{
  "best_global_step": 4000,
  "best_metric": 0.7899137546751205,
  "best_model_checkpoint": "./output_checkpoints/codebert-base-cross-entropy/checkpoint-4000",
  "epoch": 0.5119672340970178,
  "eval_steps": 2000,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0012799180852425445,
      "grad_norm": 3.6954128742218018,
      "learning_rate": 3.840409643695328e-08,
      "loss": 0.7763,
      "step": 10
    },
    {
      "epoch": 0.002559836170485089,
      "grad_norm": 3.3660645484924316,
      "learning_rate": 8.10753147002347e-08,
      "loss": 0.7763,
      "step": 20
    },
    {
      "epoch": 0.0038397542557276334,
      "grad_norm": 4.491067886352539,
      "learning_rate": 1.2374653296351612e-07,
      "loss": 0.7889,
      "step": 30
    },
    {
      "epoch": 0.005119672340970178,
      "grad_norm": 1.5920521020889282,
      "learning_rate": 1.6641775122679754e-07,
      "loss": 0.7888,
      "step": 40
    },
    {
      "epoch": 0.006399590426212722,
      "grad_norm": 4.410898208618164,
      "learning_rate": 2.0908896949007894e-07,
      "loss": 0.7692,
      "step": 50
    },
    {
      "epoch": 0.007679508511455267,
      "grad_norm": 2.5669264793395996,
      "learning_rate": 2.517601877533604e-07,
      "loss": 0.7735,
      "step": 60
    },
    {
      "epoch": 0.008959426596697812,
      "grad_norm": 6.190986156463623,
      "learning_rate": 2.944314060166418e-07,
      "loss": 0.7679,
      "step": 70
    },
    {
      "epoch": 0.010239344681940356,
      "grad_norm": 5.736413478851318,
      "learning_rate": 3.371026242799232e-07,
      "loss": 0.744,
      "step": 80
    },
    {
      "epoch": 0.011519262767182901,
      "grad_norm": 4.278866767883301,
      "learning_rate": 3.7977384254320464e-07,
      "loss": 0.7542,
      "step": 90
    },
    {
      "epoch": 0.012799180852425445,
      "grad_norm": 5.565455436706543,
      "learning_rate": 4.22445060806486e-07,
      "loss": 0.752,
      "step": 100
    },
    {
      "epoch": 0.01407909893766799,
      "grad_norm": 4.031781196594238,
      "learning_rate": 4.651162790697675e-07,
      "loss": 0.7614,
      "step": 110
    },
    {
      "epoch": 0.015359017022910534,
      "grad_norm": 2.484126329421997,
      "learning_rate": 5.077874973330489e-07,
      "loss": 0.7664,
      "step": 120
    },
    {
      "epoch": 0.016638935108153077,
      "grad_norm": 2.7525155544281006,
      "learning_rate": 5.504587155963304e-07,
      "loss": 0.7521,
      "step": 130
    },
    {
      "epoch": 0.017918853193395624,
      "grad_norm": 1.6374385356903076,
      "learning_rate": 5.931299338596117e-07,
      "loss": 0.7436,
      "step": 140
    },
    {
      "epoch": 0.019198771278638168,
      "grad_norm": 2.370774269104004,
      "learning_rate": 6.358011521228932e-07,
      "loss": 0.7429,
      "step": 150
    },
    {
      "epoch": 0.02047868936388071,
      "grad_norm": 2.8948566913604736,
      "learning_rate": 6.784723703861745e-07,
      "loss": 0.7333,
      "step": 160
    },
    {
      "epoch": 0.021758607449123255,
      "grad_norm": 2.7481186389923096,
      "learning_rate": 7.21143588649456e-07,
      "loss": 0.7294,
      "step": 170
    },
    {
      "epoch": 0.023038525534365802,
      "grad_norm": 2.0159404277801514,
      "learning_rate": 7.638148069127374e-07,
      "loss": 0.731,
      "step": 180
    },
    {
      "epoch": 0.024318443619608346,
      "grad_norm": 3.6898131370544434,
      "learning_rate": 8.064860251760189e-07,
      "loss": 0.7153,
      "step": 190
    },
    {
      "epoch": 0.02559836170485089,
      "grad_norm": 3.2842342853546143,
      "learning_rate": 8.491572434393003e-07,
      "loss": 0.7245,
      "step": 200
    },
    {
      "epoch": 0.02559836170485089,
      "eval_accuracy": 0.47695,
      "eval_loss": 0.7148504853248596,
      "eval_macro_f1": 0.3229290091066048,
      "eval_precision": 0.238475,
      "eval_recall": 0.5,
      "eval_runtime": 1641.324,
      "eval_samples_per_second": 60.926,
      "eval_steps_per_second": 0.476,
      "step": 200
    },
    {
      "epoch": 0.026878279790093433,
      "grad_norm": 2.294558048248291,
      "learning_rate": 8.918284617025817e-07,
      "loss": 0.718,
      "step": 210
    },
    {
      "epoch": 0.02815819787533598,
      "grad_norm": 2.9467175006866455,
      "learning_rate": 9.344996799658632e-07,
      "loss": 0.7076,
      "step": 220
    },
    {
      "epoch": 0.029438115960578524,
      "grad_norm": 2.6889193058013916,
      "learning_rate": 9.771708982291445e-07,
      "loss": 0.7019,
      "step": 230
    },
    {
      "epoch": 0.030718034045821067,
      "grad_norm": 0.7491384744644165,
      "learning_rate": 1.0198421164924258e-06,
      "loss": 0.7046,
      "step": 240
    },
    {
      "epoch": 0.031997952131063614,
      "grad_norm": 2.0363433361053467,
      "learning_rate": 1.0625133347557074e-06,
      "loss": 0.7085,
      "step": 250
    },
    {
      "epoch": 0.033277870216306155,
      "grad_norm": 1.1431337594985962,
      "learning_rate": 1.1051845530189888e-06,
      "loss": 0.7008,
      "step": 260
    },
    {
      "epoch": 0.0345577883015487,
      "grad_norm": 3.7735233306884766,
      "learning_rate": 1.1478557712822702e-06,
      "loss": 0.7015,
      "step": 270
    },
    {
      "epoch": 0.03583770638679125,
      "grad_norm": 0.7889328002929688,
      "learning_rate": 1.1905269895455517e-06,
      "loss": 0.699,
      "step": 280
    },
    {
      "epoch": 0.03711762447203379,
      "grad_norm": 0.8796226382255554,
      "learning_rate": 1.233198207808833e-06,
      "loss": 0.6922,
      "step": 290
    },
    {
      "epoch": 0.038397542557276336,
      "grad_norm": 0.7966652512550354,
      "learning_rate": 1.2758694260721145e-06,
      "loss": 0.6949,
      "step": 300
    },
    {
      "epoch": 0.039677460642518876,
      "grad_norm": 0.7029069066047668,
      "learning_rate": 1.318540644335396e-06,
      "loss": 0.695,
      "step": 310
    },
    {
      "epoch": 0.04095737872776142,
      "grad_norm": 2.04327392578125,
      "learning_rate": 1.3612118625986772e-06,
      "loss": 0.6944,
      "step": 320
    },
    {
      "epoch": 0.04223729681300397,
      "grad_norm": 0.918025553226471,
      "learning_rate": 1.4038830808619588e-06,
      "loss": 0.6863,
      "step": 330
    },
    {
      "epoch": 0.04351721489824651,
      "grad_norm": 0.9528706073760986,
      "learning_rate": 1.4465542991252401e-06,
      "loss": 0.6907,
      "step": 340
    },
    {
      "epoch": 0.04479713298348906,
      "grad_norm": 1.071359634399414,
      "learning_rate": 1.4892255173885215e-06,
      "loss": 0.6873,
      "step": 350
    },
    {
      "epoch": 0.046077051068731605,
      "grad_norm": 1.9875625371932983,
      "learning_rate": 1.531896735651803e-06,
      "loss": 0.6942,
      "step": 360
    },
    {
      "epoch": 0.047356969153974145,
      "grad_norm": 2.0222439765930176,
      "learning_rate": 1.5745679539150842e-06,
      "loss": 0.6898,
      "step": 370
    },
    {
      "epoch": 0.04863688723921669,
      "grad_norm": 1.7827857732772827,
      "learning_rate": 1.6172391721783658e-06,
      "loss": 0.6909,
      "step": 380
    },
    {
      "epoch": 0.04991680532445923,
      "grad_norm": 1.065674901008606,
      "learning_rate": 1.6599103904416472e-06,
      "loss": 0.6877,
      "step": 390
    },
    {
      "epoch": 0.05119672340970178,
      "grad_norm": 0.8346193432807922,
      "learning_rate": 1.7025816087049288e-06,
      "loss": 0.6889,
      "step": 400
    },
    {
      "epoch": 0.05119672340970178,
      "eval_accuracy": 0.54043,
      "eval_loss": 0.6827595233917236,
      "eval_macro_f1": 0.41833840347598905,
