision": 0.8690909090909091,
      "eval_recall": 0.8409570724841661,
      "eval_runtime": 2.9969,
      "eval_samples_per_second": 2160.934,
      "eval_steps_per_second": 8.676,
      "step": 2500
    },
    {
      "epoch": 0.3216917654597885,
      "grad_norm": 0.2127694934606552,
      "learning_rate": 1.7153633027788252e-05,
      "loss": 0.065,
      "step": 2510
    },
    {
      "epoch": 0.3229734059596283,
      "grad_norm": 0.184393048286438,
      "learning_rate": 1.7122295368462375e-05,
      "loss": 0.0679,
      "step": 2520
    },
    {
      "epoch": 0.3242550464594681,
      "grad_norm": 0.20343032479286194,
      "learning_rate": 1.7090815108634283e-05,
      "loss": 0.0662,
      "step": 2530
    },
    {
      "epoch": 0.32553668695930793,
      "grad_norm": 0.1871260553598404,
      "learning_rate": 1.7059192878592482e-05,
      "loss": 0.0661,
      "step": 2540
    },
    {
      "epoch": 0.3268183274591477,
      "grad_norm": 0.1933838278055191,
      "learning_rate": 1.7027429311467945e-05,
      "loss": 0.0689,
      "step": 2550
    },
    {
      "epoch": 0.3280999679589875,
      "grad_norm": 0.19737669825553894,
      "learning_rate": 1.699552504322147e-05,
      "loss": 0.0651,
      "step": 2560
    },
    {
      "epoch": 0.3293816084588273,
      "grad_norm": 0.17696569859981537,
      "learning_rate": 1.6963480712630936e-05,
      "loss": 0.0666,
      "step": 2570
    },
    {
      "epoch": 0.3306632489586671,
      "grad_norm": 0.20725034177303314,
      "learning_rate": 1.69312969612785e-05,
      "loss": 0.0675,
      "step": 2580
    },
    {
      "epoch": 0.33194488945850686,
      "grad_norm": 0.1974012404680252,
      "learning_rate": 1.6898974433537755e-05,
      "loss": 0.067,
      "step": 2590
    },
    {
      "epoch": 0.3332265299583467,
      "grad_norm": 0.1718105524778366,
      "learning_rate": 1.6866513776560844e-05,
      "loss": 0.0659,
      "step": 2600
    },
    {
      "epoch": 0.3345081704581865,
      "grad_norm": 0.19242900609970093,
      "learning_rate": 1.6833915640265485e-05,
      "loss": 0.0669,
      "step": 2610
    },
    {
      "epoch": 0.3357898109580263,
      "grad_norm": 0.20493321120738983,
      "learning_rate": 1.6801180677321985e-05,
      "loss": 0.0613,
      "step": 2620
    },
    {
      "epoch": 0.33707145145786604,
      "grad_norm": 0.1979932188987732,
      "learning_rate": 1.6768309543140126e-05,
      "loss": 0.0643,
      "step": 2630
    },
    {
      "epoch": 0.33835309195770585,
      "grad_norm": 0.17953254282474518,
      "learning_rate": 1.6735302895856093e-05,
      "loss": 0.0641,
      "step": 2640
    },
    {
      "epoch": 0.33963473245754566,
      "grad_norm": 0.18145443499088287,
      "learning_rate": 1.6702161396319266e-05,
      "loss": 0.0663,
      "step": 2650
    },
    {
      "epoch": 0.34091637295738547,
      "grad_norm": 0.198946014046669,
      "learning_rate": 1.666888570807899e-05,
      "loss": 0.0649,
      "step": 2660
    },
    {
      "epoch": 0.3421980134572252,
      "grad_norm": 0.17763356864452362,
      "learning_rate": 1.6635476497371314e-05,
      "loss": 0.0671,
      "step": 2670
    },
    {
      "epoch": 0.34347965395706503,
      "grad_norm": 0.2077655792236328,
      "learning_rate": 1.6601934433105612e-05,
      "loss": 0.066,
      "step": 2680
    },
    {
      "epoch": 0.34476129445690484,
      "grad_norm": 0.17525456845760345,
      "learning_rate": 1.6568260186851236e-05,
      "loss": 0.0686,
      "step": 2690
    },
    {
      "epoch": 0.34604293495674465,
      "grad_norm": 0.20130480825901031,
      "learning_rate": 1.6534454432824034e-05,
      "loss": 0.0628,
      "step": 2700
    },
    {
      "epoch": 0.3473245754565844,
      "grad_norm": 0.20377837121486664,
      "learning_rate": 1.6500517847872864e-05,
      "loss": 0.0659,
      "step": 2710
    },
    {
      "epoch": 0.3486062159564242,
      "grad_norm": 0.19869670271873474,
      "learning_rate": 1.6466451111466044e-05,
      "loss": 0.0664,
      "step": 2720
    },
    {
      "epoch": 0.349887856456264,
      "grad_norm": 0.177223801612854,
      "learning_rate": 1.643225490567775e-05,
      "loss": 0.0653,
      "step": 2730
    },
    {
      "epoch": 0.35116949695610383,
      "grad_norm": 0.17571556568145752,
      "learning_rate": 1.639792991517435e-05,
      "loss": 0.067,
      "step": 2740
    },
    {
      "epoch": 0.3524511374559436,
      "grad_norm": 0.18533557653427124,
      "learning_rate": 1.6363476827200705e-05,
      "loss": 0.067,
      "step": 2750
    },
    {
      "epoch": 0.3537327779557834,
      "grad_norm": 0.2359856367111206,
      "learning_rate": 1.6328896331566403e-05,
      "loss": 0.0689,
      "step": 2760
    },
    {
      "epoch": 0.3550144184556232,
      "grad_norm": 0.18869027495384216,
      "learning_rate": 1.6294189120631956e-05,
      "loss": 0.065,
      "step": 2770
    },
    {
      "epoch": 0.356296058955463,
      "grad_norm": 0.17907142639160156,
      "learning_rate": 1.6259355889294927e-05,
      "loss": 0.0677,
      "step": 2780
    },
    {
      "epoch": 0.35757769945530277,
      "grad_norm": 0.20352387428283691,
      "learning_rate": 1.6224397334976023e-05,
      "loss": 0.0667,
      "step": 2790
    },
    {
      "epoch": 0.3588593399551426,
      "grad_norm": 0.17456890642642975,
      "learning_rate": 1.6189314157605136e-05,
      "loss": 0.0658,
      "step": 2800
    },
    {
      "epoch": 0.3601409804549824,
      "grad_norm": 0.20184475183486938,
      "learning_rate": 1.615410705960732e-05,
      "loss": 0.0665,
      "step": 2810
    },
    {
      "epoch": 0.3614226209548222,
      "grad_norm": 0.19572722911834717,
      "learning_rate": 1.6118776745888726e-05,
      "loss": 0.0662,
      "step": 2820
    },
    {
      "epoch": 0.36270426145466195,
      "grad_norm": 0.18092940747737885,
      "learning_rate": 1.608332392382251e-05,
      "loss": 0.0619,
      "step": 2830
    },
    {
      "epoch": 0.36398590195450176,
      "grad_norm": 0.21048863232135773,
      "learning_rate": 1.6047749303234637e-05,
      "loss": 0.0647,
      "step": 2840
    },
    {
      "epoch": 0.36526754245434156,
      "grad_norm": 0.22709940373897552,
      "learning_rate": 1.601205359638969e-05,
      "loss": 0.0638,
      "step": 2850
    },
    {
      "epoch": 0.3665491829541814,
      "grad_norm": 0.17789261043071747,
      "learning_rate": 1.597623751797662e-05,
      "loss": 0.0644,
      "step": 2860
    },
    {
      "epoch": 0.3678308234540211,
      "grad_norm": 0.1861618459224701,
      "learning_rate": 1.5940301785094398e-05,
      "loss": 0.0643,
      "step": 2870
    },
    {
      "epoch": 0.36911246395386094,
      "grad_norm": 0.2651076018810272,
      "learning_rate": 1.5904247117237703e-05,
      "loss": 0.0653,
      "step": 2880
    },
    {
      "epoch": 0.37039410445370075,
      "grad_norm": 0.19479645788669586,
      "learning_rate": 1.586807423628248e-05,
      "loss": 0.0669,
      "step": 2890
    },
    {
      "epoch": 0.37167574495354055,
      "grad_norm": 0.19867412745952606,
      "learning_rate": 1.583178386647151e-05,
      "loss": 0.0644,
      "step": 2900
    },
    {
      "epoch": 0.3729573854533803,
      "grad_norm": 0.2065819799900055,
      "learning_rate": 1.57953767343999e-05,
      "loss": 0.0688,
      "step": 2910
    },
    {
      "epoch": 0.3742390259532201,
      "grad_norm": 0.16808786988258362,
      "learning_rate": 1.5758853569000536e-05,
      "loss": 0.0639,
      "step": 2920
    },
    {
      "epoch": 0.3755206664530599,
      "grad_norm": 0.19506502151489258,
      "learning_rate": 1.572221510152949e-05,
      "loss": 0.0656,
      "step": 2930
    },
    {
      "epoch": 0.37680230695289973,
      "grad_norm": 0.20851702988147736,
      "learning_rate": 1.5685462065551375e-05,
      "loss": 0.0648,
      "step": 2940
    },
    {
      "epoch": 0.3780839474527395,
      "grad_norm": 0.18304401636123657,
      "learning_rate": 1.5648595196924665e-05,
      "loss": 0.0665,
      "step": 2950
    },
    {
      "epoch": 0.3793655879525793,
      "grad_norm": 0.19657114148139954,
      "learning_rate": 1.5611615233786957e-05,
      "loss": 0.0651,
      "step": 2960
    },
    {
      "epoch": 0.3806472284524191,
      "grad_norm": 0.17807355523109436,
      "learning_rate": 1.5574522916540188e-05,
      "loss": 0.0673,
      "step": 2970
    },
    {
      "epoch": 0.3819288689522589,
      "grad_norm": 0.1849467009305954,
      "learning_rate": 1.5537318987835824e-05,
      "loss": 0.0671,
      "step": 2980
    },
    {
      "epoch": 0.38321050945209867,
      "grad_norm": 0.19102276861667633,
      "learning_rate": 1.5500004192559976e-05,
      "loss": 0.0663,
      "step": 2990
    },
    {
      "epoch": 0.3844921499519385,
      "grad_norm": 0.19463396072387695,
      "learning_rate": 1.5462579277818498e-05,
      "loss": 0.0667,
      "step": 3000
    },
    {
      "epoch": 0.3844921499519385,
      "eval_f1": 0.8683329347046161,
      "eval_loss": 0.07221938669681549,
      "eval_precision": 0.8857038302024884,
      "eval_recall": 0.851630307295332,
      "eval_runtime": 2.9907,
      "eval_samples_per_second": 2165.391,
      "eval_steps_per_second": 8.694,
      "step": 3000
    },
    {
      "epoch": 0.3857737904517783,
      "grad_norm": 0.19808001816272736,
      "learning_rate": 1.5425044992922015e-05,
      "loss": 0.0665,
      "step": 3010
    },
    {
      "epoch": 0.3870554309516181,
      "grad_norm": 0.19178569316864014,
      "learning_rate": 1.5387402089370948e-05,
      "loss": 0.0647,
      "step": 3020
    },
    {
      "epoch": 0.38833707145145785,
      "grad_norm": 0.20021072030067444,
      "learning_rate": 1.5349651320840423e-05,
      "loss": 0.0668,
      "step": 3030
    },
    {
      "epoch": 0.38961871195129766,
      "grad_norm": 0.21625490486621857,
      "learning_rate": 1.531179344316522e-05,
      "loss": 0.0659,
      "step": 3040
    },
    {
      "epoch": 0.39090035245113747,
      "grad_norm": 0.18152420222759247,
      "learning_rate": 1.527382921432464e-05,
      "loss": 0.0665,
      "step": 3050
    },
    {
      "epoch": 0.3921819929509773,
      "grad_norm": 0.2041441947221756,
      "learning_rate": 1.5235759394427287e-05,
      "loss": 0.067,
      "step": 3060
    },
    {
      "epoch": 0.39346363345081703,
      "grad_norm": 0.18800385296344757,
      "learning_rate": 1.5197584745695904e-05,
      "loss": 0.0643,
      "step": 3070
    },
    {
      "epoch": 0.39474527395065684,
      "grad_norm": 0.21839666366577148,
      "learning_rate": 1.515930603245207e-05,
      "loss": 0.0651,
      "step": 3080
    },
    {
      "epoch": 0.39602691445049665,
      "grad_norm": 0.20969919860363007,
      "learning_rate": 1.512092402110092e-05,
      "loss": 0.0671,
      "step": 3090
    },
    {
      "epoch": 0.39730855495033646,
      "grad_norm": 0.2534889578819275,
      "learning_rate": 1.508243948011579e-05,
      "loss": 0.0642,
      "step": 3100
    },
    {
      "epoch": 0.3985901954501762,
      "grad_norm": 0.18450938165187836,
      "learning_rate": 1.5043853180022838e-05,
      "loss": 0.0658,
      "step": 3110
    },
    {
      "epoch": 0.399871835950016,
      "grad_norm": 0.18956618010997772,
      "learning_rate": 1.5005165893385612e-05,
      "loss": 0.0663,
      "step": 3120
    },
    {
      "epoch": 0.40115347644985583,
      "grad_norm": 0.16871580481529236,
      "learning_rate": 1.4966378394789581e-05,
      "loss": 0.0658,
      "step": 3130
    },
    {
      "epoch": 0.40243511694969564,
      "grad_norm": 0.18960537016391754,
      "learning_rate": 1.4927491460826626e-05,
      "loss": 0.0653,
      "step": 3140
    },
    {
      "epoch": 0.4037167574495354,
      "grad_norm": 0.20116572082042694,
      "learning_rate": 1.48885058700795e-05,
      "loss": 0.0664,
      "step": 3150
    },
    {
      "epoch": 0.4049983979493752,
      "grad_norm": 0.2119298279285431,
      "learning_rate": 1.4849422403106228e-05,
      "loss": 0.0649,