      "eval_precision": 0.5839205994464474,
      "eval_recall": 0.5204152767503025,
      "eval_runtime": 1634.6244,
      "eval_samples_per_second": 61.176,
      "eval_steps_per_second": 0.478,
      "step": 400
    },
    {
      "epoch": 0.052476641494944326,
      "grad_norm": 0.8074935674667358,
      "learning_rate": 1.7452528269682101e-06,
      "loss": 0.684,
      "step": 410
    },
    {
      "epoch": 0.053756559580186866,
      "grad_norm": 0.8965069651603699,
      "learning_rate": 1.7879240452314913e-06,
      "loss": 0.6922,
      "step": 420
    },
    {
      "epoch": 0.05503647766542941,
      "grad_norm": 2.597215414047241,
      "learning_rate": 1.8305952634947729e-06,
      "loss": 0.6909,
      "step": 430
    },
    {
      "epoch": 0.05631639575067196,
      "grad_norm": 0.8233373165130615,
      "learning_rate": 1.8732664817580542e-06,
      "loss": 0.6831,
      "step": 440
    },
    {
      "epoch": 0.0575963138359145,
      "grad_norm": 1.0624489784240723,
      "learning_rate": 1.915937700021336e-06,
      "loss": 0.6817,
      "step": 450
    },
    {
      "epoch": 0.05887623192115705,
      "grad_norm": 2.0623559951782227,
      "learning_rate": 1.958608918284617e-06,
      "loss": 0.6819,
      "step": 460
    },
    {
      "epoch": 0.06015615000639959,
      "grad_norm": 0.912433922290802,
      "learning_rate": 2.0012801365478988e-06,
      "loss": 0.686,
      "step": 470
    },
    {
      "epoch": 0.061436068091642135,
      "grad_norm": 0.6926271319389343,
      "learning_rate": 2.04395135481118e-06,
      "loss": 0.6834,
      "step": 480
    },
    {
      "epoch": 0.06271598617688467,
      "grad_norm": 1.498634934425354,
      "learning_rate": 2.0866225730744615e-06,
      "loss": 0.6805,
      "step": 490
    },
    {
      "epoch": 0.06399590426212723,
      "grad_norm": 1.1818156242370605,
      "learning_rate": 2.129293791337743e-06,
      "loss": 0.6805,
      "step": 500
    },
    {
      "epoch": 0.06527582234736977,
      "grad_norm": 1.4476051330566406,
      "learning_rate": 2.1719650096010242e-06,
      "loss": 0.6773,
      "step": 510
    },
    {
      "epoch": 0.06655574043261231,
      "grad_norm": 1.1143743991851807,
      "learning_rate": 2.214636227864306e-06,
      "loss": 0.6801,
      "step": 520
    },
    {
      "epoch": 0.06783565851785486,
      "grad_norm": 1.7030247449874878,
      "learning_rate": 2.257307446127587e-06,
      "loss": 0.6794,
      "step": 530
    },
    {
      "epoch": 0.0691155766030974,
      "grad_norm": 0.8592290282249451,
      "learning_rate": 2.2999786643908685e-06,
      "loss": 0.6736,
      "step": 540
    },
    {
      "epoch": 0.07039549468833994,
      "grad_norm": 0.8231940865516663,
      "learning_rate": 2.34264988265415e-06,
      "loss": 0.67,
      "step": 550
    },
    {
      "epoch": 0.0716754127735825,
      "grad_norm": 0.88897705078125,
      "learning_rate": 2.3853211009174317e-06,
      "loss": 0.685,
      "step": 560
    },
    {
      "epoch": 0.07295533085882504,
      "grad_norm": 1.6297541856765747,
      "learning_rate": 2.427992319180713e-06,
      "loss": 0.6797,
      "step": 570
    },
    {
      "epoch": 0.07423524894406758,
      "grad_norm": 2.9297447204589844,
      "learning_rate": 2.470663537443994e-06,
      "loss": 0.6766,
      "step": 580
    },
    {
      "epoch": 0.07551516702931012,
      "grad_norm": 0.9277703762054443,
      "learning_rate": 2.5133347557072756e-06,
      "loss": 0.6824,
      "step": 590
    },
    {
      "epoch": 0.07679508511455267,
      "grad_norm": 0.5883333683013916,
      "learning_rate": 2.556005973970557e-06,
      "loss": 0.6818,
      "step": 600
    },
    {
      "epoch": 0.07679508511455267,
      "eval_accuracy": 0.63909,
      "eval_loss": 0.6700075268745422,
      "eval_macro_f1": 0.621848143053056,
      "eval_precision": 0.6524313071088819,
      "eval_recall": 0.6305863955136095,
      "eval_runtime": 1633.9732,
      "eval_samples_per_second": 61.201,
      "eval_steps_per_second": 0.479,
      "step": 600
    },
    {
      "epoch": 0.07807500319979521,
      "grad_norm": 0.7149448394775391,
      "learning_rate": 2.5986771922338383e-06,
      "loss": 0.6746,
      "step": 610
    },
    {
      "epoch": 0.07935492128503775,
      "grad_norm": 2.0167856216430664,
      "learning_rate": 2.64134841049712e-06,
      "loss": 0.682,
      "step": 620
    },
    {
      "epoch": 0.0806348393702803,
      "grad_norm": 1.303001880645752,
      "learning_rate": 2.6840196287604015e-06,
      "loss": 0.6735,
      "step": 630
    },
    {
      "epoch": 0.08191475745552285,
      "grad_norm": 2.0516321659088135,
      "learning_rate": 2.7266908470236826e-06,
      "loss": 0.6705,
      "step": 640
    },
    {
      "epoch": 0.08319467554076539,
      "grad_norm": 2.242105484008789,
      "learning_rate": 2.7693620652869642e-06,
      "loss": 0.6798,
      "step": 650
    },
    {
      "epoch": 0.08447459362600794,
      "grad_norm": 1.1836538314819336,
      "learning_rate": 2.812033283550246e-06,
      "loss": 0.6684,
      "step": 660
    },
    {
      "epoch": 0.08575451171125048,
      "grad_norm": 0.8640874624252319,
      "learning_rate": 2.8547045018135274e-06,
      "loss": 0.6783,
      "step": 670
    },
    {
      "epoch": 0.08703442979649302,
      "grad_norm": 0.6901177763938904,
      "learning_rate": 2.897375720076808e-06,
      "loss": 0.6724,
      "step": 680
    },
    {
      "epoch": 0.08831434788173557,
      "grad_norm": 1.3443949222564697,
      "learning_rate": 2.9400469383400897e-06,
      "loss": 0.67,
      "step": 690
    },
    {
      "epoch": 0.08959426596697811,
      "grad_norm": 0.7775189280509949,
      "learning_rate": 2.9827181566033713e-06,
      "loss": 0.6733,
      "step": 700
    },
    {
      "epoch": 0.09087418405222066,
      "grad_norm": 1.16916823387146,
      "learning_rate": 3.025389374866653e-06,
      "loss": 0.6713,
      "step": 710
    },
    {
      "epoch": 0.09215410213746321,
      "grad_norm": 1.707152009010315,
      "learning_rate": 3.068060593129934e-06,
      "loss": 0.6659,
      "step": 720
    },
    {
      "epoch": 0.09343402022270575,
      "grad_norm": 0.9780622124671936,
      "learning_rate": 3.1107318113932156e-06,
      "loss": 0.6639,
      "step": 730
    },
    {
      "epoch": 0.09471393830794829,
      "grad_norm": 2.3687374591827393,
      "learning_rate": 3.153403029656497e-06,
      "loss": 0.6666,
      "step": 740
    },
    {
      "epoch": 0.09599385639319083,
      "grad_norm": 0.8506460189819336,
      "learning_rate": 3.1960742479197783e-06,
      "loss": 0.6778,
      "step": 750
    },
    {
      "epoch": 0.09727377447843338,
      "grad_norm": 1.0102107524871826,
      "learning_rate": 3.23874546618306e-06,
      "loss": 0.6635,
      "step": 760
    },
    {
      "epoch": 0.09855369256367592,
      "grad_norm": 1.1480194330215454,
      "learning_rate": 3.2814166844463415e-06,
      "loss": 0.6709,
      "step": 770
    },
    {
      "epoch": 0.09983361064891846,
      "grad_norm": 1.3119179010391235,
      "learning_rate": 3.324087902709623e-06,
      "loss": 0.6564,
      "step": 780
    },
    {
      "epoch": 0.10111352873416102,
      "grad_norm": 0.681377649307251,
      "learning_rate": 3.3667591209729038e-06,
      "loss": 0.6656,
      "step": 790
    },
    {
      "epoch": 0.10239344681940356,
      "grad_norm": 1.022257685661316,
      "learning_rate": 3.4094303392361854e-06,
      "loss": 0.6709,
      "step": 800
    },
    {
      "epoch": 0.10239344681940356,
      "eval_accuracy": 0.66247,
      "eval_loss": 0.6552017331123352,
      "eval_macro_f1": 0.6514270557181998,
      "eval_precision": 0.6718270655389242,
      "eval_recall": 0.6556580720913894,
      "eval_runtime": 1633.5478,
      "eval_samples_per_second": 61.216,
      "eval_steps_per_second": 0.479,
      "step": 800
    },
    {
      "epoch": 0.1036733649046461,
      "grad_norm": 1.0268975496292114,
      "learning_rate": 3.452101557499467e-06,
      "loss": 0.6602,
      "step": 810
    },
    {
      "epoch": 0.10495328298988865,
      "grad_norm": 2.6206603050231934,
      "learning_rate": 3.494772775762748e-06,
      "loss": 0.6705,
      "step": 820
    },
    {
      "epoch": 0.10623320107513119,
      "grad_norm": 1.035157561302185,
      "learning_rate": 3.5374439940260297e-06,
      "loss": 0.6643,
      "step": 830
    },
    {
      "epoch": 0.10751311916037373,
      "grad_norm": 1.736412763595581,
      "learning_rate": 3.5801152122893113e-06,
      "loss": 0.6612,
      "step": 840
    },
    {
      "epoch": 0.10879303724561629,
      "grad_norm": 0.6906964778900146,
      "learning_rate": 3.622786430552593e-06,
      "loss": 0.6576,
      "step": 850
    },
    {
      "epoch": 0.11007295533085883,
      "grad_norm": 0.8148729801177979,
      "learning_rate": 3.665457648815874e-06,
      "loss": 0.6561,
      "step": 860
    },
    {
      "epoch": 0.11135287341610137,
      "grad_norm": 1.0217723846435547,
      "learning_rate": 3.7081288670791556e-06,
      "loss": 0.6581,
      "step": 870
    },
    {
      "epoch": 0.11263279150134392,
      "grad_norm": 1.2743115425109863,
      "learning_rate": 3.750800085342437e-06,
      "loss": 0.6606,
      "step": 880
    },
    {
      "epoch": 0.11391270958658646,
      "grad_norm": 0.6405361294746399,
      "learning_rate": 3.793471303605718e-06,
      "loss": 0.652,
      "step": 890
    },
    {
      "epoch": 0.115192627671829,
      "grad_norm": 0.8357161283493042,
      "learning_rate": 3.836142521869e-06,
      "loss": 0.6558,
      "step": 900
    },
    {
      "epoch": 0.11647254575707154,
      "grad_norm": 1.3505173921585083,
      "learning_rate": 3.8788137401322815e-06,
      "loss": 0.6576,
      "step": 910
    },
    {
      "epoch": 0.1177524638423141,
      "grad_norm": 1.2253319025039673,
      "learning_rate": 3.921484958395563e-06,
      "loss": 0.6548,
      "step": 920
    },
    {
      "epoch": 0.11903238192755664,
      "grad_norm": 0.6269568800926208,
      "learning_rate": 3.964156176658844e-06,
      "loss": 0.6562,
      "step": 930
    },
    {
      "epoch": 0.12031230001279918,
      "grad_norm": 0.9413424134254456,
      "learning_rate": 4.006827394922125e-06,
      "loss": 0.6461,
      "step": 940
    },
    {
      "epoch": 0.12159221809804173,
      "grad_norm": 1.0903513431549072,
      "learning_rate": 4.049498613185407e-06,
      "loss": 0.6457,
      "step": 950
    },
    {
      "epoch": 0.12287213618328427,
      "grad_norm": 2.0023152828216553,
      "learning_rate": 4.092169831448688e-06,
      "loss": 0.6458,
      "step": 960
    },
    {
      "epoch": 0.12415205426852681,
      "grad_norm": 0.9638109803199768,
      "learning_rate": 4.134841049711969e-06,
      "loss": 0.6486,
      "step": 970
    },
    {
      "epoch": 0.12543197235376935,
      "grad_norm": 0.9894546270370483,
      "learning_rate": 4.177512267975251e-06,
      "loss": 0.6453,
      "step": 980
    },
    {
      "epoch": 0.1267118904390119,
      "grad_norm": 1.2289057970046997,
      "learning_rate": 4.220183486238532e-06,
      "loss": 0.6605,
      "step": 990
    },
    {
      "epoch": 0.12799180852425446,
      "grad_norm": 1.119125247001648,
      "learning_rate": 4.262854704501814e-06,
      "loss": 0.6564,
      "step": 1000
    },
    {
      "epoch": 0.12799180852425446,
      "eval_accuracy": 0.69068,
      "eval_loss": 0.6396247148513794,
      "eval_macro_f1": 0.689588440856387,
      "eval_precision": 0.6900076024604327,
      "eval_recall": 0.6894114140312133,
      "eval_runtime": 1634.4321,
      "eval_samples_per_second": 61.183,
      "eval_steps_per_second": 0.478,
      "step": 1000
    },
    {
      "epoch": 0.12927172660949698,
      "grad_norm": 2.02966046333313,
      "learning_rate": 4.3055259227650956e-06,
      "loss": 0.646,
      "step": 1010
    },
    {
      "epoch": 0.13055164469473954,
      "grad_norm": 0.7430800199508667,
      "learning_rate": 4.348197141028377e-06,
      "loss": 0.6453,
      "step": 1020
    },
    {
      "epoch": 0.1318315627799821,
      "grad_norm": 1.031962275505066,
      "learning_rate": 4.390868359291658e-06,
      "loss": 0.6425,
      "step": 1030
    },
    {
      "epoch": 0.13311148086522462,
      "grad_norm": 1.0249106884002686,
      "learning_rate": 4.4335395775549394e-06,
      "loss": 0.6426,
      "step": 1040
    },
    {
      "epoch": 0.13439139895046717,
      "grad_norm": 0.7375448942184448,
      "learning_rate": 4.476210795818221e-06,
      "loss": 0.6401,
      "step": 1050
    },
    {
      "epoch": 0.13567131703570973,
      "grad_norm": 1.295983076095581,
      "learning_rate": 4.518882014081503e-06,
      "loss": 0.6449,
      "step": 1060
    },
    {
      "epoch": 0.13695123512095225,
      "grad_norm": 1.6978998184204102,
      "learning_rate": 4.561553232344783e-06,
      "loss": 0.6484,
      "step": 1070
    },
    {
      "epoch": 0.1382311532061948,
      "grad_norm": 0.7747512459754944,
      "learning_rate": 4.604224450608065e-06,
      "loss": 0.6526,
      "step": 1080
    },
    {
      "epoch": 0.13951107129143736,
      "grad_norm": 0.7895913124084473,
      "learning_rate": 4.6468956688713465e-06,
      "loss": 0.6311,
      "step": 1090
    },
    {
      "epoch": 0.1407909893766799,
      "grad_norm": 0.6975072026252747,
      "learning_rate": 4.689566887134628e-06,
      "loss": 0.6454,
      "step": 1100
    },
    {
      "epoch": 0.14207090746192244,
      "grad_norm": 0.8316811919212341,
      "learning_rate": 4.73223810539791e-06,
      "loss": 0.6447,
      "step": 1110
    },
    {
      "epoch": 0.143350825547165,
      "grad_norm": 0.6887441277503967,
      "learning_rate": 4.774909323661191e-06,
      "loss": 0.6325,
      "step": 1120
    },
    {
      "epoch": 0.14463074363240752,
      "grad_norm": 1.0218420028686523,
      "learning_rate": 4.817580541924473e-06,
      "loss": 0.6308,
      "step": 1130
    },
    {
      "epoch": 0.14591066171765008,
      "grad_norm": 0.7663633823394775,
      "learning_rate": 4.8602517601877535e-06,
      "loss": 0.6227,
      "step": 1140
    },
    {
      "epoch": 0.1471905798028926,
      "grad_norm": 1.5115224123001099,
      "learning_rate": 4.902922978451035e-06,
      "loss": 0.6315,
      "step": 1150
    },
    {
      "epoch": 0.14847049788813516,
      "grad_norm": 0.8011078238487244,
      "learning_rate": 4.945594196714317e-06,
      "loss": 0.6428,
      "step": 1160
    },
    {
      "epoch": 0.1497504159733777,
      "grad_norm": 1.3246116638183594,
      "learning_rate": 4.9882654149775974e-06,
      "loss": 0.6397,
      "step": 1170
    },
    {
      "epoch": 0.15103033405862024,
      "grad_norm": 0.979498565196991,
      "learning_rate": 5.030936633240879e-06,
      "loss": 0.6308,
      "step": 1180
    },
    {
      "epoch": 0.1523102521438628,
      "grad_norm": 0.8756135106086731,
      "learning_rate": 5.073607851504161e-06,
      "loss": 0.6373,
      "step": 1190
    },
    {
      "epoch": 0.15359017022910534,
      "grad_norm": 0.845967710018158,
      "learning_rate": 5.116279069767442e-06,
      "loss": 0.6322,
      "step": 1200
    },
    {
      "epoch": 0.15359017022910534,
      "eval_accuracy": 0.69931,
      "eval_loss": 0.6224722862243652,
      "eval_macro_f1": 0.6985627236986651,