      "step": 3160
    },
    {
      "epoch": 0.406280038449215,
      "grad_norm": 0.19871027767658234,
      "learning_rate": 1.4810241842424491e-05,
      "loss": 0.0669,
      "step": 3170
    },
    {
      "epoch": 0.4075616789490548,
      "grad_norm": 0.19416335225105286,
      "learning_rate": 1.4770964972495938e-05,
      "loss": 0.0628,
      "step": 3180
    },
    {
      "epoch": 0.40884331944889457,
      "grad_norm": 0.18964780867099762,
      "learning_rate": 1.4731592579710507e-05,
      "loss": 0.0634,
      "step": 3190
    },
    {
      "epoch": 0.4101249599487344,
      "grad_norm": 0.18982967734336853,
      "learning_rate": 1.4692125452370664e-05,
      "loss": 0.0651,
      "step": 3200
    },
    {
      "epoch": 0.4114066004485742,
      "grad_norm": 0.20831823348999023,
      "learning_rate": 1.4652564380675616e-05,
      "loss": 0.0671,
      "step": 3210
    },
    {
      "epoch": 0.41268824094841394,
      "grad_norm": 0.190037801861763,
      "learning_rate": 1.4612910156705497e-05,
      "loss": 0.0697,
      "step": 3220
    },
    {
      "epoch": 0.41396988144825375,
      "grad_norm": 0.20448094606399536,
      "learning_rate": 1.457316357440552e-05,
      "loss": 0.0656,
      "step": 3230
    },
    {
      "epoch": 0.41525152194809356,
      "grad_norm": 0.18915881216526031,
      "learning_rate": 1.4533325429570052e-05,
      "loss": 0.0651,
      "step": 3240
    },
    {
      "epoch": 0.41653316244793337,
      "grad_norm": 0.19686464965343475,
      "learning_rate": 1.4493396519826717e-05,
      "loss": 0.0661,
      "step": 3250
    },
    {
      "epoch": 0.4178148029477731,
      "grad_norm": 0.17895036935806274,
      "learning_rate": 1.4453377644620389e-05,
      "loss": 0.0677,
      "step": 3260
    },
    {
      "epoch": 0.41909644344761293,
      "grad_norm": 0.17503611743450165,
      "learning_rate": 1.441326960519721e-05,
      "loss": 0.0644,
      "step": 3270
    },
    {
      "epoch": 0.42037808394745274,
      "grad_norm": 0.18629562854766846,
      "learning_rate": 1.4373073204588556e-05,
      "loss": 0.0661,
      "step": 3280
    },
    {
      "epoch": 0.42165972444729255,
      "grad_norm": 0.17520345747470856,
      "learning_rate": 1.4332789247594923e-05,
      "loss": 0.0661,
      "step": 3290
    },
    {
      "epoch": 0.4229413649471323,
      "grad_norm": 0.1670594960451126,
      "learning_rate": 1.4292418540769845e-05,
      "loss": 0.0643,
      "step": 3300
    },
    {
      "epoch": 0.4242230054469721,
      "grad_norm": 0.17656877636909485,
      "learning_rate": 1.425196189240374e-05,
      "loss": 0.0654,
      "step": 3310
    },
    {
      "epoch": 0.4255046459468119,
      "grad_norm": 0.20574980974197388,
      "learning_rate": 1.4211420112507714e-05,
      "loss": 0.0665,
      "step": 3320
    },
    {
      "epoch": 0.42678628644665173,
      "grad_norm": 0.17492710053920746,
      "learning_rate": 1.4170794012797357e-05,
      "loss": 0.0669,
      "step": 3330
    },
    {
      "epoch": 0.4280679269464915,
      "grad_norm": 0.18844270706176758,
      "learning_rate": 1.4130084406676488e-05,
      "loss": 0.0644,
      "step": 3340
    },
    {
      "epoch": 0.4293495674463313,
      "grad_norm": 0.18853574991226196,
      "learning_rate": 1.4089292109220852e-05,
      "loss": 0.0654,
      "step": 3350
    },
    {
      "epoch": 0.4306312079461711,
      "grad_norm": 0.19601227343082428,
      "learning_rate": 1.4048417937161833e-05,
      "loss": 0.0658,
      "step": 3360
    },
    {
      "epoch": 0.4319128484460109,
      "grad_norm": 0.19078759849071503,
      "learning_rate": 1.4007462708870076e-05,
      "loss": 0.0655,
      "step": 3370
    },
    {
      "epoch": 0.43319448894585066,
      "grad_norm": 0.187064990401268,
      "learning_rate": 1.3966427244339111e-05,
      "loss": 0.0656,
      "step": 3380
    },
    {
      "epoch": 0.4344761294456905,
      "grad_norm": 0.17603109776973724,
      "learning_rate": 1.3925312365168934e-05,
      "loss": 0.0629,
      "step": 3390
    },
    {
      "epoch": 0.4357577699455303,
      "grad_norm": 0.19391457736492157,
      "learning_rate": 1.3884118894549562e-05,
      "loss": 0.0634,
      "step": 3400
    },
    {
      "epoch": 0.4370394104453701,
      "grad_norm": 0.19418643414974213,
      "learning_rate": 1.3842847657244535e-05,
      "loss": 0.064,
      "step": 3410
    },
    {
      "epoch": 0.43832105094520984,
      "grad_norm": 0.21033746004104614,
      "learning_rate": 1.3801499479574431e-05,
      "loss": 0.0648,
      "step": 3420
    },
    {
      "epoch": 0.43960269144504965,
      "grad_norm": 0.18090617656707764,
      "learning_rate": 1.3760075189400295e-05,
      "loss": 0.0658,
      "step": 3430
    },
    {
      "epoch": 0.44088433194488946,
      "grad_norm": 0.19315758347511292,
      "learning_rate": 1.3718575616107072e-05,
      "loss": 0.0673,
      "step": 3440
    },
    {
      "epoch": 0.44216597244472927,
      "grad_norm": 0.21332131326198578,
      "learning_rate": 1.3677001590587011e-05,
      "loss": 0.0653,
      "step": 3450
    },
    {
      "epoch": 0.443447612944569,
      "grad_norm": 0.16083082556724548,
      "learning_rate": 1.3635353945223022e-05,
      "loss": 0.0618,
      "step": 3460
    },
    {
      "epoch": 0.44472925344440883,
      "grad_norm": 0.1910344362258911,
      "learning_rate": 1.3593633513872e-05,
      "loss": 0.063,
      "step": 3470
    },
    {
      "epoch": 0.44601089394424864,
      "grad_norm": 0.21517983078956604,
      "learning_rate": 1.3551841131848144e-05,
      "loss": 0.0632,
      "step": 3480
    },
    {
      "epoch": 0.44729253444408845,
      "grad_norm": 0.18821708858013153,
      "learning_rate": 1.3509977635906241e-05,
      "loss": 0.065,
      "step": 3490
    },
    {
      "epoch": 0.4485741749439282,
      "grad_norm": 0.19319505989551544,
      "learning_rate": 1.3468043864224878e-05,
      "loss": 0.0649,
      "step": 3500
    },
    {
      "epoch": 0.4485741749439282,
      "eval_f1": 0.8754380754380754,
      "eval_loss": 0.07064414024353027,
      "eval_precision": 0.8868696594054639,
      "eval_recall": 0.8642974431151771,
      "eval_runtime": 2.9906,
      "eval_samples_per_second": 2165.465,
      "eval_steps_per_second": 8.694,
      "step": 3500
    },
    {
      "epoch": 0.449855815443768,
      "grad_norm": 0.19362811744213104,
      "learning_rate": 1.3426040656389696e-05,
      "loss": 0.0659,
      "step": 3510
    },
    {
      "epoch": 0.4511374559436078,
      "grad_norm": 0.19557055830955505,
      "learning_rate": 1.3383968853376568e-05,
      "loss": 0.0674,
      "step": 3520
    },
    {
      "epoch": 0.45241909644344763,
      "grad_norm": 0.17842736840248108,
      "learning_rate": 1.3341829297534745e-05,
      "loss": 0.0647,
      "step": 3530
    },
    {
      "epoch": 0.4537007369432874,
      "grad_norm": 0.19172975420951843,
      "learning_rate": 1.3299622832570025e-05,
      "loss": 0.0653,
      "step": 3540
    },
    {
      "epoch": 0.4549823774431272,
      "grad_norm": 0.17918632924556732,
      "learning_rate": 1.3257350303527829e-05,
      "loss": 0.0662,
      "step": 3550
    },
    {
      "epoch": 0.456264017942967,
      "grad_norm": 0.1870853751897812,
      "learning_rate": 1.3215012556776287e-05,
      "loss": 0.0666,
      "step": 3560
    },
    {
      "epoch": 0.4575456584428068,
      "grad_norm": 0.18598654866218567,
      "learning_rate": 1.317261043998932e-05,
      "loss": 0.0645,
      "step": 3570
    },
    {
      "epoch": 0.45882729894264657,
      "grad_norm": 0.20888212323188782,
      "learning_rate": 1.3130144802129624e-05,
      "loss": 0.0631,
      "step": 3580
    },
    {
      "epoch": 0.4601089394424864,
      "grad_norm": 0.176805317401886,
      "learning_rate": 1.3087616493431705e-05,
      "loss": 0.0645,
      "step": 3590
    },
    {
      "epoch": 0.4613905799423262,
      "grad_norm": 0.17067846655845642,
      "learning_rate": 1.3045026365384854e-05,
      "loss": 0.0663,
      "step": 3600
    },
    {
      "epoch": 0.462672220442166,
      "grad_norm": 0.18683072924613953,
      "learning_rate": 1.3002375270716076e-05,
      "loss": 0.0629,
      "step": 3610
    },
    {
      "epoch": 0.46395386094200575,
      "grad_norm": 0.16378134489059448,
      "learning_rate": 1.2959664063373044e-05,
      "loss": 0.0653,
      "step": 3620
    },
    {
      "epoch": 0.46523550144184556,
      "grad_norm": 0.1760198473930359,
      "learning_rate": 1.2916893598506981e-05,
      "loss": 0.0655,
      "step": 3630
    },
    {
      "epoch": 0.46651714194168536,
      "grad_norm": 0.1707535684108734,
      "learning_rate": 1.2874064732455552e-05,
      "loss": 0.0631,
      "step": 3640
    },
    {
      "epoch": 0.4677987824415252,
      "grad_norm": 0.18154825270175934,
      "learning_rate": 1.2831178322725706e-05,
      "loss": 0.0657,
      "step": 3650
    },
    {
      "epoch": 0.4690804229413649,
      "grad_norm": 0.20046283304691315,
      "learning_rate": 1.2788235227976529e-05,
      "loss": 0.065,
      "step": 3660
    },
    {
      "epoch": 0.47036206344120474,
      "grad_norm": 0.19358018040657043,
      "learning_rate": 1.2745236308002018e-05,
      "loss": 0.066,
      "step": 3670
    },
    {
      "epoch": 0.47164370394104455,
      "grad_norm": 0.185184046626091,
      "learning_rate": 1.2702182423713904e-05,
      "loss": 0.0656,
      "step": 3680
    },
    {
      "epoch": 0.47292534444088435,
      "grad_norm": 0.2039334625005722,
      "learning_rate": 1.2659074437124388e-05,
      "loss": 0.0643,
      "step": 3690
    },
    {
      "epoch": 0.4742069849407241,
      "grad_norm": 0.2065436691045761,
      "learning_rate": 1.2615913211328894e-05,
      "loss": 0.0652,
      "step": 3700
    },
    {
      "epoch": 0.4754886254405639,
      "grad_norm": 0.17350079119205475,
      "learning_rate": 1.2572699610488783e-05,
      "loss": 0.0654,
      "step": 3710
    },
    {
      "epoch": 0.4767702659404037,
      "grad_norm": 0.20188377797603607,
      "learning_rate": 1.2529434499814058e-05,
      "loss": 0.0653,
      "step": 3720
    },
    {
      "epoch": 0.47805190644024353,
      "grad_norm": 0.18335194885730743,
      "learning_rate": 1.2486118745546035e-05,
      "loss": 0.0655,
      "step": 3730
    },
    {
      "epoch": 0.4793335469400833,
      "grad_norm": 0.194752037525177,
      "learning_rate": 1.244275321494e-05,
      "loss": 0.0649,
      "step": 3740
    },
    {
      "epoch": 0.4806151874399231,
      "grad_norm": 0.20098841190338135,
      "learning_rate": 1.2399338776247846e-05,
      "loss": 0.0641,
      "step": 3750
    },
    {
      "epoch": 0.4818968279397629,
      "grad_norm": 0.2057989090681076,
      "learning_rate": 1.2355876298700693e-05,
      "loss": 0.0643,
      "step": 3760
    },
    {
      "epoch": 0.4831784684396027,
      "grad_norm": 0.18653416633605957,
      "learning_rate": 1.2312366652491476e-05,
      "loss": 0.0635,
      "step": 3770
    },
    {
      "epoch": 0.48446010893944247,
      "grad_norm": 0.19041769206523895,
      "learning_rate": 1.2268810708757533e-05,
      "loss": 0.065,
      "step": 3780
    },
    {
      "epoch": 0.4857417494392823,
      "grad_norm": 0.19951310753822327,
      "learning_rate": 1.2225209339563144e-05,
      "loss": 0.0641,
      "step": 3790
    },
    {
      "epoch": 0.4870233899391221,
      "grad_norm": 0.1979627162218094,
      "learning_rate": 1.2181563417882104e-05,
      "loss": 0.0647,
      "step": 3800
    },
    {
      "epoch": 0.4883050304389619,
      "grad_norm": 0.20096543431282043,
      "learning_rate": 1.2137873817580213e-05,
      "loss": 0.0647,
      "step": 3810
    },
    {
      "epoch": 0.48958667093880165,
      "grad_norm": 0.19415560364723206,
      "learning_rate": 1.2094141413397785e-05,
      "loss": 0.0645,
      "step": 3820
    },
    {
      "epoch": 0.49086831143864146,
      "grad_norm": 0.180967316031456,
      "learning_rate": 1.2050367080932159e-05,