      "eval_precision": 0.6986454338713279,
      "eval_recall": 0.6984991383538209,
      "eval_runtime": 692.1526,
      "eval_samples_per_second": 144.477,
      "eval_steps_per_second": 2.258,
      "step": 1200
    },
    {
      "epoch": 0.15487008831434787,
      "grad_norm": 2.0670483112335205,
      "learning_rate": 1.032010243277849e-05,
      "loss": 0.6386,
      "step": 1210
    },
    {
      "epoch": 0.15615000639959042,
      "grad_norm": 2.482168436050415,
      "learning_rate": 1.0405463081519421e-05,
      "loss": 0.6281,
      "step": 1220
    },
    {
      "epoch": 0.15742992448483298,
      "grad_norm": 1.4530706405639648,
      "learning_rate": 1.049082373026035e-05,
      "loss": 0.6277,
      "step": 1230
    },
    {
      "epoch": 0.1587098425700755,
      "grad_norm": 2.9233577251434326,
      "learning_rate": 1.0576184379001282e-05,
      "loss": 0.6363,
      "step": 1240
    },
    {
      "epoch": 0.15998976065531806,
      "grad_norm": 0.6975642442703247,
      "learning_rate": 1.0661545027742212e-05,
      "loss": 0.6344,
      "step": 1250
    },
    {
      "epoch": 0.1612696787405606,
      "grad_norm": 2.3634731769561768,
      "learning_rate": 1.0746905676483142e-05,
      "loss": 0.6217,
      "step": 1260
    },
    {
      "epoch": 0.16254959682580314,
      "grad_norm": 0.8812585473060608,
      "learning_rate": 1.0832266325224072e-05,
      "loss": 0.6311,
      "step": 1270
    },
    {
      "epoch": 0.1638295149110457,
      "grad_norm": 1.8986798524856567,
      "learning_rate": 1.0917626973965003e-05,
      "loss": 0.6304,
      "step": 1280
    },
    {
      "epoch": 0.16510943299628825,
      "grad_norm": 0.8174984455108643,
      "learning_rate": 1.1002987622705935e-05,
      "loss": 0.6176,
      "step": 1290
    },
    {
      "epoch": 0.16638935108153077,
      "grad_norm": 1.0268642902374268,
      "learning_rate": 1.1088348271446865e-05,
      "loss": 0.6233,
      "step": 1300
    },
    {
      "epoch": 0.16766926916677333,
      "grad_norm": 1.047821044921875,
      "learning_rate": 1.1173708920187793e-05,
      "loss": 0.6127,
      "step": 1310
    },
    {
      "epoch": 0.16894918725201588,
      "grad_norm": 0.9875057935714722,
      "learning_rate": 1.1259069568928724e-05,
      "loss": 0.6211,
      "step": 1320
    },
    {
      "epoch": 0.1702291053372584,
      "grad_norm": 1.0082708597183228,
      "learning_rate": 1.1344430217669656e-05,
      "loss": 0.6202,
      "step": 1330
    },
    {
      "epoch": 0.17150902342250096,
      "grad_norm": 0.7229795455932617,
      "learning_rate": 1.1429790866410586e-05,
      "loss": 0.6071,
      "step": 1340
    },
    {
      "epoch": 0.17278894150774352,
      "grad_norm": 0.7687681317329407,
      "learning_rate": 1.1515151515151517e-05,
      "loss": 0.6218,
      "step": 1350
    },
    {
      "epoch": 0.17406885959298604,
      "grad_norm": 1.1643378734588623,
      "learning_rate": 1.1600512163892445e-05,
      "loss": 0.6208,
      "step": 1360
    },
    {
      "epoch": 0.1753487776782286,
      "grad_norm": 0.7338509559631348,
      "learning_rate": 1.1685872812633377e-05,
      "loss": 0.6105,
      "step": 1370
    },
    {
      "epoch": 0.17662869576347115,
      "grad_norm": 0.7444656491279602,
      "learning_rate": 1.1771233461374307e-05,
      "loss": 0.6183,
      "step": 1380
    },
    {
      "epoch": 0.17790861384871368,
      "grad_norm": 1.1902283430099487,
      "learning_rate": 1.1856594110115238e-05,
      "loss": 0.6209,
      "step": 1390
    },
    {
      "epoch": 0.17918853193395623,
      "grad_norm": 1.0872228145599365,
      "learning_rate": 1.194195475885617e-05,
      "loss": 0.6172,
      "step": 1400
    },
    {
      "epoch": 0.18046845001919878,
      "grad_norm": 1.8439041376113892,
      "learning_rate": 1.2027315407597098e-05,
      "loss": 0.6086,
      "step": 1410
    },
    {
      "epoch": 0.1817483681044413,
      "grad_norm": 0.8699899911880493,
      "learning_rate": 1.211267605633803e-05,
      "loss": 0.6247,
      "step": 1420
    },
    {
      "epoch": 0.18302828618968386,
      "grad_norm": 1.1468650102615356,
      "learning_rate": 1.2198036705078959e-05,
      "loss": 0.6114,
      "step": 1430
    },
    {
      "epoch": 0.18430820427492642,
      "grad_norm": 1.0204615592956543,
      "learning_rate": 1.228339735381989e-05,
      "loss": 0.6015,
      "step": 1440
    },
    {
      "epoch": 0.18558812236016894,
      "grad_norm": 2.660583257675171,
      "learning_rate": 1.236875800256082e-05,
      "loss": 0.6002,
      "step": 1450
    },
    {
      "epoch": 0.1868680404454115,
      "grad_norm": 1.1916371583938599,
      "learning_rate": 1.2454118651301752e-05,
      "loss": 0.6076,
      "step": 1460
    },
    {
      "epoch": 0.18814795853065402,
      "grad_norm": 0.8097987174987793,
      "learning_rate": 1.253947930004268e-05,
      "loss": 0.6126,
      "step": 1470
    },
    {
      "epoch": 0.18942787661589658,
      "grad_norm": 1.978747010231018,
      "learning_rate": 1.2624839948783612e-05,
      "loss": 0.5962,
      "step": 1480
    },
    {
      "epoch": 0.19070779470113913,
      "grad_norm": 1.0975425243377686,
      "learning_rate": 1.2710200597524543e-05,
      "loss": 0.587,
      "step": 1490
    },
    {
      "epoch": 0.19198771278638166,
      "grad_norm": 2.150351047515869,
      "learning_rate": 1.2795561246265473e-05,
      "loss": 0.5962,
      "step": 1500
    },
    {
      "epoch": 0.19198771278638166,
      "eval_accuracy": 0.72526,
      "eval_loss": 0.5823243856430054,
      "eval_macro_f1": 0.7252418839884757,
      "eval_precision": 0.7275813809678697,
      "eval_recall": 0.7271797416587706,
      "eval_runtime": 708.4119,
      "eval_samples_per_second": 141.161,
      "eval_steps_per_second": 1.104,
      "step": 1500
    },
    {
      "epoch": 0.1932676308716242,
      "grad_norm": 1.8355706930160522,
      "learning_rate": 1.2880921895006405e-05,
      "loss": 0.6133,
      "step": 1510
    },
    {
      "epoch": 0.19454754895686677,
      "grad_norm": 1.4349753856658936,
      "learning_rate": 1.2966282543747333e-05,
      "loss": 0.6062,
      "step": 1520
    },
    {
      "epoch": 0.1958274670421093,
      "grad_norm": 1.8322596549987793,
      "learning_rate": 1.3051643192488264e-05,
      "loss": 0.5982,
      "step": 1530
    },
    {
      "epoch": 0.19710738512735185,
      "grad_norm": 1.509338617324829,
      "learning_rate": 1.3137003841229194e-05,
      "loss": 0.5961,
      "step": 1540
    },
    {
      "epoch": 0.1983873032125944,
      "grad_norm": 0.9844911098480225,
      "learning_rate": 1.3222364489970126e-05,
      "loss": 0.583,
      "step": 1550
    },
    {
      "epoch": 0.19966722129783693,
      "grad_norm": 1.035900354385376,
      "learning_rate": 1.3307725138711055e-05,
      "loss": 0.598,
      "step": 1560
    },
    {
      "epoch": 0.20094713938307948,
      "grad_norm": 0.7891924977302551,
      "learning_rate": 1.3393085787451985e-05,
      "loss": 0.6064,
      "step": 1570
    },
    {
      "epoch": 0.20222705746832204,
      "grad_norm": 1.4326112270355225,
      "learning_rate": 1.3478446436192915e-05,
      "loss": 0.5954,
      "step": 1580
    },
    {
      "epoch": 0.20350697555356456,
      "grad_norm": 0.9054266214370728,
      "learning_rate": 1.3563807084933847e-05,
      "loss": 0.59,
      "step": 1590
    },
    {
      "epoch": 0.20478689363880712,
      "grad_norm": 1.0216469764709473,
      "learning_rate": 1.3649167733674778e-05,
      "loss": 0.5845,
      "step": 1600
    },
    {
      "epoch": 0.20606681172404967,
      "grad_norm": 1.0591177940368652,
      "learning_rate": 1.3734528382415708e-05,
      "loss": 0.599,
      "step": 1610
    },
    {
      "epoch": 0.2073467298092922,
      "grad_norm": 0.7826931476593018,
      "learning_rate": 1.3819889031156638e-05,
      "loss": 0.5898,
      "step": 1620
    },
    {
      "epoch": 0.20862664789453475,
      "grad_norm": 0.755007803440094,
      "learning_rate": 1.3905249679897568e-05,
      "loss": 0.5769,
      "step": 1630
    },
    {
      "epoch": 0.2099065659797773,
      "grad_norm": 1.3526090383529663,
      "learning_rate": 1.3990610328638499e-05,
      "loss": 0.5819,
      "step": 1640
    },
    {
      "epoch": 0.21118648406501983,
      "grad_norm": 0.7078200578689575,
      "learning_rate": 1.4075970977379429e-05,
      "loss": 0.5884,
      "step": 1650
    },
    {
      "epoch": 0.21246640215026238,
      "grad_norm": 0.80736243724823,
      "learning_rate": 1.416133162612036e-05,
      "loss": 0.5991,
      "step": 1660
    },
    {
      "epoch": 0.21374632023550494,
      "grad_norm": 0.6882274150848389,
      "learning_rate": 1.4246692274861289e-05,
      "loss": 0.5825,
      "step": 1670
    },
    {
      "epoch": 0.21502623832074746,
      "grad_norm": 0.7223654985427856,
      "learning_rate": 1.433205292360222e-05,
      "loss": 0.5689,
      "step": 1680
    },
    {
      "epoch": 0.21630615640599002,
      "grad_norm": 0.6446322798728943,
      "learning_rate": 1.4417413572343152e-05,
      "loss": 0.5773,
      "step": 1690
    },
    {
      "epoch": 0.21758607449123257,
      "grad_norm": 1.505007266998291,
      "learning_rate": 1.4502774221084081e-05,
      "loss": 0.5677,
      "step": 1700
    },
    {
      "epoch": 0.2188659925764751,
      "grad_norm": 1.1939284801483154,
      "learning_rate": 1.4588134869825013e-05,
      "loss": 0.5805,
      "step": 1710
    },
    {
      "epoch": 0.22014591066171765,
      "grad_norm": 2.848264217376709,
      "learning_rate": 1.4673495518565943e-05,
      "loss": 0.5712,
      "step": 1720
    },
    {
      "epoch": 0.2214258287469602,
      "grad_norm": 1.2117749452590942,
      "learning_rate": 1.4758856167306873e-05,
      "loss": 0.582,
      "step": 1730
    },
    {
      "epoch": 0.22270574683220273,
      "grad_norm": 2.2297441959381104,
      "learning_rate": 1.4844216816047802e-05,
      "loss": 0.5751,
      "step": 1740
    },
    {
      "epoch": 0.2239856649174453,
      "grad_norm": 1.211978554725647,
      "learning_rate": 1.4929577464788734e-05,
      "loss": 0.5607,
      "step": 1750
    },
    {
      "epoch": 0.22526558300268784,
      "grad_norm": 0.8025451898574829,
      "learning_rate": 1.5014938113529664e-05,
      "loss": 0.5803,
      "step": 1760
    },
    {
      "epoch": 0.22654550108793037,
      "grad_norm": 1.2300900220870972,
      "learning_rate": 1.5100298762270595e-05,
      "loss": 0.5827,
      "step": 1770
    },
    {
      "epoch": 0.22782541917317292,
      "grad_norm": 1.0382864475250244,
      "learning_rate": 1.5185659411011523e-05,
      "loss": 0.5916,
      "step": 1780
    },
    {
      "epoch": 0.22910533725841545,
      "grad_norm": 1.4876656532287598,
      "learning_rate": 1.5271020059752453e-05,
      "loss": 0.5536,
      "step": 1790
    },
    {
      "epoch": 0.230385255343658,
      "grad_norm": 1.1409252882003784,
      "learning_rate": 1.5356380708493386e-05,
      "loss": 0.5484,
      "step": 1800
    },
    {
      "epoch": 0.23166517342890056,
      "grad_norm": 0.7180280089378357,
      "learning_rate": 1.5441741357234316e-05,
      "loss": 0.5712,
      "step": 1810
    },
    {
      "epoch": 0.23294509151414308,
      "grad_norm": 1.5890289545059204,
      "learning_rate": 1.5527102005975246e-05,
      "loss": 0.5758,
      "step": 1820
    },
    {
      "epoch": 0.23422500959938564,
      "grad_norm": 0.6635769605636597,
      "learning_rate": 1.5612462654716176e-05,
      "loss": 0.5543,
      "step": 1830
    },
    {
      "epoch": 0.2355049276846282,
      "grad_norm": 0.8864602446556091,
      "learning_rate": 1.5697823303457106e-05,
      "loss": 0.5536,
      "step": 1840
    },
    {
      "epoch": 0.23678484576987072,
      "grad_norm": 0.8372675776481628,
      "learning_rate": 1.578318395219804e-05,
      "loss": 0.5743,
      "step": 1850
    },
    {
      "epoch": 0.23806476385511327,
      "grad_norm": 0.7532811164855957,
      "learning_rate": 1.586854460093897e-05,
      "loss": 0.5593,
      "step": 1860
    },
    {
      "epoch": 0.23934468194035582,
      "grad_norm": 0.7231799960136414,
      "learning_rate": 1.59539052496799e-05,
      "loss": 0.5831,
      "step": 1870
    },
    {
      "epoch": 0.24062460002559835,
      "grad_norm": 1.124228596687317,
      "learning_rate": 1.603926589842083e-05,
      "loss": 0.5936,
      "step": 1880
    },
    {
      "epoch": 0.2419045181108409,
      "grad_norm": 1.0107067823410034,
      "learning_rate": 1.612462654716176e-05,
      "loss": 0.5598,
      "step": 1890
    },
    {
      "epoch": 0.24318443619608346,
      "grad_norm": 1.6117154359817505,
      "learning_rate": 1.620998719590269e-05,
      "loss": 0.5605,
      "step": 1900
    },
    {
      "epoch": 0.24446435428132599,
      "grad_norm": 1.5564496517181396,
      "learning_rate": 1.629534784464362e-05,
      "loss": 0.5666,
      "step": 1910
    },
    {
      "epoch": 0.24574427236656854,
      "grad_norm": 0.8275917768478394,
      "learning_rate": 1.638070849338455e-05,
      "loss": 0.5597,
      "step": 1920
    },
    {
      "epoch": 0.2470241904518111,
      "grad_norm": 0.8956324458122253,
      "learning_rate": 1.646606914212548e-05,
      "loss": 0.5453,
      "step": 1930
    },
    {
      "epoch": 0.24830410853705362,
      "grad_norm": 1.9644063711166382,
      "learning_rate": 1.655142979086641e-05,
      "loss": 0.5732,
      "step": 1940
    },
    {
      "epoch": 0.24958402662229617,
      "grad_norm": 0.7495090365409851,
      "learning_rate": 1.663679043960734e-05,
      "loss": 0.5549,
      "step": 1950
    },
    {
      "epoch": 0.2508639447075387,
      "grad_norm": 0.9346863627433777,
      "learning_rate": 1.6722151088348274e-05,
      "loss": 0.5803,
      "step": 1960
    },
    {
      "epoch": 0.2521438627927813,
      "grad_norm": 0.5847868919372559,
      "learning_rate": 1.6807511737089204e-05,
      "loss": 0.5705,
      "step": 1970
    },
    {
      "epoch": 0.2534237808780238,
      "grad_norm": 1.5892306566238403,
      "learning_rate": 1.6892872385830134e-05,
      "loss": 0.5371,
      "step": 1980
    },
    {
      "epoch": 0.25470369896326633,
      "grad_norm": 0.8782947063446045,
      "learning_rate": 1.6978233034571063e-05,
      "loss": 0.565,
      "step": 1990
    },
    {
      "epoch": 0.2559836170485089,
      "grad_norm": 1.0143548250198364,
      "learning_rate": 1.7063593683311993e-05,
      "loss": 0.557,
      "step": 2000
    },
    {
      "epoch": 0.2559836170485089,
      "eval_accuracy": 0.74427,
      "eval_loss": 0.5308485627174377,
      "eval_macro_f1": 0.743516265066894,
      "eval_precision": 0.7538937627554241,
      "eval_recall": 0.7483595020974525,
      "eval_runtime": 708.2272,
      "eval_samples_per_second": 141.198,
      "eval_steps_per_second": 1.104,
      "step": 2000
    },
    {
      "epoch": 0.25726353513375144,
      "grad_norm": 0.9080437421798706,
      "learning_rate": 1.7148954332052926e-05,
      "loss": 0.5478,
      "step": 2010
    },
    {
      "epoch": 0.25854345321899397,
      "grad_norm": 1.1889078617095947,
      "learning_rate": 1.7234314980793856e-05,
      "loss": 0.5394,
      "step": 2020
    },
    {
      "epoch": 0.25982337130423655,
      "grad_norm": 1.7290714979171753,