      "loss": 0.0629,
      "step": 3830
    },
    {
      "epoch": 0.49214995193848127,
      "grad_norm": 0.18070431053638458,
      "learning_rate": 1.2006551696620135e-05,
      "loss": 0.0649,
      "step": 3840
    },
    {
      "epoch": 0.4934315924383211,
      "grad_norm": 0.174631729722023,
      "learning_rate": 1.1962696137720443e-05,
      "loss": 0.0649,
      "step": 3850
    },
    {
      "epoch": 0.49471323293816083,
      "grad_norm": 0.1981905698776245,
      "learning_rate": 1.1918801282296184e-05,
      "loss": 0.0648,
      "step": 3860
    },
    {
      "epoch": 0.49599487343800064,
      "grad_norm": 0.19266869127750397,
      "learning_rate": 1.1874868009197236e-05,
      "loss": 0.0653,
      "step": 3870
    },
    {
      "epoch": 0.49727651393784045,
      "grad_norm": 0.19425834715366364,
      "learning_rate": 1.1830897198042662e-05,
      "loss": 0.0646,
      "step": 3880
    },
    {
      "epoch": 0.49855815443768026,
      "grad_norm": 0.20095306634902954,
      "learning_rate": 1.1786889729203113e-05,
      "loss": 0.0651,
      "step": 3890
    },
    {
      "epoch": 0.49983979493752,
      "grad_norm": 0.16680414974689484,
      "learning_rate": 1.1742846483783177e-05,
      "loss": 0.0633,
      "step": 3900
    },
    {
      "epoch": 0.5011214354373599,
      "grad_norm": 0.2134004682302475,
      "learning_rate": 1.1698768343603753e-05,
      "loss": 0.066,
      "step": 3910
    },
    {
      "epoch": 0.5024030759371996,
      "grad_norm": 0.19804078340530396,
      "learning_rate": 1.16546561911844e-05,
      "loss": 0.0654,
      "step": 3920
    },
    {
      "epoch": 0.5036847164370394,
      "grad_norm": 0.1972855180501938,
      "learning_rate": 1.1610510909725644e-05,
      "loss": 0.0654,
      "step": 3930
    },
    {
      "epoch": 0.5049663569368792,
      "grad_norm": 0.18941302597522736,
      "learning_rate": 1.1566333383091333e-05,
      "loss": 0.0656,
      "step": 3940
    },
    {
      "epoch": 0.506247997436719,
      "grad_norm": 0.2275722473859787,
      "learning_rate": 1.15221244957909e-05,
      "loss": 0.0642,
      "step": 3950
    },
    {
      "epoch": 0.5075296379365588,
      "grad_norm": 0.19969482719898224,
      "learning_rate": 1.1477885132961679e-05,
      "loss": 0.0657,
      "step": 3960
    },
    {
      "epoch": 0.5088112784363986,
      "grad_norm": 0.1895623505115509,
      "learning_rate": 1.1433616180351176e-05,
      "loss": 0.0653,
      "step": 3970
    },
    {
      "epoch": 0.5100929189362384,
      "grad_norm": 0.19780370593070984,
      "learning_rate": 1.1389318524299332e-05,
      "loss": 0.0646,
      "step": 3980
    },
    {
      "epoch": 0.5113745594360782,
      "grad_norm": 0.18038791418075562,
      "learning_rate": 1.1344993051720792e-05,
      "loss": 0.0642,
      "step": 3990
    },
    {
      "epoch": 0.5126561999359179,
      "grad_norm": 0.2090374231338501,
      "learning_rate": 1.1300640650087114e-05,
      "loss": 0.0649,
      "step": 4000
    },
    {
      "epoch": 0.5126561999359179,
      "eval_f1": 0.8851620576865894,
      "eval_loss": 0.06776601821184158,
      "eval_precision": 0.8978163831584027,
      "eval_recall": 0.8728594886230354,
      "eval_runtime": 2.9842,
      "eval_samples_per_second": 2170.11,
      "eval_steps_per_second": 8.713,
      "step": 4000
    },
    {
      "epoch": 0.5139378404357577,
      "grad_norm": 0.21133767068386078,
      "learning_rate": 1.1256262207409038e-05,
      "loss": 0.0651,
      "step": 4010
    },
    {
      "epoch": 0.5152194809355976,
      "grad_norm": 0.20390534400939941,
      "learning_rate": 1.1211858612218689e-05,
      "loss": 0.0661,
      "step": 4020
    },
    {
      "epoch": 0.5165011214354374,
      "grad_norm": 0.16816197335720062,
      "learning_rate": 1.1167430753551779e-05,
      "loss": 0.0642,
      "step": 4030
    },
    {
      "epoch": 0.5177827619352772,
      "grad_norm": 0.17969034612178802,
      "learning_rate": 1.1122979520929826e-05,
      "loss": 0.0619,
      "step": 4040
    },
    {
      "epoch": 0.519064402435117,
      "grad_norm": 0.17689917981624603,
      "learning_rate": 1.1078505804342327e-05,
      "loss": 0.0659,
      "step": 4050
    },
    {
      "epoch": 0.5203460429349568,
      "grad_norm": 0.19302503764629364,
      "learning_rate": 1.1034010494228951e-05,
      "loss": 0.0631,
      "step": 4060
    },
    {
      "epoch": 0.5216276834347965,
      "grad_norm": 0.21062147617340088,
      "learning_rate": 1.0989494481461707e-05,
      "loss": 0.0665,
      "step": 4070
    },
    {
      "epoch": 0.5229093239346363,
      "grad_norm": 0.18397943675518036,
      "learning_rate": 1.0944958657327101e-05,
      "loss": 0.064,
      "step": 4080
    },
    {
      "epoch": 0.5241909644344761,
      "grad_norm": 0.18781358003616333,
      "learning_rate": 1.0900403913508301e-05,
      "loss": 0.0657,
      "step": 4090
    },
    {
      "epoch": 0.5254726049343159,
      "grad_norm": 0.20603561401367188,
      "learning_rate": 1.085583114206728e-05,
      "loss": 0.0626,
      "step": 4100
    },
    {
      "epoch": 0.5267542454341557,
      "grad_norm": 0.18262669444084167,
      "learning_rate": 1.0811241235426947e-05,
      "loss": 0.0626,
      "step": 4110
    },
    {
      "epoch": 0.5280358859339955,
      "grad_norm": 0.18097414076328278,
      "learning_rate": 1.0766635086353298e-05,
      "loss": 0.0626,
      "step": 4120
    },
    {
      "epoch": 0.5293175264338353,
      "grad_norm": 0.1961183249950409,
      "learning_rate": 1.0722013587937528e-05,
      "loss": 0.063,
      "step": 4130
    },
    {
      "epoch": 0.5305991669336751,
      "grad_norm": 0.18248215317726135,
      "learning_rate": 1.0677377633578137e-05,
      "loss": 0.0656,
      "step": 4140
    },
    {
      "epoch": 0.5318808074335148,
      "grad_norm": 0.17086833715438843,
      "learning_rate": 1.0632728116963082e-05,
      "loss": 0.0644,
      "step": 4150
    },
    {
      "epoch": 0.5331624479333547,
      "grad_norm": 0.19590778648853302,
      "learning_rate": 1.0588065932051843e-05,
      "loss": 0.0638,
      "step": 4160
    },
    {
      "epoch": 0.5344440884331945,
      "grad_norm": 0.18840567767620087,
      "learning_rate": 1.0543391973057537e-05,
      "loss": 0.0639,
      "step": 4170
    },
    {
      "epoch": 0.5357257289330343,
      "grad_norm": 0.18893346190452576,
      "learning_rate": 1.049870713442904e-05,
      "loss": 0.0607,
      "step": 4180
    },
    {
      "epoch": 0.5370073694328741,
      "grad_norm": 0.19564594328403473,
      "learning_rate": 1.0454012310833034e-05,
      "loss": 0.0649,
      "step": 4190
    },
    {
      "epoch": 0.5382890099327139,
      "grad_norm": 0.19073913991451263,
      "learning_rate": 1.0409308397136128e-05,
      "loss": 0.0652,
      "step": 4200
    },
    {
      "epoch": 0.5395706504325537,
      "grad_norm": 0.1852230280637741,
      "learning_rate": 1.036459628838693e-05,
      "loss": 0.0643,
      "step": 4210
    },
    {
      "epoch": 0.5408522909323935,
      "grad_norm": 0.18030205368995667,
      "learning_rate": 1.0319876879798123e-05,
      "loss": 0.065,
      "step": 4220
    },
    {
      "epoch": 0.5421339314322332,
      "grad_norm": 0.19292110204696655,
      "learning_rate": 1.0275151066728548e-05,
      "loss": 0.061,
      "step": 4230
    },
    {
      "epoch": 0.543415571932073,
      "grad_norm": 0.1795249581336975,
      "learning_rate": 1.0230419744665278e-05,
      "loss": 0.0648,
      "step": 4240
    },
    {
      "epoch": 0.5446972124319128,
      "grad_norm": 0.18629731237888336,
      "learning_rate": 1.0185683809205675e-05,
      "loss": 0.0652,
      "step": 4250
    },
    {
      "epoch": 0.5459788529317526,
      "grad_norm": 0.19541800022125244,
      "learning_rate": 1.0140944156039481e-05,
      "loss": 0.0656,
      "step": 4260
    },
    {
      "epoch": 0.5472604934315924,
      "grad_norm": 0.201457679271698,
      "learning_rate": 1.0096201680930867e-05,
      "loss": 0.0651,
      "step": 4270
    },
    {
      "epoch": 0.5485421339314323,
      "grad_norm": 0.20298662781715393,
      "learning_rate": 1.0051457279700502e-05,
      "loss": 0.0657,
      "step": 4280
    },
    {
      "epoch": 0.5498237744312721,
      "grad_norm": 0.2091987133026123,
      "learning_rate": 1.0006711848207625e-05,
      "loss": 0.0642,
      "step": 4290
    },
    {
      "epoch": 0.5511054149311119,
      "grad_norm": 0.195583313703537,
      "learning_rate": 9.961966282332094e-06,
      "loss": 0.064,
      "step": 4300
    },
    {
      "epoch": 0.5523870554309516,
      "grad_norm": 0.16657018661499023,
      "learning_rate": 9.917221477956472e-06,
      "loss": 0.0629,
      "step": 4310
    },
    {
      "epoch": 0.5536686959307914,
      "grad_norm": 0.18883733451366425,
      "learning_rate": 9.87247833094806e-06,
      "loss": 0.0645,
      "step": 4320
    },
    {
      "epoch": 0.5549503364306312,
      "grad_norm": 0.18239256739616394,
      "learning_rate": 9.827737737140983e-06,
      "loss": 0.0639,
      "step": 4330
    },
    {
      "epoch": 0.556231976930471,
      "grad_norm": 0.20366910099983215,
      "learning_rate": 9.783000592318249e-06,
      "loss": 0.0649,
      "step": 4340
    },
    {
      "epoch": 0.5575136174303108,
      "grad_norm": 0.19524288177490234,
      "learning_rate": 9.738267792193806e-06,
      "loss": 0.0636,
      "step": 4350
    },
    {
      "epoch": 0.5587952579301506,
      "grad_norm": 0.19309386610984802,
      "learning_rate": 9.693540232394613e-06,
      "loss": 0.0648,
      "step": 4360
    },
    {
      "epoch": 0.5600768984299904,
      "grad_norm": 0.2024122029542923,
      "learning_rate": 9.648818808442716e-06,
      "loss": 0.0648,
      "step": 4370
    },
    {
      "epoch": 0.5613585389298302,
      "grad_norm": 0.1884394735097885,
      "learning_rate": 9.604104415737309e-06,
      "loss": 0.0656,
      "step": 4380
    },
    {
      "epoch": 0.5626401794296699,
      "grad_norm": 0.20835010707378387,
      "learning_rate": 9.559397949536799e-06,
      "loss": 0.0632,
      "step": 4390
    },
    {
      "epoch": 0.5639218199295097,
      "grad_norm": 0.189108744263649,
      "learning_rate": 9.514700304940901e-06,
      "loss": 0.0635,
      "step": 4400
    },
    {
      "epoch": 0.5652034604293495,
      "grad_norm": 0.1688823103904724,
      "learning_rate": 9.470012376872706e-06,
      "loss": 0.0644,
      "step": 4410
    },
    {
      "epoch": 0.5664851009291894,
      "grad_norm": 0.19286887347698212,
      "learning_rate": 9.425335060060757e-06,
      "loss": 0.0637,
      "step": 4420
    },
    {
      "epoch": 0.5677667414290292,
      "grad_norm": 0.18644267320632935,
      "learning_rate": 9.380669249021146e-06,
      "loss": 0.061,
      "step": 4430
    },
    {
      "epoch": 0.569048381928869,
      "grad_norm": 0.2195032835006714,
      "learning_rate": 9.336015838039598e-06,
      "loss": 0.0663,
      "step": 4440
    },
    {
      "epoch": 0.5703300224287088,
      "grad_norm": 0.20251427590847015,
      "learning_rate": 9.291375721153579e-06,
      "loss": 0.066,
      "step": 4450
    },
    {
      "epoch": 0.5716116629285486,
      "grad_norm": 0.1976834088563919,
      "learning_rate": 9.24674979213436e-06,
      "loss": 0.0632,
      "step": 4460
    },
    {
      "epoch": 0.5728933034283883,
      "grad_norm": 0.18137118220329285,
      "learning_rate": 9.202138944469168e-06,
      "loss": 0.0637,
      "step": 4470
    },
    {
      "epoch": 0.5741749439282281,
      "grad_norm": 0.18933522701263428,
      "learning_rate": 9.157544071343275e-06,
      "loss": 0.0635,
      "step": 4480
    },
    {
      "epoch": 0.5754565844280679,
      "grad_norm": 0.20421184599399567,
      "learning_rate": 9.112966065622095e-06,
      "loss": 0.0649,
      "step": 4490
    },
    {
      "epoch": 0.5767382249279077,
      "grad_norm": 0.18742768466472626,
      "learning_rate": 9.068405819833352e-06,
      "loss": 0.0639,