      "learning_rate": 1.7319675629534786e-05,
      "loss": 0.5385,
      "step": 2030
    },
    {
      "epoch": 0.2611032893894791,
      "grad_norm": 0.9356957077980042,
      "learning_rate": 1.7405036278275716e-05,
      "loss": 0.5392,
      "step": 2040
    },
    {
      "epoch": 0.2623832074747216,
      "grad_norm": 0.765798032283783,
      "learning_rate": 1.7490396927016646e-05,
      "loss": 0.5442,
      "step": 2050
    },
    {
      "epoch": 0.2636631255599642,
      "grad_norm": 1.6606382131576538,
      "learning_rate": 1.7575757575757576e-05,
      "loss": 0.5331,
      "step": 2060
    },
    {
      "epoch": 0.2649430436452067,
      "grad_norm": 0.7928575277328491,
      "learning_rate": 1.766111822449851e-05,
      "loss": 0.5483,
      "step": 2070
    },
    {
      "epoch": 0.26622296173044924,
      "grad_norm": 2.2261486053466797,
      "learning_rate": 1.774647887323944e-05,
      "loss": 0.5536,
      "step": 2080
    },
    {
      "epoch": 0.2675028798156918,
      "grad_norm": 0.9795323610305786,
      "learning_rate": 1.783183952198037e-05,
      "loss": 0.5584,
      "step": 2090
    },
    {
      "epoch": 0.26878279790093434,
      "grad_norm": 2.373581886291504,
      "learning_rate": 1.7917200170721298e-05,
      "loss": 0.5475,
      "step": 2100
    },
    {
      "epoch": 0.27006271598617687,
      "grad_norm": 0.9658742547035217,
      "learning_rate": 1.8002560819462228e-05,
      "loss": 0.5577,
      "step": 2110
    },
    {
      "epoch": 0.27134263407141945,
      "grad_norm": 0.8572520017623901,
      "learning_rate": 1.808792146820316e-05,
      "loss": 0.5301,
      "step": 2120
    },
    {
      "epoch": 0.272622552156662,
      "grad_norm": 0.7225477695465088,
      "learning_rate": 1.817328211694409e-05,
      "loss": 0.5626,
      "step": 2130
    },
    {
      "epoch": 0.2739024702419045,
      "grad_norm": 1.4276026487350464,
      "learning_rate": 1.825864276568502e-05,
      "loss": 0.5208,
      "step": 2140
    },
    {
      "epoch": 0.2751823883271471,
      "grad_norm": 1.7469464540481567,
      "learning_rate": 1.834400341442595e-05,
      "loss": 0.5669,
      "step": 2150
    },
    {
      "epoch": 0.2764623064123896,
      "grad_norm": 1.250931978225708,
      "learning_rate": 1.842936406316688e-05,
      "loss": 0.5577,
      "step": 2160
    },
    {
      "epoch": 0.27774222449763214,
      "grad_norm": 1.011313557624817,
      "learning_rate": 1.8514724711907814e-05,
      "loss": 0.5494,
      "step": 2170
    },
    {
      "epoch": 0.2790221425828747,
      "grad_norm": 0.7606518268585205,
      "learning_rate": 1.8600085360648744e-05,
      "loss": 0.5391,
      "step": 2180
    },
    {
      "epoch": 0.28030206066811725,
      "grad_norm": 0.8173381090164185,
      "learning_rate": 1.8685446009389673e-05,
      "loss": 0.5262,
      "step": 2190
    },
    {
      "epoch": 0.2815819787533598,
      "grad_norm": 0.7178583145141602,
      "learning_rate": 1.8770806658130603e-05,
      "loss": 0.5182,
      "step": 2200
    },
    {
      "epoch": 0.28286189683860236,
      "grad_norm": 0.6852149367332458,
      "learning_rate": 1.8856167306871533e-05,
      "loss": 0.528,
      "step": 2210
    },
    {
      "epoch": 0.2841418149238449,
      "grad_norm": 1.9807002544403076,
      "learning_rate": 1.8941527955612463e-05,
      "loss": 0.5315,
      "step": 2220
    },
    {
      "epoch": 0.2854217330090874,
      "grad_norm": 0.7938176989555359,
      "learning_rate": 1.9026888604353396e-05,
      "loss": 0.5324,
      "step": 2230
    },
    {
      "epoch": 0.28670165109433,
      "grad_norm": 0.8181535005569458,
      "learning_rate": 1.9112249253094326e-05,
      "loss": 0.51,
      "step": 2240
    },
    {
      "epoch": 0.2879815691795725,
      "grad_norm": 0.8722071051597595,
      "learning_rate": 1.9197609901835256e-05,
      "loss": 0.5373,
      "step": 2250
    },
    {
      "epoch": 0.28926148726481504,
      "grad_norm": 1.371565818786621,
      "learning_rate": 1.9282970550576186e-05,
      "loss": 0.5227,
      "step": 2260
    },
    {
      "epoch": 0.29054140535005757,
      "grad_norm": 0.6769284605979919,
      "learning_rate": 1.9368331199317115e-05,
      "loss": 0.5421,
      "step": 2270
    },
    {
      "epoch": 0.29182132343530015,
      "grad_norm": 2.1016697883605957,
      "learning_rate": 1.945369184805805e-05,
      "loss": 0.5299,
      "step": 2280
    },
    {
      "epoch": 0.2931012415205427,
      "grad_norm": 1.7011477947235107,
      "learning_rate": 1.953905249679898e-05,
      "loss": 0.5403,
      "step": 2290
    },
    {
      "epoch": 0.2943811596057852,
      "grad_norm": 1.0961430072784424,
      "learning_rate": 1.962441314553991e-05,
      "loss": 0.534,
      "step": 2300
    },
    {
      "epoch": 0.2956610776910278,
      "grad_norm": 0.8536306023597717,
      "learning_rate": 1.9709773794280838e-05,
      "loss": 0.5562,
      "step": 2310
    },
    {
      "epoch": 0.2969409957762703,
      "grad_norm": 1.046201467514038,
      "learning_rate": 1.9795134443021768e-05,
      "loss": 0.5418,
      "step": 2320
    },
    {
      "epoch": 0.29822091386151284,
      "grad_norm": 1.38905930519104,
      "learning_rate": 1.9880495091762698e-05,
      "loss": 0.5557,
      "step": 2330
    },
    {
      "epoch": 0.2995008319467554,
      "grad_norm": 2.44215989112854,
      "learning_rate": 1.996585574050363e-05,
      "loss": 0.5285,
      "step": 2340
    },
    {
      "epoch": 0.30078075003199795,
      "grad_norm": 1.4736480712890625,
      "learning_rate": 1.9994311717861208e-05,
      "loss": 0.513,
      "step": 2350
    },
    {
      "epoch": 0.30206066811724047,
      "grad_norm": 0.7968178987503052,
      "learning_rate": 1.9984831247629883e-05,
      "loss": 0.5249,
      "step": 2360
    },
    {
      "epoch": 0.30334058620248305,
      "grad_norm": 1.6530416011810303,
      "learning_rate": 1.997535077739856e-05,
      "loss": 0.5329,
      "step": 2370
    },
    {
      "epoch": 0.3046205042877256,
      "grad_norm": 1.0949759483337402,
      "learning_rate": 1.9965870307167238e-05,
      "loss": 0.5164,
      "step": 2380
    },
    {
      "epoch": 0.3059004223729681,
      "grad_norm": 1.2724019289016724,
      "learning_rate": 1.9956389836935916e-05,
      "loss": 0.5293,
      "step": 2390
    },
    {
      "epoch": 0.3071803404582107,
      "grad_norm": 1.8785524368286133,
      "learning_rate": 1.994690936670459e-05,
      "loss": 0.5286,
      "step": 2400
    },
    {
      "epoch": 0.3084602585434532,
      "grad_norm": 2.1517200469970703,
      "learning_rate": 1.9937428896473267e-05,
      "loss": 0.5273,
      "step": 2410
    },
    {
      "epoch": 0.30974017662869574,
      "grad_norm": 0.999515175819397,
      "learning_rate": 1.9927948426241942e-05,
      "loss": 0.5115,
      "step": 2420
    },
    {
      "epoch": 0.3110200947139383,
      "grad_norm": 2.214357852935791,
      "learning_rate": 1.991846795601062e-05,
      "loss": 0.5145,
      "step": 2430
    },
    {
      "epoch": 0.31230001279918085,
      "grad_norm": 1.8795071840286255,
      "learning_rate": 1.9908987485779297e-05,
      "loss": 0.509,
      "step": 2440
    },
    {
      "epoch": 0.3135799308844234,
      "grad_norm": 0.9374184608459473,
      "learning_rate": 1.989950701554797e-05,
      "loss": 0.5112,
      "step": 2450
    },
    {
      "epoch": 0.31485984896966596,
      "grad_norm": 0.7281107306480408,
      "learning_rate": 1.989002654531665e-05,
      "loss": 0.5029,
      "step": 2460
    },
    {
      "epoch": 0.3161397670549085,
      "grad_norm": 2.6096737384796143,