      "step": 4500
    },
    {
      "epoch": 0.5767382249279077,
      "eval_f1": 0.8894003777148253,
      "eval_loss": 0.06641285866498947,
      "eval_precision": 0.8951057258256118,
      "eval_recall": 0.8837673000234577,
      "eval_runtime": 2.9904,
      "eval_samples_per_second": 2165.589,
      "eval_steps_per_second": 8.694,
      "step": 4500
    },
    {
      "epoch": 0.5780198654277475,
      "grad_norm": 0.17334450781345367,
      "learning_rate": 9.02386422614918e-06,
      "loss": 0.0633,
      "step": 4510
    },
    {
      "epoch": 0.5793015059275873,
      "grad_norm": 0.16847829520702362,
      "learning_rate": 8.979342176368247e-06,
      "loss": 0.0655,
      "step": 4520
    },
    {
      "epoch": 0.5805831464274271,
      "grad_norm": 0.17423571646213531,
      "learning_rate": 8.934840561897944e-06,
      "loss": 0.0652,
      "step": 4530
    },
    {
      "epoch": 0.581864786927267,
      "grad_norm": 0.18410921096801758,
      "learning_rate": 8.890360273736504e-06,
      "loss": 0.0607,
      "step": 4540
    },
    {
      "epoch": 0.5831464274271067,
      "grad_norm": 0.1774194836616516,
      "learning_rate": 8.84590220245516e-06,
      "loss": 0.0645,
      "step": 4550
    },
    {
      "epoch": 0.5844280679269465,
      "grad_norm": 0.17994000017642975,
      "learning_rate": 8.80146723818034e-06,
      "loss": 0.0653,
      "step": 4560
    },
    {
      "epoch": 0.5857097084267863,
      "grad_norm": 0.1958846151828766,
      "learning_rate": 8.757056270575823e-06,
      "loss": 0.0634,
      "step": 4570
    },
    {
      "epoch": 0.5869913489266261,
      "grad_norm": 0.19152416288852692,
      "learning_rate": 8.712670188824937e-06,
      "loss": 0.0617,
      "step": 4580
    },
    {
      "epoch": 0.5882729894264659,
      "grad_norm": 0.20121438801288605,
      "learning_rate": 8.668309881612746e-06,
      "loss": 0.0635,
      "step": 4590
    },
    {
      "epoch": 0.5895546299263057,
      "grad_norm": 0.2359953075647354,
      "learning_rate": 8.623976237108271e-06,
      "loss": 0.0615,
      "step": 4600
    },
    {
      "epoch": 0.5908362704261455,
      "grad_norm": 0.17650754749774933,
      "learning_rate": 8.579670142946701e-06,
      "loss": 0.065,
      "step": 4610
    },
    {
      "epoch": 0.5921179109259853,
      "grad_norm": 0.17972248792648315,
      "learning_rate": 8.53539248621161e-06,
      "loss": 0.0648,
      "step": 4620
    },
    {
      "epoch": 0.593399551425825,
      "grad_norm": 0.17738933861255646,
      "learning_rate": 8.491144153417217e-06,
      "loss": 0.0676,
      "step": 4630
    },
    {
      "epoch": 0.5946811919256648,
      "grad_norm": 0.19836388528347015,
      "learning_rate": 8.446926030490623e-06,
      "loss": 0.0643,
      "step": 4640
    },
    {
      "epoch": 0.5959628324255046,
      "grad_norm": 0.1889767348766327,
      "learning_rate": 8.40273900275407e-06,
      "loss": 0.065,
      "step": 4650
    },
    {
      "epoch": 0.5972444729253444,
      "grad_norm": 0.1591385453939438,
      "learning_rate": 8.358583954907228e-06,
      "loss": 0.063,
      "step": 4660
    },
    {
      "epoch": 0.5985261134251842,
      "grad_norm": 0.1870860904455185,
      "learning_rate": 8.314461771009478e-06,
      "loss": 0.0629,
      "step": 4670
    },
    {
      "epoch": 0.5998077539250241,
      "grad_norm": 0.18506884574890137,
      "learning_rate": 8.270373334462193e-06,
      "loss": 0.0649,
      "step": 4680
    },
    {
      "epoch": 0.6010893944248639,
      "grad_norm": 0.15451203286647797,
      "learning_rate": 8.226319527991088e-06,
      "loss": 0.0628,
      "step": 4690
    },
    {
      "epoch": 0.6023710349247037,
      "grad_norm": 0.1646854281425476,
      "learning_rate": 8.182301233628506e-06,
      "loss": 0.063,
      "step": 4700
    },
    {
      "epoch": 0.6036526754245434,
      "grad_norm": 0.18889601528644562,
      "learning_rate": 8.138319332695788e-06,
      "loss": 0.0639,
      "step": 4710
    },
    {
      "epoch": 0.6049343159243832,
      "grad_norm": 0.18762964010238647,
      "learning_rate": 8.094374705785613e-06,
      "loss": 0.0647,
      "step": 4720
    },
    {
      "epoch": 0.606215956424223,
      "grad_norm": 0.17451703548431396,
      "learning_rate": 8.050468232744367e-06,
      "loss": 0.0619,
      "step": 4730
    },
    {
      "epoch": 0.6074975969240628,
      "grad_norm": 0.1720801442861557,
      "learning_rate": 8.006600792654546e-06,
      "loss": 0.0631,
      "step": 4740
    },
    {
      "epoch": 0.6087792374239026,
      "grad_norm": 0.19186915457248688,
      "learning_rate": 7.962773263817114e-06,
      "loss": 0.064,
      "step": 4750
    },
    {
      "epoch": 0.6100608779237424,
      "grad_norm": 0.1681632250547409,
      "learning_rate": 7.918986523733964e-06,
      "loss": 0.0635,
      "step": 4760
    },
    {
      "epoch": 0.6113425184235822,
      "grad_norm": 0.17930343747138977,
      "learning_rate": 7.875241449090327e-06,
      "loss": 0.0637,
      "step": 4770
    },
    {
      "epoch": 0.612624158923422,
      "grad_norm": 0.20653793215751648,
      "learning_rate": 7.831538915737204e-06,
      "loss": 0.0648,
      "step": 4780
    },
    {
      "epoch": 0.6139057994232617,
      "grad_norm": 0.18635816872119904,
      "learning_rate": 7.787879798673869e-06,
      "loss": 0.0652,
      "step": 4790
    },
    {
      "epoch": 0.6151874399231015,
      "grad_norm": 0.19335713982582092,
      "learning_rate": 7.744264972030319e-06,
      "loss": 0.0616,
      "step": 4800
    },
    {
      "epoch": 0.6164690804229414,
      "grad_norm": 0.18472839891910553,
      "learning_rate": 7.700695309049768e-06,
      "loss": 0.0644,
      "step": 4810
    },
    {
      "epoch": 0.6177507209227812,
      "grad_norm": 0.1953488439321518,
      "learning_rate": 7.657171682071198e-06,
      "loss": 0.0649,
      "step": 4820
    },
    {
      "epoch": 0.619032361422621,
      "grad_norm": 0.19706253707408905,
      "learning_rate": 7.613694962511863e-06,
      "loss": 0.0636,
      "step": 4830
    },
    {
      "epoch": 0.6203140019224608,
      "grad_norm": 0.19095227122306824,
      "learning_rate": 7.57026602084984e-06,
      "loss": 0.0608,
      "step": 4840
    },
    {
      "epoch": 0.6215956424223006,
      "grad_norm": 0.18028537929058075,
      "learning_rate": 7.52688572660663e-06,
      "loss": 0.0618,
      "step": 4850
    },
    {
      "epoch": 0.6228772829221403,
      "grad_norm": 0.19275839626789093,
      "learning_rate": 7.483554948329716e-06,
      "loss": 0.0629,
      "step": 4860
    },
    {
      "epoch": 0.6241589234219801,
      "grad_norm": 0.18970589339733124,
      "learning_rate": 7.440274553575194e-06,
      "loss": 0.0636,
      "step": 4870
    },
    {
      "epoch": 0.6254405639218199,
      "grad_norm": 0.18009796738624573,
      "learning_rate": 7.397045408890388e-06,
      "loss": 0.0649,
      "step": 4880
    },
    {
      "epoch": 0.6267222044216597,
      "grad_norm": 0.1802019476890564,
      "learning_rate": 7.353868379796518e-06,
      "loss": 0.0642,
      "step": 4890
    },
    {
      "epoch": 0.6280038449214995,
      "grad_norm": 0.1904575377702713,
      "learning_rate": 7.3107443307713555e-06,
      "loss": 0.064,
      "step": 4900
    },
    {
      "epoch": 0.6292854854213393,
      "grad_norm": 0.17662177979946136,
      "learning_rate": 7.267674125231919e-06,
      "loss": 0.0643,
      "step": 4910
    },
    {
      "epoch": 0.6305671259211791,
      "grad_norm": 0.20192337036132812,
      "learning_rate": 7.224658625517191e-06,
      "loss": 0.0658,
      "step": 4920
    },
    {
      "epoch": 0.631848766421019,
      "grad_norm": 0.2027018964290619,
      "learning_rate": 7.181698692870854e-06,
      "loss": 0.0633,
      "step": 4930
    },
    {
      "epoch": 0.6331304069208586,
      "grad_norm": 0.18302646279335022,
      "learning_rate": 7.138795187424035e-06,
      "loss": 0.0649,
      "step": 4940
    },
    {
      "epoch": 0.6344120474206985,
      "grad_norm": 0.19563300907611847,
      "learning_rate": 7.0959489681780945e-06,
      "loss": 0.0658,
      "step": 4950
    },
    {
      "epoch": 0.6356936879205383,
      "grad_norm": 0.19320793449878693,
      "learning_rate": 7.053160892987434e-06,
      "loss": 0.0618,
      "step": 4960
    },
    {
      "epoch": 0.6369753284203781,
      "grad_norm": 0.18894124031066895,
      "learning_rate": 7.010431818542298e-06,
      "loss": 0.0629,
      "step": 4970
    },
    {
      "epoch": 0.6382569689202179,
      "grad_norm": 0.18019933998584747,
      "learning_rate": 6.967762600351646e-06,
      "loss": 0.065,
      "step": 4980
    },
    {
      "epoch": 0.6395386094200577,
      "grad_norm": 0.19297577440738678,
      "learning_rate": 6.9251540927260115e-06,
      "loss": 0.0624,
      "step": 4990
    },
    {
      "epoch": 0.6408202499198975,
      "grad_norm": 0.20897281169891357,
      "learning_rate": 6.8826071487603926e-06,
      "loss": 0.0642,
      "step": 5000
    },
    {
      "epoch": 0.6408202499198975,
      "eval_f1": 0.8921365599668659,
      "eval_loss": 0.06497912108898163,
      "eval_precision": 0.900179104477612,
      "eval_recall": 0.8842364532019704,
      "eval_runtime": 2.9972,
      "eval_samples_per_second": 2160.691,
      "eval_steps_per_second": 8.675,
      "step": 5000
    },
    {
      "epoch": 0.6421018904197373,
      "grad_norm": 0.16397173702716827,
      "learning_rate": 6.840122620317185e-06,
      "loss": 0.0637,
      "step": 5010
    },
    {
      "epoch": 0.643383530919577,
      "grad_norm": 0.1813378483057022,
      "learning_rate": 6.797701358009114e-06,
      "loss": 0.0653,
      "step": 5020
    },
    {
      "epoch": 0.6446651714194168,
      "grad_norm": 0.17247265577316284,
      "learning_rate": 6.755344211182221e-06,
      "loss": 0.0653,
      "step": 5030
    },
    {
      "epoch": 0.6459468119192566,
      "grad_norm": 0.20150932669639587,
      "learning_rate": 6.7130520278988255e-06,
      "loss": 0.064,
      "step": 5040
    },
    {
      "epoch": 0.6472284524190964,
      "grad_norm": 0.17291051149368286,
      "learning_rate": 6.670825654920579e-06,
      "loss": 0.0623,
      "step": 5050
    },
    {
      "epoch": 0.6485100929189362,
      "grad_norm": 0.1938193440437317,
      "learning_rate": 6.628665937691503e-06,
      "loss": 0.0632,
      "step": 5060
    },
    {
      "epoch": 0.649791733418776,
      "grad_norm": 0.2040683478116989,
      "learning_rate": 6.586573720321043e-06,
      "loss": 0.0624,
      "step": 5070
    },
    {
      "epoch": 0.6510733739186159,
      "grad_norm": 0.20392175018787384,
      "learning_rate": 6.544549845567195e-06,
      "loss": 0.0643,
      "step": 5080
    },
    {
      "epoch": 0.6523550144184557,
      "grad_norm": 0.16794058680534363,
      "learning_rate": 6.502595154819617e-06,
      "loss": 0.0637,
      "step": 5090
    },
    {
      "epoch": 0.6536366549182954,
      "grad_norm": 0.17543956637382507,
      "learning_rate": 6.460710488082774e-06,
      "loss": 0.0647,
      "step": 5100
    },
    {
      "epoch": 0.6549182954181352,
      "grad_norm": 0.1936439424753189,
      "learning_rate": 6.418896683959146e-06,
      "loss": 0.0641,
      "step": 5110
    },
    {
      "epoch": 0.656199935917975,
      "grad_norm": 0.2110176831483841,
      "learning_rate": 6.377154579632416e-06,
      "loss": 0.0645,
      "step": 5120
    },
    {
      "epoch": 0.6574815764178148,
      "grad_norm": 0.17867045104503632,
      "learning_rate": 6.3354850108507084e-06,
      "loss": 0.0636,
      "step": 5130
    },
    {
      "epoch": 0.6587632169176546,
      "grad_norm": 0.18331512808799744,
      "learning_rate": 6.293888811909874e-06,
      "loss": 0.0627,
      "step": 5140
    },
    {
      "epoch": 0.6600448574174944,
      "grad_norm": 0.17740613222122192,
      "learning_rate": 6.252366815636768e-06,
      "loss": 0.0607,
      "step": 5150
    },
    {
      "epoch": 0.6613264979173342,
      "grad_norm": 0.18722304701805115,