      "learning_rate": 1.9880546075085327e-05,
      "loss": 0.5302,
      "step": 2470
    },
    {
      "epoch": 0.317419685140151,
      "grad_norm": 0.7463436126708984,
      "learning_rate": 1.9871065604854004e-05,
      "loss": 0.5249,
      "step": 2480
    },
    {
      "epoch": 0.3186996032253936,
      "grad_norm": 0.7995405197143555,
      "learning_rate": 1.986158513462268e-05,
      "loss": 0.5171,
      "step": 2490
    },
    {
      "epoch": 0.3199795213106361,
      "grad_norm": 0.873103678226471,
      "learning_rate": 1.9852104664391356e-05,
      "loss": 0.5125,
      "step": 2500
    },
    {
      "epoch": 0.32125943939587864,
      "grad_norm": 1.5319750308990479,
      "learning_rate": 1.984262419416003e-05,
      "loss": 0.527,
      "step": 2510
    },
    {
      "epoch": 0.3225393574811212,
      "grad_norm": 1.717421531677246,
      "learning_rate": 1.983314372392871e-05,
      "loss": 0.5415,
      "step": 2520
    },
    {
      "epoch": 0.32381927556636375,
      "grad_norm": 0.7133586406707764,
      "learning_rate": 1.9823663253697386e-05,
      "loss": 0.5251,
      "step": 2530
    },
    {
      "epoch": 0.3250991936516063,
      "grad_norm": 0.7959030270576477,
      "learning_rate": 1.981418278346606e-05,
      "loss": 0.5127,
      "step": 2540
    },
    {
      "epoch": 0.32637911173684886,
      "grad_norm": 2.4905223846435547,
      "learning_rate": 1.9804702313234738e-05,
      "loss": 0.4944,
      "step": 2550
    },
    {
      "epoch": 0.3276590298220914,
      "grad_norm": 1.0039845705032349,
      "learning_rate": 1.9795221843003412e-05,
      "loss": 0.534,
      "step": 2560
    },
    {
      "epoch": 0.3289389479073339,
      "grad_norm": 1.2181288003921509,
      "learning_rate": 1.978574137277209e-05,
      "loss": 0.5198,
      "step": 2570
    },
    {
      "epoch": 0.3302188659925765,
      "grad_norm": 0.9166068434715271,
      "learning_rate": 1.9776260902540768e-05,
      "loss": 0.5373,
      "step": 2580
    },
    {
      "epoch": 0.331498784077819,
      "grad_norm": 0.938994288444519,
      "learning_rate": 1.9766780432309445e-05,
      "loss": 0.5179,
      "step": 2590
    },
    {
      "epoch": 0.33277870216306155,
      "grad_norm": 0.8827023506164551,
      "learning_rate": 1.975729996207812e-05,
      "loss": 0.5249,
      "step": 2600
    },
    {
      "epoch": 0.33405862024830413,
      "grad_norm": 0.8544501066207886,
      "learning_rate": 1.9747819491846797e-05,
      "loss": 0.5172,
      "step": 2610
    },
    {
      "epoch": 0.33533853833354665,
      "grad_norm": 0.7890387177467346,
      "learning_rate": 1.9738339021615475e-05,
      "loss": 0.5207,
      "step": 2620
    },
    {
      "epoch": 0.3366184564187892,
      "grad_norm": 0.7724694013595581,
      "learning_rate": 1.9728858551384153e-05,
      "loss": 0.5065,
      "step": 2630
    },
    {
      "epoch": 0.33789837450403176,
      "grad_norm": 0.7064672708511353,
      "learning_rate": 1.9719378081152827e-05,
      "loss": 0.4828,
      "step": 2640
    },
    {
      "epoch": 0.3391782925892743,
      "grad_norm": 0.9625025391578674,
      "learning_rate": 1.97098976109215e-05,
      "loss": 0.5108,
      "step": 2650
    },
    {
      "epoch": 0.3404582106745168,
      "grad_norm": 2.1604244709014893,
      "learning_rate": 1.970041714069018e-05,
      "loss": 0.5129,
      "step": 2660
    },
    {
      "epoch": 0.3417381287597594,
      "grad_norm": 0.8561355471611023,
      "learning_rate": 1.9690936670458856e-05,
      "loss": 0.521,
      "step": 2670
    },
    {
      "epoch": 0.3430180468450019,
      "grad_norm": 0.8122753500938416,
      "learning_rate": 1.9681456200227534e-05,
      "loss": 0.5039,
      "step": 2680
    },
    {
      "epoch": 0.34429796493024445,
      "grad_norm": 0.848838210105896,
      "learning_rate": 1.967197572999621e-05,
      "loss": 0.5559,
      "step": 2690
    },
    {
      "epoch": 0.34557788301548703,
      "grad_norm": 0.8424274325370789,
      "learning_rate": 1.9662495259764886e-05,
      "loss": 0.5299,
      "step": 2700
    },
    {
      "epoch": 0.34685780110072956,
      "grad_norm": 1.0475687980651855,
      "learning_rate": 1.9653014789533564e-05,
      "loss": 0.5003,
      "step": 2710
    },
    {
      "epoch": 0.3481377191859721,
      "grad_norm": 0.9720590710639954,
      "learning_rate": 1.964353431930224e-05,
      "loss": 0.4976,
      "step": 2720
    },
    {
      "epoch": 0.34941763727121466,
      "grad_norm": 1.1562892198562622,
      "learning_rate": 1.9634053849070916e-05,
      "loss": 0.4728,
      "step": 2730
    },
    {
      "epoch": 0.3506975553564572,
      "grad_norm": 1.2779773473739624,
      "learning_rate": 1.9624573378839593e-05,
      "loss": 0.4933,
      "step": 2740
    },
    {
      "epoch": 0.3519774734416997,
      "grad_norm": 1.047945499420166,
      "learning_rate": 1.9615092908608268e-05,
      "loss": 0.518,
      "step": 2750
    },
    {
      "epoch": 0.3532573915269423,
      "grad_norm": 0.7196894884109497,
      "learning_rate": 1.9605612438376945e-05,
      "loss": 0.5006,
      "step": 2760
    },
    {
      "epoch": 0.3545373096121848,
      "grad_norm": 0.9644359350204468,
      "learning_rate": 1.9596131968145623e-05,
      "loss": 0.5083,
      "step": 2770
    },
    {
      "epoch": 0.35581722769742735,
      "grad_norm": 0.9720232486724854,
      "learning_rate": 1.9586651497914297e-05,
      "loss": 0.5194,
      "step": 2780
    },
    {
      "epoch": 0.35709714578266993,
      "grad_norm": 0.8277772665023804,
      "learning_rate": 1.9577171027682975e-05,
      "loss": 0.5121,
      "step": 2790
    },
    {
      "epoch": 0.35837706386791246,
      "grad_norm": 0.7811515927314758,
      "learning_rate": 1.956769055745165e-05,
      "loss": 0.5106,
      "step": 2800
    },
    {
      "epoch": 0.359656981953155,
      "grad_norm": 0.7438198328018188,
      "learning_rate": 1.9558210087220327e-05,
      "loss": 0.5183,
      "step": 2810
    },
    {
      "epoch": 0.36093690003839757,
      "grad_norm": 1.1424624919891357,
      "learning_rate": 1.9548729616989005e-05,
      "loss": 0.4973,
      "step": 2820
    },
    {
      "epoch": 0.3622168181236401,
      "grad_norm": 1.0706868171691895,
      "learning_rate": 1.9539249146757682e-05,
      "loss": 0.5063,
      "step": 2830
    },
    {
      "epoch": 0.3634967362088826,
      "grad_norm": 1.1213792562484741,
      "learning_rate": 1.9529768676526357e-05,
      "loss": 0.4991,
      "step": 2840
    },
    {
      "epoch": 0.3647766542941252,
      "grad_norm": 1.0018831491470337,
      "learning_rate": 1.9520288206295034e-05,
      "loss": 0.5004,
      "step": 2850
    },
    {
      "epoch": 0.36605657237936773,
      "grad_norm": 0.8369762897491455,
      "learning_rate": 1.9510807736063712e-05,
      "loss": 0.5341,
      "step": 2860
    },
    {
      "epoch": 0.36733649046461025,
      "grad_norm": 0.8711100220680237,
      "learning_rate": 1.9501327265832386e-05,
      "loss": 0.4788,
      "step": 2870
    },
    {
      "epoch": 0.36861640854985284,
      "grad_norm": 1.0853824615478516,
      "learning_rate": 1.9491846795601064e-05,
      "loss": 0.5419,
      "step": 2880
    },
    {
      "epoch": 0.36989632663509536,
      "grad_norm": 1.4816603660583496,
      "learning_rate": 1.9482366325369738e-05,
      "loss": 0.5252,
      "step": 2890
    },
    {
      "epoch": 0.3711762447203379,
      "grad_norm": 1.3375904560089111,
      "learning_rate": 1.9472885855138416e-05,
      "loss": 0.5127,
      "step": 2900
    },
    {
      "epoch": 0.3724561628055804,
      "grad_norm": 0.7361944913864136,
      "learning_rate": 1.9463405384907093e-05,