      "learning_rate": 6.2109198533725836e-06,
      "loss": 0.0649,
      "step": 5160
    },
    {
      "epoch": 0.662608138417174,
      "grad_norm": 0.1758435070514679,
      "learning_rate": 6.169548754956201e-06,
      "loss": 0.0619,
      "step": 5170
    },
    {
      "epoch": 0.6638897789170137,
      "grad_norm": 0.2046230286359787,
      "learning_rate": 6.128254348707579e-06,
      "loss": 0.0659,
      "step": 5180
    },
    {
      "epoch": 0.6651714194168535,
      "grad_norm": 0.19040079414844513,
      "learning_rate": 6.087037461411176e-06,
      "loss": 0.0613,
      "step": 5190
    },
    {
      "epoch": 0.6664530599166933,
      "grad_norm": 0.19141903519630432,
      "learning_rate": 6.045898918299373e-06,
      "loss": 0.062,
      "step": 5200
    },
    {
      "epoch": 0.6677347004165332,
      "grad_norm": 0.19412516057491302,
      "learning_rate": 6.004839543035981e-06,
      "loss": 0.0632,
      "step": 5210
    },
    {
      "epoch": 0.669016340916373,
      "grad_norm": 0.19847750663757324,
      "learning_rate": 5.9638601576997305e-06,
      "loss": 0.0635,
      "step": 5220
    },
    {
      "epoch": 0.6702979814162128,
      "grad_norm": 0.17892880737781525,
      "learning_rate": 5.922961582767812e-06,
      "loss": 0.0654,
      "step": 5230
    },
    {
      "epoch": 0.6715796219160526,
      "grad_norm": 0.2220013439655304,
      "learning_rate": 5.882144637099465e-06,
      "loss": 0.0654,
      "step": 5240
    },
    {
      "epoch": 0.6728612624158924,
      "grad_norm": 0.20190030336380005,
      "learning_rate": 5.841410137919563e-06,
      "loss": 0.0623,
      "step": 5250
    },
    {
      "epoch": 0.6741429029157321,
      "grad_norm": 0.18294011056423187,
      "learning_rate": 5.8007589008022605e-06,
      "loss": 0.0634,
      "step": 5260
    },
    {
      "epoch": 0.6754245434155719,
      "grad_norm": 0.18562212586402893,
      "learning_rate": 5.76019173965467e-06,
      "loss": 0.0639,
      "step": 5270
    },
    {
      "epoch": 0.6767061839154117,
      "grad_norm": 0.17406517267227173,
      "learning_rate": 5.719709466700558e-06,
      "loss": 0.0632,
      "step": 5280
    },
    {
      "epoch": 0.6779878244152515,
      "grad_norm": 0.19202564656734467,
      "learning_rate": 5.679312892464074e-06,
      "loss": 0.063,
      "step": 5290
    },
    {
      "epoch": 0.6792694649150913,
      "grad_norm": 0.18389640748500824,
      "learning_rate": 5.639002825753546e-06,
      "loss": 0.0643,
      "step": 5300
    },
    {
      "epoch": 0.6805511054149311,
      "grad_norm": 0.18304041028022766,
      "learning_rate": 5.598780073645267e-06,
      "loss": 0.0633,
      "step": 5310
    },
    {
      "epoch": 0.6818327459147709,
      "grad_norm": 0.1976161152124405,
      "learning_rate": 5.558645441467347e-06,
      "loss": 0.0632,
      "step": 5320
    },
    {
      "epoch": 0.6831143864146108,
      "grad_norm": 0.1801530420780182,
      "learning_rate": 5.51859973278358e-06,
      "loss": 0.0637,
      "step": 5330
    },
    {
      "epoch": 0.6843960269144505,
      "grad_norm": 0.20090307295322418,
      "learning_rate": 5.478643749377364e-06,
      "loss": 0.0661,
      "step": 5340
    },
    {
      "epoch": 0.6856776674142903,
      "grad_norm": 0.1645522266626358,
      "learning_rate": 5.438778291235647e-06,
      "loss": 0.0618,
      "step": 5350
    },
    {
      "epoch": 0.6869593079141301,
      "grad_norm": 0.19986960291862488,
      "learning_rate": 5.39900415653289e-06,
      "loss": 0.065,
      "step": 5360
    },
    {
      "epoch": 0.6882409484139699,
      "grad_norm": 0.18243259191513062,
      "learning_rate": 5.359322141615124e-06,
      "loss": 0.0646,
      "step": 5370
    },
    {
      "epoch": 0.6895225889138097,
      "grad_norm": 0.17021586000919342,
      "learning_rate": 5.319733040983972e-06,
      "loss": 0.06,
      "step": 5380
    },
    {
      "epoch": 0.6908042294136495,
      "grad_norm": 0.18343709409236908,
      "learning_rate": 5.280237647280759e-06,
      "loss": 0.0637,
      "step": 5390
    },
    {
      "epoch": 0.6920858699134893,
      "grad_norm": 0.19553858041763306,
      "learning_rate": 5.24083675127064e-06,
      "loss": 0.0623,
      "step": 5400
    },
    {
      "epoch": 0.6933675104133291,
      "grad_norm": 0.18324916064739227,
      "learning_rate": 5.201531141826759e-06,
      "loss": 0.0651,
      "step": 5410
    },
    {
      "epoch": 0.6946491509131688,
      "grad_norm": 0.16518688201904297,
      "learning_rate": 5.162321605914461e-06,
      "loss": 0.0662,
      "step": 5420
    },
    {
      "epoch": 0.6959307914130086,
      "grad_norm": 0.19472633302211761,
      "learning_rate": 5.123208928575538e-06,
      "loss": 0.0647,
      "step": 5430
    },
    {
      "epoch": 0.6972124319128484,
      "grad_norm": 0.17345207929611206,
      "learning_rate": 5.084193892912506e-06,
      "loss": 0.0634,
      "step": 5440
    },
    {
      "epoch": 0.6984940724126882,
      "grad_norm": 0.19299833476543427,
      "learning_rate": 5.0452772800729375e-06,
      "loss": 0.0632,
      "step": 5450
    },
    {
      "epoch": 0.699775712912528,
      "grad_norm": 0.17639560997486115,
      "learning_rate": 5.006459869233795e-06,
      "loss": 0.0625,
      "step": 5460
    },
    {
      "epoch": 0.7010573534123679,
      "grad_norm": 0.1841004192829132,
      "learning_rate": 4.967742437585859e-06,
      "loss": 0.0654,
      "step": 5470
    },
    {
      "epoch": 0.7023389939122077,
      "grad_norm": 0.17446312308311462,
      "learning_rate": 4.929125760318159e-06,
      "loss": 0.0632,
      "step": 5480
    },
    {
      "epoch": 0.7036206344120475,
      "grad_norm": 0.190720796585083,
      "learning_rate": 4.890610610602437e-06,
      "loss": 0.0659,
      "step": 5490
    },
    {
      "epoch": 0.7049022749118872,
      "grad_norm": 0.19830749928951263,
      "learning_rate": 4.852197759577688e-06,
      "loss": 0.0621,
      "step": 5500
    },
    {
      "epoch": 0.7049022749118872,
      "eval_f1": 0.8970788647271435,
      "eval_loss": 0.06400919705629349,
      "eval_precision": 0.9064782660759191,
      "eval_recall": 0.8878723903354445,
      "eval_runtime": 2.989,
      "eval_samples_per_second": 2166.596,
      "eval_steps_per_second": 8.699,
      "step": 5500
    },
    {
      "epoch": 0.706183915411727,
      "grad_norm": 0.20006248354911804,
      "learning_rate": 4.813887976334722e-06,
      "loss": 0.0631,
      "step": 5510
    },
    {
      "epoch": 0.7074655559115668,
      "grad_norm": 0.19379836320877075,
      "learning_rate": 4.775682027900739e-06,
      "loss": 0.0642,
      "step": 5520
    },
    {
      "epoch": 0.7087471964114066,
      "grad_norm": 0.18635933101177216,
      "learning_rate": 4.737580679223996e-06,
      "loss": 0.0632,
      "step": 5530
    },
    {
      "epoch": 0.7100288369112464,
      "grad_norm": 0.18606948852539062,
      "learning_rate": 4.699584693158494e-06,
      "loss": 0.0643,
      "step": 5540
    },
    {
      "epoch": 0.7113104774110862,
      "grad_norm": 0.18975263833999634,
      "learning_rate": 4.661694830448673e-06,
      "loss": 0.0635,
      "step": 5550
    },
    {
      "epoch": 0.712592117910926,
      "grad_norm": 0.18707899749279022,
      "learning_rate": 4.623911849714226e-06,
      "loss": 0.0622,
      "step": 5560
    },
    {
      "epoch": 0.7138737584107658,
      "grad_norm": 0.1804761439561844,
      "learning_rate": 4.586236507434876e-06,
      "loss": 0.0623,
      "step": 5570
    },
    {
      "epoch": 0.7151553989106055,
      "grad_norm": 0.19166141748428345,
      "learning_rate": 4.548669557935233e-06,
      "loss": 0.0616,
      "step": 5580
    },
    {
      "epoch": 0.7164370394104453,
      "grad_norm": 0.19050787389278412,
      "learning_rate": 4.511211753369712e-06,
      "loss": 0.0656,
      "step": 5590
    },
    {
      "epoch": 0.7177186799102852,
      "grad_norm": 0.18800762295722961,
      "learning_rate": 4.473863843707454e-06,
      "loss": 0.0629,
      "step": 5600
    },
    {
      "epoch": 0.719000320410125,
      "grad_norm": 0.17600564658641815,
      "learning_rate": 4.436626576717313e-06,
      "loss": 0.0643,
      "step": 5610
    },
    {
      "epoch": 0.7202819609099648,
      "grad_norm": 0.18252992630004883,
      "learning_rate": 4.399500697952898e-06,
      "loss": 0.0649,
      "step": 5620
    },
    {
      "epoch": 0.7215636014098046,
      "grad_norm": 0.21690472960472107,
      "learning_rate": 4.362486950737626e-06,
      "loss": 0.0652,
      "step": 5630
    },
    {
      "epoch": 0.7228452419096444,
      "grad_norm": 0.19096529483795166,
      "learning_rate": 4.325586076149858e-06,
      "loss": 0.0637,
      "step": 5640
    },
    {
      "epoch": 0.7241268824094841,
      "grad_norm": 0.17736491560935974,
      "learning_rate": 4.288798813008039e-06,
      "loss": 0.0634,
      "step": 5650
    },
    {
      "epoch": 0.7254085229093239,
      "grad_norm": 0.19415591657161713,
      "learning_rate": 4.2521258978559324e-06,
      "loss": 0.0629,
      "step": 5660
    },
    {
      "epoch": 0.7266901634091637,
      "grad_norm": 0.2018127739429474,
      "learning_rate": 4.215568064947854e-06,
      "loss": 0.0625,
      "step": 5670
    },
    {
      "epoch": 0.7279718039090035,
      "grad_norm": 0.17937569320201874,
      "learning_rate": 4.179126046233977e-06,
      "loss": 0.0633,
      "step": 5680
    },
    {
      "epoch": 0.7292534444088433,
      "grad_norm": 0.21321403980255127,
      "learning_rate": 4.142800571345678e-06,
      "loss": 0.064,
      "step": 5690
    },
    {
      "epoch": 0.7305350849086831,
      "grad_norm": 0.16154085099697113,
      "learning_rate": 4.106592367580931e-06,
      "loss": 0.0628,
      "step": 5700
    },
    {
      "epoch": 0.7318167254085229,
      "grad_norm": 0.20596788823604584,
      "learning_rate": 4.070502159889731e-06,
      "loss": 0.0655,
      "step": 5710
    },
    {
      "epoch": 0.7330983659083627,
      "grad_norm": 0.18368017673492432,
      "learning_rate": 4.034530670859598e-06,
      "loss": 0.0639,
      "step": 5720
    },
    {
      "epoch": 0.7343800064082024,
      "grad_norm": 0.184966579079628,
      "learning_rate": 3.998678620701102e-06,
      "loss": 0.0619,
      "step": 5730
    },
    {
      "epoch": 0.7356616469080423,
      "grad_norm": 0.19185282289981842,
      "learning_rate": 3.9629467272334495e-06,
      "loss": 0.0658,
      "step": 5740
    },
    {
      "epoch": 0.7369432874078821,
      "grad_norm": 0.19974376261234283,
      "learning_rate": 3.927335705870089e-06,
      "loss": 0.0642,
      "step": 5750
    },
    {
      "epoch": 0.7382249279077219,
      "grad_norm": 0.17768967151641846,
      "learning_rate": 3.8918462696044145e-06,
      "loss": 0.0638,
      "step": 5760
    },
    {
      "epoch": 0.7395065684075617,
      "grad_norm": 0.19170130789279938,
      "learning_rate": 3.8564791289954805e-06,
      "loss": 0.0607,
      "step": 5770
    },
    {
      "epoch": 0.7407882089074015,
      "grad_norm": 0.21157631278038025,
      "learning_rate": 3.8212349921537626e-06,
      "loss": 0.0647,
      "step": 5780
    },
    {
      "epoch": 0.7420698494072413,
      "grad_norm": 0.16236081719398499,
      "learning_rate": 3.7861145647269994e-06,
      "loss": 0.0619,
      "step": 5790
    },
    {
      "epoch": 0.7433514899070811,
      "grad_norm": 0.18598569929599762,
      "learning_rate": 3.751118549886065e-06,
      "loss": 0.0637,
      "step": 5800
    },
    {
      "epoch": 0.7446331304069208,
      "grad_norm": 0.19498658180236816,
      "learning_rate": 3.7162476483108612e-06,
      "loss": 0.0618,
      "step": 5810
    },
    {
      "epoch": 0.7459147709067606,
      "grad_norm": 0.19161143898963928,
      "learning_rate": 3.681502558176321e-06,
      "loss": 0.0624,
      "step": 5820
    },
    {
      "epoch": 0.7471964114066004,
      "grad_norm": 0.21474111080169678,
      "learning_rate": 3.646883975138421e-06,
      "loss": 0.0652,
      "step": 5830
    },
    {
      "epoch": 0.7484780519064402,
      "grad_norm": 0.18814519047737122,
      "learning_rate": 3.612392592320233e-06,
      "loss": 0.0629,
      "step": 5840
    },
    {
      "epoch": 0.74975969240628,
      "grad_norm": 0.18717658519744873,
      "learning_rate": 3.578029100298087e-06,
      "loss": 0.063,
      "step": 5850
    },
    {
      "epoch": 0.7510413329061199,
      "grad_norm": 0.1913532316684723,
      "learning_rate": 3.5437941870877078e-06,
      "loss": 0.0623,
      "step": 5860
    },
    {
      "epoch": 0.7523229734059597,
      "grad_norm": 0.19016607105731964,
      "learning_rate": 3.509688538130448e-06,
      "loss": 0.0643,
      "step": 5870
    },
    {
      "epoch": 0.7536046139057995,
      "grad_norm": 0.19770283997058868,
      "learning_rate": 3.475712836279579e-06,
      "loss": 0.062,
      "step": 5880
    },
    {
      "epoch": 0.7548862544056392,
      "grad_norm": 0.19611884653568268,
      "learning_rate": 3.441867761786607e-06,
      "loss": 0.0617,
      "step": 5890
    },
    {
      "epoch": 0.756167894905479,
      "grad_norm": 0.176730215549469,
      "learning_rate": 3.408153992287655e-06,
      "loss": 0.0619,
      "step": 5900
    },
    {
      "epoch": 0.7574495354053188,
      "grad_norm": 0.20056287944316864,
      "learning_rate": 3.374572202789895e-06,
      "loss": 0.0632,
      "step": 5910
    },
    {
      "epoch": 0.7587311759051586,
      "grad_norm": 0.20373013615608215,
      "learning_rate": 3.3411230656580372e-06,
      "loss": 0.0639,
      "step": 5920
    },
    {
      "epoch": 0.7600128164049984,
      "grad_norm": 0.17553798854351044,
      "learning_rate": 3.307807250600864e-06,
      "loss": 0.0614,
      "step": 5930
    },
    {
      "epoch": 0.7612944569048382,
      "grad_norm": 0.18124283850193024,
      "learning_rate": 3.2746254246578167e-06,
      "loss": 0.0654,
      "step": 5940
    },
    {
      "epoch": 0.762576097404678,
      "grad_norm": 0.17987513542175293,
      "learning_rate": 3.2415782521856543e-06,
      "loss": 0.064,
      "step": 5950
    },
    {
      "epoch": 0.7638577379045178,
      "grad_norm": 0.1896025538444519,
      "learning_rate": 3.208666394845139e-06,
      "loss": 0.0647,
      "step": 5960
    },
    {
      "epoch": 0.7651393784043575,
      "grad_norm": 0.16292789578437805,
      "learning_rate": 3.1758905115877968e-06,
      "loss": 0.062,
      "step": 5970
    },
    {
      "epoch": 0.7664210189041973,
      "grad_norm": 0.1821797639131546,
      "learning_rate": 3.1432512586427155e-06,
      "loss": 0.0627,
      "step": 5980
    },
    {
      "epoch": 0.7677026594040371,
      "grad_norm": 0.18443669378757477,
      "learning_rate": 3.1107492895034197e-06,
      "loss": 0.0628,
      "step": 5990
    },
    {
      "epoch": 0.768984299903877,
      "grad_norm": 0.1898009479045868,
      "learning_rate": 3.078385254914764e-06,
      "loss": 0.0639,
      "step": 6000
    },
    {
      "epoch": 0.768984299903877,
      "eval_f1": 0.8969218068168416,
      "eval_loss": 0.0637771412730217,
      "eval_precision": 0.9019212523719166,
      "eval_recall": 0.8919774806474314,
      "eval_runtime": 3.0047,
      "eval_samples_per_second": 2155.325,
      "eval_steps_per_second": 8.653,
      "step": 6000
    },
    {
      "epoch": 0.7702659404037168,
      "grad_norm": 0.17967712879180908,
      "learning_rate": 3.0461598028599305e-06,
      "loss": 0.0617,
      "step": 6010
    },
    {
      "epoch": 0.7715475809035566,
      "grad_norm": 0.196019247174263,
      "learning_rate": 3.014073578547437e-06,
      "loss": 0.0645,
      "step": 6020
    },
    {
      "epoch": 0.7728292214033964,
      "grad_norm": 0.1881665289402008,
      "learning_rate": 2.982127224398228e-06,
      "loss": 0.0633,
      "step": 6030
    },
    {
      "epoch": 0.7741108619032362,
      "grad_norm": 0.19193509221076965,
      "learning_rate": 2.9503213800328035e-06,
      "loss": 0.0638,
      "step": 6040
    },
    {
      "epoch": 0.7753925024030759,
      "grad_norm": 0.21863171458244324,
      "learning_rate": 2.918656682258423e-06,
      "loss": 0.0623,
      "step": 6050
    },
    {
      "epoch": 0.7766741429029157,
      "grad_norm": 0.18367791175842285,
      "learning_rate": 2.8871337650563504e-06,
      "loss": 0.0646,
      "step": 6060
    },
    {
      "epoch": 0.7779557834027555,
      "grad_norm": 0.1860390156507492,
      "learning_rate": 2.855753259569153e-06,
      "loss": 0.0651,
      "step": 6070
    },
    {
      "epoch": 0.7792374239025953,
      "grad_norm": 0.18775425851345062,
      "learning_rate": 2.8245157940880784e-06,
      "loss": 0.0619,
      "step": 6080
    },
    {
      "epoch": 0.7805190644024351,
      "grad_norm": 0.19896088540554047,
      "learning_rate": 2.793421994040478e-06,
      "loss": 0.0641,
      "step": 6090
    },
    {
      "epoch": 0.7818007049022749,
      "grad_norm": 0.16440510749816895,
      "learning_rate": 2.7624724819772586e-06,
      "loss": 0.0634,
      "step": 6100
    },
    {
      "epoch": 0.7830823454021147,
      "grad_norm": 0.17060019075870514,
      "learning_rate": 2.7316678775604464e-06,
      "loss": 0.065,
      "step": 6110
    },
    {
      "epoch": 0.7843639859019546,
      "grad_norm": 0.18785084784030914,
      "learning_rate": 2.7010087975507714e-06,
      "loss": 0.0613,
      "step": 6120
    },
    {
      "epoch": 0.7856456264017943,
      "grad_norm": 0.18072864413261414,
      "learning_rate": 2.6704958557953063e-06,
      "loss": 0.0609,
      "step": 6130
    },
    {
      "epoch": 0.7869272669016341,
      "grad_norm": 0.18689529597759247,
      "learning_rate": 2.6401296632151918e-06,
      "loss": 0.0629,
      "step": 6140
    },
    {
      "epoch": 0.7882089074014739,
      "grad_norm": 0.18816889822483063,
      "learning_rate": 2.6099108277934105e-06,
      "loss": 0.06,
      "step": 6150
    },
    {
      "epoch": 0.7894905479013137,
      "grad_norm": 0.1944245547056198,
      "learning_rate": 2.579839954562585e-06,
      "loss": 0.0642,
      "step": 6160
    },
    {
      "epoch": 0.7907721884011535,
      "grad_norm": 0.1741659939289093,
      "learning_rate": 2.5499176455928933e-06,
      "loss": 0.0632,
      "step": 6170
    },
    {
      "epoch": 0.7920538289009933,
      "grad_norm": 0.18417255580425262,
      "learning_rate": 2.520144499980002e-06,
      "loss": 0.0641,
      "step": 6180
    },
    {
      "epoch": 0.7933354694008331,
      "grad_norm": 0.20959390699863434,
      "learning_rate": 2.490521113833071e-06,
      "loss": 0.0653,
      "step": 6190
    },
    {
      "epoch": 0.7946171099006729,
      "grad_norm": 0.17131665349006653,
      "learning_rate": 2.4610480802628235e-06,
      "loss": 0.0615,
      "step": 6200
    },
    {
      "epoch": 0.7958987504005126,
      "grad_norm": 0.19155119359493256,
      "learning_rate": 2.4317259893696643e-06,
      "loss": 0.0627,
      "step": 6210
    },
    {
      "epoch": 0.7971803909003524,
      "grad_norm": 0.20838290452957153,
      "learning_rate": 2.402555428231872e-06,
      "loss": 0.0653,
      "step": 6220
    },
    {
      "epoch": 0.7984620314001922,
      "grad_norm": 0.18234503269195557,
      "learning_rate": 2.3735369808938338e-06,
      "loss": 0.0649,
      "step": 6230
    },
    {
      "epoch": 0.799743671900032,
      "grad_norm": 0.16235202550888062,
      "learning_rate": 2.344671228354366e-06,
      "loss": 0.0624,
      "step": 6240
    },
    {
      "epoch": 0.8010253123998718,
      "grad_norm": 0.17800317704677582,
      "learning_rate": 2.3159587485550728e-06,
      "loss": 0.0621,
      "step": 6250
    },
    {
      "epoch": 0.8023069528997117,
      "grad_norm": 0.18719688057899475,
      "learning_rate": 2.2874001163687764e-06,
      "loss": 0.063,
      "step": 6260
    },
    {
      "epoch": 0.8035885933995515,
      "grad_norm": 0.2191469967365265,
      "learning_rate": 2.258995903588008e-06,
      "loss": 0.0637,
      "step": 6270
    },
    {
      "epoch": 0.8048702338993913,
      "grad_norm": 0.1862994134426117,
      "learning_rate": 2.23074667891356e-06,
      "loss": 0.0651,
      "step": 6280
    },
    {
      "epoch": 0.806151874399231,
      "grad_norm": 0.17251913249492645,
      "learning_rate": 2.202653007943093e-06,
      "loss": 0.064,
      "step": 6290
    },
    {
      "epoch": 0.8074335148990708,
      "grad_norm": 0.19216704368591309,
      "learning_rate": 2.1747154531598226e-06,
      "loss": 0.0612,
      "step": 6300
    },
    {
      "epoch": 0.8087151553989106,
      "grad_norm": 0.2258075773715973,
      "learning_rate": 2.146934573921249e-06,
      "loss": 0.0634,
      "step": 6310
    },
    {
      "epoch": 0.8099967958987504,
      "grad_norm": 0.19085222482681274,
      "learning_rate": 2.119310926447965e-06,
      "loss": 0.0626,
      "step": 6320
    },
    {
      "epoch": 0.8112784363985902,
      "grad_norm": 0.16878673434257507,
      "learning_rate": 2.0918450638125097e-06,
      "loss": 0.0619,
      "step": 6330
    },
    {
      "epoch": 0.81256007689843,
      "grad_norm": 0.1871701180934906,
      "learning_rate": 2.0645375359283047e-06,
      "loss": 0.0627,
      "step": 6340
    },
    {
      "epoch": 0.8138417173982698,
      "grad_norm": 0.18515709042549133,
      "learning_rate": 2.037388889538642e-06,
      "loss": 0.0646,
      "step": 6350
    },
    {
      "epoch": 0.8151233578981096,
      "grad_norm": 0.18814954161643982,
      "learning_rate": 2.0103996682057235e-06,
      "loss": 0.0614,
      "step": 6360
    },
    {
      "epoch": 0.8164049983979493,
      "grad_norm": 0.18813377618789673,
      "learning_rate": 1.983570412299799e-06,
      "loss": 0.0626,
      "step": 6370
    },
    {
      "epoch": 0.8176866388977891,
      "grad_norm": 0.176827535033226,
      "learning_rate": 1.956901658988345e-06,
      "loss": 0.0617,
      "step": 6380
    },
    {
      "epoch": 0.818968279397629,
      "grad_norm": 0.18096593022346497,
      "learning_rate": 1.930393942225283e-06,
      "loss": 0.0635,
      "step": 6390
    },
    {
      "epoch": 0.8202499198974688,
      "grad_norm": 0.14722853899002075,
      "learning_rate": 1.9040477927403223e-06,
      "loss": 0.0633,
      "step": 6400
    },
    {
      "epoch": 0.8215315603973086,
      "grad_norm": 0.20519372820854187,
      "learning_rate": 1.8778637380283193e-06,
      "loss": 0.0647,
      "step": 6410
    },
    {
      "epoch": 0.8228132008971484,
      "grad_norm": 0.1677417904138565,
      "learning_rate": 1.8518423023387066e-06,
      "loss": 0.0617,
      "step": 6420
    },
    {
      "epoch": 0.8240948413969882,
      "grad_norm": 0.17646180093288422,
      "learning_rate": 1.8259840066650136e-06,
      "loss": 0.0634,
      "step": 6430
    },
    {
      "epoch": 0.8253764818968279,
      "grad_norm": 0.1692490577697754,
      "learning_rate": 1.800289368734436e-06,
      "loss": 0.0658,
      "step": 6440
    },
    {
      "epoch": 0.8266581223966677,
      "grad_norm": 0.20320384204387665,
      "learning_rate": 1.7747589029974454e-06,
      "loss": 0.0655,
      "step": 6450
    },
    {
      "epoch": 0.8279397628965075,
      "grad_norm": 0.17576946318149567,
      "learning_rate": 1.7493931206175185e-06,
      "loss": 0.0631,
      "step": 6460
    },
    {
      "epoch": 0.8292214033963473,
      "grad_norm": 0.1861860305070877,
      "learning_rate": 1.7241925294608853e-06,
      "loss": 0.0604,
      "step": 6470
    },
    {
      "epoch": 0.8305030438961871,
      "grad_norm": 0.18089807033538818,
      "learning_rate": 1.6991576340863669e-06,
      "loss": 0.0609,
      "step": 6480
    },
    {
      "epoch": 0.8317846843960269,
      "grad_norm": 0.19535063207149506,
      "learning_rate": 1.674288935735273e-06,
      "loss": 0.0628,
      "step": 6490
    },
    {
      "epoch": 0.8330663248958667,
      "grad_norm": 0.19545066356658936,
      "learning_rate": 1.6495869323213654e-06,
      "loss": 0.0663,
      "step": 6500
    },
    {
      "epoch": 0.8330663248958667,
      "eval_f1": 0.9011988424968995,
      "eval_loss": 0.06293080747127533,
      "eval_precision": 0.9075770191507078,
      "eval_recall": 0.8949096880131363,
      "eval_runtime": 2.9883,
      "eval_samples_per_second": 2167.101,
      "eval_steps_per_second": 8.701,
      "step": 6500
    },
    {
      "epoch": 0.8343479653957065,
      "grad_norm": 0.18452438712120056,
      "learning_rate": 1.625052118420889e-06,
      "loss": 0.0621,
      "step": 6510
    },
    {
      "epoch": 0.8356296058955462,
      "grad_norm": 0.18466158211231232,
      "learning_rate": 1.600684985262665e-06,
      "loss": 0.0614,
      "step": 6520
    },
    {
      "epoch": 0.836911246395386,
      "grad_norm": 0.20921823382377625,
      "learning_rate": 1.5764860207182642e-06,
      "loss": 0.0635,
      "step": 6530
    },
    {
      "epoch": 0.8381928868952259,
      "grad_norm": 0.2036418616771698,
      "learning_rate": 1.5524557092922377e-06,
      "loss": 0.062,
      "step": 6540
    },
    {
      "epoch": 0.8394745273950657,
      "grad_norm": 0.19418755173683167,
      "learning_rate": 1.5285945321124073e-06,
      "loss": 0.0633,
      "step": 6550
    },
    {
      "epoch": 0.8407561678949055,
      "grad_norm": 0.21860581636428833,
      "learning_rate": 1.504902966920243e-06,
      "loss": 0.0654,
      "step": 6560
    },
    {
      "epoch": 0.8420378083947453,
      "grad_norm": 0.18533924221992493,
      "learning_rate": 1.4813814880612942e-06,
      "loss": 0.0629,
      "step": 6570
    },
    {
      "epoch": 0.8433194488945851,
      "grad_norm": 0.2107323408126831,
      "learning_rate": 1.4580305664756856e-06,
      "loss": 0.0642,
      "step": 6580
    },
    {
      "epoch": 0.8446010893944249,
      "grad_norm": 0.17865046858787537,
      "learning_rate": 1.4348506696887e-06,
      "loss": 0.0639,
      "step": 6590
    },
    {
      "epoch": 0.8458827298942646,
      "grad_norm": 0.18889763951301575,
      "learning_rate": 1.4118422618014093e-06,
      "loss": 0.0626,
      "step": 6600
    },
    {
      "epoch": 0.8471643703941044,
      "grad_norm": 0.1740342676639557,
      "learning_rate": 1.389005803481389e-06,
      "loss": 0.0624,
      "step": 6610
    },
    {
      "epoch": 0.8484460108939442,
      "grad_norm": 0.1994953602552414,
      "learning_rate": 1.366341751953487e-06,
      "loss": 0.0623,
      "step": 6620
    },
    {
      "epoch": 0.849727651393784,
      "grad_norm": 0.18933704495429993,
      "learning_rate": 1.3438505609906738e-06,
      "loss": 0.0611,
      "step": 6630
    },
    {
      "epoch": 0.8510092918936238,
      "grad_norm": 0.19953961670398712,
      "learning_rate": 1.321532680904959e-06,
      "loss": 0.0643,
      "step": 6640
    },
    {
      "epoch": 0.8522909323934637,
      "grad_norm": 0.1963312029838562,
      "learning_rate": 1.299388558538368e-06,
      "loss": 0.062,
      "step": 6650
    },
    {
      "epoch": 0.8535725728933035,
      "grad_norm": 0.18360593914985657,
      "learning_rate": 1.277418637254002e-06,
      "loss": 0.0627,
      "step": 6660
    },
    {
      "epoch": 0.8548542133931433,
      "grad_norm": 0.1813330501317978,
      "learning_rate": 1.255623356927168e-06,
      "loss": 0.0629,
      "step": 6670
    },
    {
      "epoch": 0.856135853892983,
      "grad_norm": 0.17133601009845734,
      "learning_rate": 1.2340031539365483e-06,
      "loss": 0.0631,
      "step": 6680
    },
    {
      "epoch": 0.8574174943928228,
      "grad_norm": 0.17428767681121826,
      "learning_rate": 1.2125584611554886e-06,
      "loss": 0.0634,
      "step": 6690
    },
    {
      "epoch": 0.8586991348926626,
      "grad_norm": 0.18506939709186554,
      "learning_rate": 1.1912897079433183e-06,
      "loss": 0.0636,
      "step": 6700
    },
    {
      "epoch": 0.8599807753925024,
      "grad_norm": 0.1925283968448639,
      "learning_rate": 1.1701973201367544e-06,
      "loss": 0.0625,
      "step": 6710
    },
    {
      "epoch": 0.8612624158923422,
      "grad_norm": 0.20190182328224182,
      "learning_rate": 1.1492817200413785e-06,
      "loss": 0.0651,
      "step": 6720
    },
    {
      "epoch": 0.862544056392182,
      "grad_norm": 0.18169893324375153,
      "learning_rate": 1.128543326423187e-06,
      "loss": 0.0619,
      "step": 6730
    },
    {
      "epoch": 0.8638256968920218,
      "grad_norm": 0.212158665060997,
      "learning_rate": 1.1079825545001887e-06,
      "loss": 0.0639,
      "step": 6740
    },
    {
      "epoch": 0.8651073373918616,
      "grad_norm": 0.18513430655002594,
      "learning_rate": 1.0875998159341128e-06,
      "loss": 0.0646,
      "step": 6750
    },
    {
      "epoch": 0.8663889778917013,
      "grad_norm": 0.18722467124462128,
      "learning_rate": 1.0673955188221518e-06,
      "loss": 0.0611,
      "step": 6760
    },
    {
      "epoch": 0.8676706183915411,
      "grad_norm": 0.17140091955661774,
      "learning_rate": 1.0473700676887988e-06,
      "loss": 0.0645,
      "step": 6770
    },
    {
      "epoch": 0.868952258891381,
      "grad_norm": 0.19441086053848267,
      "learning_rate": 1.0275238634777441e-06,
      "loss": 0.0644,
      "step": 6780
    },
    {
      "epoch": 0.8702338993912208,
      "grad_norm": 0.22514182329177856,
      "learning_rate": 1.0078573035438476e-06,
      "loss": 0.0627,
      "step": 6790
    },
    {
      "epoch": 0.8715155398910606,
      "grad_norm": 0.18342842161655426,
      "learning_rate": 9.88370781645188e-07,
      "loss": 0.0621,
      "step": 6800
    },
    {
      "epoch": 0.8727971803909004,
      "grad_norm": 0.17307966947555542,
      "learning_rate": 9.690646879351706e-07,
      "loss": 0.0625,
      "step": 6810
    },
    {
      "epoch": 0.8740788208907402,
      "grad_norm": 0.1793297678232193,
      "learning_rate": 9.499394089547242e-07,
      "loss": 0.0605,
      "step": 6820
    },
    {
      "epoch": 0.87536046139058,
      "grad_norm": 0.16885530948638916,
      "learning_rate": 9.309953276245565e-07,
      "loss": 0.0631,
      "step": 6830
    },
    {
      "epoch": 0.8766421018904197,
      "grad_norm": 0.1713690608739853,
      "learning_rate": 9.122328232374899e-07,
      "loss": 0.0642,
      "step": 6840
    },
    {
      "epoch": 0.8779237423902595,
      "grad_norm": 0.17423969507217407,
      "learning_rate": 8.936522714508678e-07,
      "loss": 0.0619,
      "step": 6850
    },
    {
      "epoch": 0.8792053828900993,
      "grad_norm": 0.18652468919754028,
      "learning_rate": 8.752540442790314e-07,
      "loss": 0.0625,
      "step": 6860
    },
    {
      "epoch": 0.8804870233899391,
      "grad_norm": 0.17617492377758026,
      "learning_rate": 8.570385100858692e-07,
      "loss": 0.0624,
      "step": 6870
    },
    {
      "epoch": 0.8817686638897789,
      "grad_norm": 0.18869616091251373,
      "learning_rate": 8.390060335774486e-07,
      "loss": 0.0638,
      "step": 6880
    },
    {
      "epoch": 0.8830503043896187,
      "grad_norm": 0.1805189996957779,
      "learning_rate": 8.211569757947069e-07,
      "loss": 0.0621,
      "step": 6890
    },
    {
      "epoch": 0.8843319448894585,
      "grad_norm": 0.17392995953559875,
      "learning_rate": 8.034916941062287e-07,
      "loss": 0.0657,
      "step": 6900
    },
    {
      "epoch": 0.8856135853892984,
      "grad_norm": 0.20601050555706024,
      "learning_rate": 7.860105422010867e-07,
      "loss": 0.064,
      "step": 6910
    },
    {
      "epoch": 0.886895225889138,
      "grad_norm": 0.1817730814218521,
      "learning_rate": 7.687138700817598e-07,
      "loss": 0.0618,
      "step": 6920
    },
    {
      "epoch": 0.8881768663889779,
      "grad_norm": 0.1970674693584442,
      "learning_rate": 7.516020240571286e-07,
      "loss": 0.0615,
      "step": 6930
    },
    {
      "epoch": 0.8894585068888177,
      "grad_norm": 0.17473480105400085,
      "learning_rate": 7.346753467355372e-07,
      "loss": 0.0632,
      "step": 6940
    },
    {
      "epoch": 0.8907401473886575,
      "grad_norm": 0.18870329856872559,
      "learning_rate": 7.179341770179404e-07,
      "loss": 0.0629,
      "step": 6950
    },
    {
      "epoch": 0.8920217878884973,
      "grad_norm": 0.16362690925598145,
      "learning_rate": 7.013788500911112e-07,
      "loss": 0.0633,
      "step": 6960
    },
    {
      "epoch": 0.8933034283883371,
      "grad_norm": 0.20100702345371246,
      "learning_rate": 6.850096974209353e-07,
      "loss": 0.0622,
      "step": 6970
    },
    {
      "epoch": 0.8945850688881769,
      "grad_norm": 0.170489102602005,
      "learning_rate": 6.688270467457703e-07,
      "loss": 0.0616,
      "step": 6980
    },
    {
      "epoch": 0.8958667093880167,
      "grad_norm": 0.20799265801906586,
      "learning_rate": 6.528312220698885e-07,
      "loss": 0.0646,
      "step": 6990
    },
    {
      "epoch": 0.8971483498878564,
      "grad_norm": 0.18187017738819122,
      "learning_rate": 6.370225436569799e-07,
      "loss": 0.0638,
      "step": 7000
    },
    {
      "epoch": 0.8971483498878564,
      "eval_f1": 0.9003476928516707,
      "eval_loss": 0.06308671087026596,
      "eval_precision": 0.9047731848868885,
      "eval_recall": 0.89596528266479,
      "eval_runtime": 2.9924,
      "eval_samples_per_second": 2164.151,
      "eval_steps_per_second": 8.689,
      "step": 7000
    },
    {
      "epoch": 0.8984299903876962,
      "grad_norm": 0.19084925949573517,
      "learning_rate": 6.214013280237552e-07,
      "loss": 0.0641,
      "step": 7010
    },
    {
      "epoch": 0.899711630887536,
      "grad_norm": 0.18255262076854706,
      "learning_rate": 6.059678879336006e-07,
      "loss": 0.0634,
      "step": 7020
    },
    {
      "epoch": 0.9009932713873758,
      "grad_norm": 0.2704687714576721,
      "learning_rate": 5.907225323903076e-07,
      "loss": 0.0604,
      "step": 7030
    },
    {
      "epoch": 0.9022749118872156,
      "grad_norm": 0.1986435055732727,
      "learning_rate": 5.756655666319011e-07,
      "loss": 0.0641,
      "step": 7040
    },
    {
      "epoch": 0.9035565523870555,
      "grad_norm": 0.19538725912570953,
      "learning_rate": 5.607972921245197e-07,
      "loss": 0.0617,
      "step": 7050
    },
    {
      "epoch": 0.9048381928868953,
      "grad_norm": 0.2058635950088501,
      "learning_rate": 5.461180065563787e-07,
      "loss": 0.0643,
      "step": 7060
    },
    {
      "epoch": 0.9061198333867351,
      "grad_norm": 0.1940140575170517,
      "learning_rate": 5.316280038318178e-07,
      "loss": 0.0659,
      "step": 7070
    },
    {
      "epoch": 0.9074014738865748,
      "grad_norm": 0.19371388852596283,
      "learning_rate": 5.173275740654049e-07,
      "loss": 0.0634,
      "step": 7080
    },
    {
      "epoch": 0.9086831143864146,
      "grad_norm": 0.19026151299476624,
      "learning_rate": 5.032170035761397e-07,
      "loss": 0.0638,
      "step": 7090
    },
    {
      "epoch": 0.9099647548862544,
      "grad_norm": 0.17401890456676483,
      "learning_rate": 4.892965748817102e-07,
      "loss": 0.0601,
      "step": 7100
    },
    {
      "epoch": 0.9112463953860942,
      "grad_norm": 0.19419656693935394,
      "learning_rate": 4.7556656669284553e-07,
      "loss": 0.0603,
      "step": 7110
    },
    {
      "epoch": 0.912528035885934,
      "grad_norm": 0.17826074361801147,
      "learning_rate": 4.6202725390772863e-07,
      "loss": 0.0618,
      "step": 7120
    },
    {
      "epoch": 0.9138096763857738,
      "grad_norm": 0.20400400459766388,
      "learning_rate": 4.486789076064968e-07,
      "loss": 0.0633,
      "step": 7130
    },
    {
      "epoch": 0.9150913168856136,
      "grad_norm": 0.1725650578737259,
      "learning_rate": 4.355217950458124e-07,
      "loss": 0.062,
      "step": 7140
    },
    {
      "epoch": 0.9163729573854534,
      "grad_norm": 0.19925615191459656,
      "learning_rate": 4.225561796535127e-07,
      "loss": 0.0623,
      "step": 7150
    },
    {
      "epoch": 0.9176545978852931,
      "grad_norm": 0.19225375354290009,
      "learning_rate": 4.0978232102333095e-07,
      "loss": 0.0647,
      "step": 7160
    },
    {
      "epoch": 0.9189362383851329,
      "grad_norm": 0.167951762676239,
      "learning_rate": 3.9720047490970916e-07,
      "loss": 0.0614,
      "step": 7170
    },
    {
      "epoch": 0.9202178788849728,
      "grad_norm": 0.17831383645534515,
      "learning_rate": 3.8481089322266683e-07,
      "loss": 0.0624,
      "step": 7180
    },
    {
      "epoch": 0.9214995193848126,
      "grad_norm": 0.18907809257507324,
      "learning_rate": 3.726138240227628e-07,
      "loss": 0.0659,
      "step": 7190
    },
    {
      "epoch": 0.9227811598846524,
      "grad_norm": 0.16613955795764923,
      "learning_rate": 3.606095115161279e-07,
      "loss": 0.0602,
      "step": 7200
    },
    {
      "epoch": 0.9240628003844922,
      "grad_norm": 0.19376474618911743,
      "learning_rate": 3.4879819604957564e-07,
      "loss": 0.0637,
      "step": 7210
    },
    {
      "epoch": 0.925344440884332,
      "grad_norm": 0.18962319195270538,
      "learning_rate": 3.3718011410578954e-07,
      "loss": 0.0639,
      "step": 7220
    },
    {
      "epoch": 0.9266260813841717,
      "grad_norm": 0.19449980556964874,
      "learning_rate": 3.2575549829858557e-07,
      "loss": 0.0633,
      "step": 7230
    },
    {
      "epoch": 0.9279077218840115,
      "grad_norm": 0.17876175045967102,
      "learning_rate": 3.1452457736826146e-07,
      "loss": 0.0621,
      "step": 7240
    },
    {
      "epoch": 0.9291893623838513,
      "grad_norm": 0.21437576413154602,
      "learning_rate": 3.0348757617701064e-07,
      "loss": 0.0654,
      "step": 7250
    },
    {
      "epoch": 0.9304710028836911,
      "grad_norm": 0.17090333998203278,
      "learning_rate": 2.926447157044243e-07,
      "loss": 0.0612,
      "step": 7260
    },
    {
      "epoch": 0.9317526433835309,
      "grad_norm": 0.17806924879550934,
      "learning_rate": 2.8199621304306425e-07,
      "loss": 0.0641,
      "step": 7270
    },
    {
      "epoch": 0.9330342838833707,
      "grad_norm": 0.1961851716041565,
      "learning_rate": 2.715422813941193e-07,
      "loss": 0.064,
      "step": 7280
    },
    {
      "epoch": 0.9343159243832105,
      "grad_norm": 0.1685817539691925,
      "learning_rate": 2.612831300631291e-07,
      "loss": 0.063,
      "step": 7290
    },
    {
      "epoch": 0.9355975648830503,
      "grad_norm": 0.18308007717132568,
      "learning_rate": 2.5121896445580383e-07,
      "loss": 0.0615,
      "step": 7300
    },
    {
      "epoch": 0.93687920538289,
      "grad_norm": 0.19477280974388123,
      "learning_rate": 2.413499860739088e-07,
      "loss": 0.0623,
      "step": 7310
    },
    {
      "epoch": 0.9381608458827299,
      "grad_norm": 0.21266412734985352,
      "learning_rate": 2.316763925112242e-07,
      "loss": 0.0637,
      "step": 7320
    },
    {
      "epoch": 0.9394424863825697,
      "grad_norm": 0.1711306869983673,
      "learning_rate": 2.2219837744959284e-07,
      "loss": 0.062,
      "step": 7330
    },
    {
      "epoch": 0.9407241268824095,
      "grad_norm": 0.1893559992313385,
      "learning_rate": 2.1291613065504313e-07,
      "loss": 0.0619,
      "step": 7340
    },
    {
      "epoch": 0.9420057673822493,
      "grad_norm": 0.1796388030052185,
      "learning_rate": 2.0382983797399224e-07,
      "loss": 0.0629,
      "step": 7350
    },
    {
      "epoch": 0.9432874078820891,
      "grad_norm": 0.19146645069122314,
      "learning_rate": 1.9493968132951456e-07,
      "loss": 0.0645,
      "step": 7360
    },
    {
      "epoch": 0.9445690483819289,
      "grad_norm": 0.16710014641284943,
      "learning_rate": 1.8624583871771352e-07,
      "loss": 0.063,
      "step": 7370
    },
    {
      "epoch": 0.9458506888817687,
      "grad_norm": 0.29610446095466614,
      "learning_rate": 1.777484842041488e-07,
      "loss": 0.0612,
      "step": 7380
    },
    {
      "epoch": 0.9471323293816084,
      "grad_norm": 0.16617882251739502,
      "learning_rate": 1.6944778792034822e-07,
      "loss": 0.0638,
      "step": 7390
    },
    {
      "epoch": 0.9484139698814482,
      "grad_norm": 0.20489302277565002,
      "learning_rate": 1.6134391606041354e-07,
      "loss": 0.0638,
      "step": 7400
    },
    {
      "epoch": 0.949695610381288,
      "grad_norm": 0.19994662702083588,
      "learning_rate": 1.5343703087768225e-07,
      "loss": 0.0642,
      "step": 7410
    },
    {
      "epoch": 0.9509772508811278,
      "grad_norm": 0.1873672902584076,
      "learning_rate": 1.4572729068148106e-07,
      "loss": 0.0625,
      "step": 7420
    },
    {
      "epoch": 0.9522588913809676,
      "grad_norm": 0.18355967104434967,
      "learning_rate": 1.3821484983396194e-07,
      "loss": 0.0623,
      "step": 7430
    },
    {
      "epoch": 0.9535405318808075,
      "grad_norm": 0.1817985624074936,
      "learning_rate": 1.308998587470056e-07,
      "loss": 0.0641,
      "step": 7440
    },
    {
      "epoch": 0.9548221723806473,
      "grad_norm": 0.19310711324214935,
      "learning_rate": 1.2378246387920845e-07,
      "loss": 0.0619,
      "step": 7450
    },
    {
      "epoch": 0.9561038128804871,
      "grad_norm": 0.18324962258338928,
      "learning_rate": 1.1686280773295699e-07,
      "loss": 0.0643,
      "step": 7460
    },
    {
      "epoch": 0.9573854533803268,
      "grad_norm": 0.20702552795410156,
      "learning_rate": 1.1014102885157252e-07,
      "loss": 0.0663,
      "step": 7470
    },
    {
      "epoch": 0.9586670938801666,
      "grad_norm": 0.1761350929737091,
      "learning_rate": 1.0361726181653209e-07,
      "loss": 0.0637,
      "step": 7480
    },
    {
      "epoch": 0.9599487343800064,
      "grad_norm": 0.17296813428401947,
      "learning_rate": 9.729163724478074e-08,
      "loss": 0.0619,
      "step": 7490
    },
    {
      "epoch": 0.9612303748798462,
      "grad_norm": 0.20203419029712677,
      "learning_rate": 9.116428178611248e-08,
      "loss": 0.0641,
      "step": 7500
    },
    {
      "epoch": 0.9612303748798462,
      "eval_f1": 0.9000766373872546,
      "eval_loss": 0.06302953511476517,
      "eval_precision": 0.9048239895697523,
      "eval_recall": 0.8953788411916491,
      "eval_runtime": 2.9885,
      "eval_samples_per_second": 2167.001,
      "eval_steps_per_second": 8.7,
      "step": 7500
    },
    {
      "epoch": 0.962512015379686,
      "grad_norm": 0.19427528977394104,
      "learning_rate": 8.523531812063446e-08,
      "loss": 0.0604,
      "step": 7510
    },
    {
      "epoch": 0.9637936558795258,
      "grad_norm": 0.19381150603294373,
      "learning_rate": 7.950486495631349e-08,
      "loss": 0.0646,
      "step": 7520
    },
    {
      "epoch": 0.9650752963793656,
      "grad_norm": 0.18169087171554565,
      "learning_rate": 7.397303702659675e-08,
      "loss": 0.063,
      "step": 7530
    },
    {
      "epoch": 0.9663569368792054,
      "grad_norm": 0.18334563076496124,
      "learning_rate": 6.863994508811478e-08,
      "loss": 0.0627,
      "step": 7540
    },
    {
      "epoch": 0.9676385773790451,
      "grad_norm": 0.19370083510875702,
      "learning_rate": 6.350569591846434e-08,
      "loss": 0.0633,
      "step": 7550
    },
    {
      "epoch": 0.9689202178788849,
      "grad_norm": 0.18291835486888885,
      "learning_rate": 5.8570392314074576e-08,
      "loss": 0.0639,
      "step": 7560
    },
    {
      "epoch": 0.9702018583787247,
      "grad_norm": 0.18133868277072906,
      "learning_rate": 5.3834133088139784e-08,
      "loss": 0.0644,
      "step": 7570
    },
    {
      "epoch": 0.9714834988785646,
      "grad_norm": 0.1771089881658554,
      "learning_rate": 4.929701306864876e-08,
      "loss": 0.0611,
      "step": 7580
    },
    {
      "epoch": 0.9727651393784044,
      "grad_norm": 0.17924901843070984,
      "learning_rate": 4.4959123096482984e-08,
      "loss": 0.0636,
      "step": 7590
    },
    {
      "epoch": 0.9740467798782442,
      "grad_norm": 0.1930427998304367,
      "learning_rate": 4.0820550023601416e-08,
      "loss": 0.064,
      "step": 7600
    },
    {
      "epoch": 0.975328420378084,
      "grad_norm": 0.18830351531505585,
      "learning_rate": 3.68813767112941e-08,
      "loss": 0.0631,
      "step": 7610
    },
    {
      "epoch": 0.9766100608779238,
      "grad_norm": 0.19777809083461761,
      "learning_rate": 3.31416820285313e-08,
      "loss": 0.0634,
      "step": 7620
    },
    {
      "epoch": 0.9778917013777635,
      "grad_norm": 0.17489665746688843,
      "learning_rate": 2.960154085037803e-08,
      "loss": 0.0635,
      "step": 7630
    },
    {
      "epoch": 0.9791733418776033,
      "grad_norm": 0.18989871442317963,
      "learning_rate": 2.6261024056501995e-08,
      "loss": 0.0635,
      "step": 7640
    },
    {
      "epoch": 0.9804549823774431,
      "grad_norm": 0.19326794147491455,
      "learning_rate": 2.3120198529745785e-08,
      "loss": 0.0629,
      "step": 7650
    },
    {
      "epoch": 0.9817366228772829,
      "grad_norm": 0.16711817681789398,
      "learning_rate": 2.0179127154793533e-08,
      "loss": 0.0626,
      "step": 7660
    },
    {
      "epoch": 0.9830182633771227,
      "grad_norm": 0.17362891137599945,
      "learning_rate": 1.7437868816911894e-08,
      "loss": 0.0631,
      "step": 7670
    },
    {
      "epoch": 0.9842999038769625,
      "grad_norm": 0.17052555084228516,
      "learning_rate": 1.4896478400767688e-08,
      "loss": 0.0623,
      "step": 7680
    },
    {
      "epoch": 0.9855815443768023,
      "grad_norm": 0.1941647231578827,
      "learning_rate": 1.2555006789334301e-08,
      "loss": 0.0631,
      "step": 7690
    },
    {
      "epoch": 0.9868631848766422,
      "grad_norm": 0.18964990973472595,
      "learning_rate": 1.0413500862864745e-08,
      "loss": 0.0635,
      "step": 7700
    },
    {
      "epoch": 0.9881448253764819,
      "grad_norm": 0.21469241380691528,
      "learning_rate": 8.47200349796129e-09,
      "loss": 0.0633,
      "step": 7710
    },
    {
      "epoch": 0.9894264658763217,
      "grad_norm": 0.174968883395195,
      "learning_rate": 6.730553566713926e-09,
      "loss": 0.0627,
      "step": 7720
    },
    {
      "epoch": 0.9907081063761615,
      "grad_norm": 0.17662419378757477,
      "learning_rate": 5.189185935919883e-09,
      "loss": 0.0615,
      "step": 7730
    },
    {
      "epoch": 0.9919897468760013,
      "grad_norm": 0.2064337432384491,
      "learning_rate": 3.847931466388621e-09,
      "loss": 0.0614,
      "step": 7740
    },
    {
      "epoch": 0.9932713873758411,
      "grad_norm": 0.2198801338672638,
      "learning_rate": 2.7068170123234483e-09,
      "loss": 0.0633,
      "step": 7750
    },
    {
      "epoch": 0.9945530278756809,
      "grad_norm": 0.20898166298866272,
      "learning_rate": 1.765865420779722e-09,
      "loss": 0.0641,
      "step": 7760
    },
    {
      "epoch": 0.9958346683755207,
      "grad_norm": 0.17097249627113342,
      "learning_rate": 1.0250955312152144e-09,
      "loss": 0.0628,
      "step": 7770
    },
    {
      "epoch": 0.9971163088753605,
      "grad_norm": 0.15358439087867737,
      "learning_rate": 4.845221751070827e-10,
      "loss": 0.0634,
      "step": 7780
    },
    {
      "epoch": 0.9983979493752002,
      "grad_norm": 0.18262039124965668,
      "learning_rate": 1.4415617565433082e-10,
      "loss": 0.0618,
      "step": 7790
    },
    {
      "epoch": 0.99967958987504,
      "grad_norm": 0.21999365091323853,
      "learning_rate": 4.004347565755851e-12,
      "loss": 0.0642,
      "step": 7800
    },
    {
      "epoch": 0.999935917975008,
      "step": 7802,
      "total_flos": 4.995590848700744e+17,
      "train_loss": 0.0661678893047093,
      "train_runtime": 4443.5679,
      "train_samples_per_second": 449.496,
      "train_steps_per_second": 1.756
    }
  ],
  "logging_steps": 10,
  "max_steps": 7802,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.995590848700744e+17,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}