      "loss": 0.5278,
      "step": 2910
    },
    {
      "epoch": 0.373736080890823,
      "grad_norm": 1.1056461334228516,
      "learning_rate": 1.945392491467577e-05,
      "loss": 0.5072,
      "step": 2920
    },
    {
      "epoch": 0.3750159989760655,
      "grad_norm": 0.921783447265625,
      "learning_rate": 1.9444444444444445e-05,
      "loss": 0.5161,
      "step": 2930
    },
    {
      "epoch": 0.37629591706130805,
      "grad_norm": 1.15033757686615,
      "learning_rate": 1.9434963974213123e-05,
      "loss": 0.5329,
      "step": 2940
    },
    {
      "epoch": 0.37757583514655063,
      "grad_norm": 1.3983162641525269,
      "learning_rate": 1.94254835039818e-05,
      "loss": 0.5034,
      "step": 2950
    },
    {
      "epoch": 0.37885575323179316,
      "grad_norm": 1.0419155359268188,
      "learning_rate": 1.9416003033750475e-05,
      "loss": 0.5281,
      "step": 2960
    },
    {
      "epoch": 0.3801356713170357,
      "grad_norm": 2.478982448577881,
      "learning_rate": 1.9406522563519153e-05,
      "loss": 0.4843,
      "step": 2970
    },
    {
      "epoch": 0.38141558940227827,
      "grad_norm": 0.7638375163078308,
      "learning_rate": 1.9397042093287827e-05,
      "loss": 0.5214,
      "step": 2980
    },
    {
      "epoch": 0.3826955074875208,
      "grad_norm": 2.18105411529541,
      "learning_rate": 1.9387561623056505e-05,
      "loss": 0.5373,
      "step": 2990
    },
    {
      "epoch": 0.3839754255727633,
      "grad_norm": 1.0936481952667236,
      "learning_rate": 1.9378081152825182e-05,
      "loss": 0.5187,
      "step": 3000
    },
    {
      "epoch": 0.3852553436580059,
      "grad_norm": 1.0057384967803955,
      "learning_rate": 1.936860068259386e-05,
      "loss": 0.5179,
      "step": 3010
    },
    {
      "epoch": 0.3865352617432484,
      "grad_norm": 1.9525461196899414,
      "learning_rate": 1.9359120212362534e-05,
      "loss": 0.4964,
      "step": 3020
    },
    {
      "epoch": 0.38781517982849095,
      "grad_norm": 0.9308017492294312,
      "learning_rate": 1.9349639742131212e-05,
      "loss": 0.4751,
      "step": 3030
    },
    {
      "epoch": 0.38909509791373353,
      "grad_norm": 0.6847271919250488,
      "learning_rate": 1.9340159271899886e-05,
      "loss": 0.5049,
      "step": 3040
    },
    {
      "epoch": 0.39037501599897606,
      "grad_norm": 0.9826173186302185,
      "learning_rate": 1.9330678801668564e-05,
      "loss": 0.5089,
      "step": 3050
    },
    {
      "epoch": 0.3916549340842186,
      "grad_norm": 1.1526622772216797,
      "learning_rate": 1.932119833143724e-05,
      "loss": 0.5263,
      "step": 3060
    },
    {
      "epoch": 0.39293485216946117,
      "grad_norm": 1.3890351057052612,
      "learning_rate": 1.9311717861205916e-05,
      "loss": 0.5293,
      "step": 3070
    },
    {
      "epoch": 0.3942147702547037,
      "grad_norm": 1.8960402011871338,
      "learning_rate": 1.9302237390974593e-05,
      "loss": 0.4944,
      "step": 3080
    },
    {
      "epoch": 0.3954946883399462,
      "grad_norm": 3.4239537715911865,
      "learning_rate": 1.929275692074327e-05,
      "loss": 0.5281,
      "step": 3090
    },
    {
      "epoch": 0.3967746064251888,
      "grad_norm": 0.8984873294830322,
      "learning_rate": 1.928327645051195e-05,
      "loss": 0.5169,
      "step": 3100
    },
    {
      "epoch": 0.39805452451043133,
      "grad_norm": 1.3759665489196777,
      "learning_rate": 1.9273795980280623e-05,
      "loss": 0.5086,
      "step": 3110
    },
    {
      "epoch": 0.39933444259567386,
      "grad_norm": 1.0449821949005127,
      "learning_rate": 1.92643155100493e-05,
      "loss": 0.4921,
      "step": 3120
    },
    {
      "epoch": 0.40061436068091644,
      "grad_norm": 0.9045126438140869,
      "learning_rate": 1.9254835039817975e-05,
      "loss": 0.4888,
      "step": 3130
    },
    {
      "epoch": 0.40189427876615896,
      "grad_norm": 0.8578636050224304,
      "learning_rate": 1.9245354569586653e-05,
      "loss": 0.4956,
      "step": 3140
    },
    {
      "epoch": 0.4031741968514015,
      "grad_norm": 1.2654173374176025,
      "learning_rate": 1.923587409935533e-05,
      "loss": 0.4942,
      "step": 3150
    },
    {
      "epoch": 0.40445411493664407,
      "grad_norm": 1.6073803901672363,
      "learning_rate": 1.9226393629124008e-05,
      "loss": 0.5037,
      "step": 3160
    },
    {
      "epoch": 0.4057340330218866,
      "grad_norm": 0.9294481873512268,
      "learning_rate": 1.9216913158892682e-05,
      "loss": 0.4836,
      "step": 3170
    },
    {
      "epoch": 0.4070139511071291,
      "grad_norm": 1.1646478176116943,
      "learning_rate": 1.9207432688661357e-05,
      "loss": 0.5189,
      "step": 3180
    },
    {
      "epoch": 0.4082938691923717,
      "grad_norm": 0.8620973825454712,
      "learning_rate": 1.9197952218430034e-05,
      "loss": 0.5002,
      "step": 3190
    },
    {
      "epoch": 0.40957378727761423,
      "grad_norm": 0.7730439305305481,
      "learning_rate": 1.9188471748198712e-05,
      "loss": 0.4802,
      "step": 3200
    },
    {
      "epoch": 0.41085370536285676,
      "grad_norm": 0.706767201423645,
      "learning_rate": 1.917899127796739e-05,
      "loss": 0.535,
      "step": 3210
    },
    {
      "epoch": 0.41213362344809934,
      "grad_norm": 1.1379191875457764,
      "learning_rate": 1.9169510807736064e-05,
      "loss": 0.5186,
      "step": 3220
    },
    {
      "epoch": 0.41341354153334187,
      "grad_norm": 0.6369794011116028,
      "learning_rate": 1.916003033750474e-05,
      "loss": 0.4896,
      "step": 3230
    },
    {
      "epoch": 0.4146934596185844,
      "grad_norm": 3.0110867023468018,
      "learning_rate": 1.915054986727342e-05,
      "loss": 0.5275,
      "step": 3240
    },
    {
      "epoch": 0.415973377703827,
      "grad_norm": 1.3241641521453857,
      "learning_rate": 1.9141069397042097e-05,
      "loss": 0.5078,
      "step": 3250
    },
    {
      "epoch": 0.4172532957890695,
      "grad_norm": 0.942676305770874,
      "learning_rate": 1.913158892681077e-05,
      "loss": 0.4901,
      "step": 3260
    },
    {
      "epoch": 0.418533213874312,
      "grad_norm": 2.2048139572143555,
      "learning_rate": 1.9122108456579446e-05,
      "loss": 0.5272,
      "step": 3270
    },
    {
      "epoch": 0.4198131319595546,
      "grad_norm": 1.8954252004623413,
      "learning_rate": 1.9112627986348123e-05,
      "loss": 0.4853,
      "step": 3280
    },
    {
      "epoch": 0.42109305004479713,
      "grad_norm": 0.8221827745437622,
      "learning_rate": 1.91031475161168e-05,
      "loss": 0.4919,
      "step": 3290
    },
    {
      "epoch": 0.42237296813003966,
      "grad_norm": 0.742501437664032,
      "learning_rate": 1.909366704588548e-05,
      "loss": 0.5139,
      "step": 3300
    },
    {
      "epoch": 0.42365288621528224,
      "grad_norm": 1.9924613237380981,
      "learning_rate": 1.9084186575654153e-05,
      "loss": 0.4803,
      "step": 3310
    },
    {
      "epoch": 0.42493280430052477,
      "grad_norm": 0.9771905541419983,
      "learning_rate": 1.907470610542283e-05,
      "loss": 0.475,
      "step": 3320
    },
    {
      "epoch": 0.4262127223857673,
      "grad_norm": 1.5012576580047607,
      "learning_rate": 1.9065225635191508e-05,
      "loss": 0.4941,
      "step": 3330
    },
    {
      "epoch": 0.4274926404710099,
      "grad_norm": 0.6949377059936523,
      "learning_rate": 1.9055745164960186e-05,
      "loss": 0.512,
      "step": 3340
    },
    {
      "epoch": 0.4287725585562524,
      "grad_norm": 0.9960921406745911,
      "learning_rate": 1.904626469472886e-05,
      "loss": 0.4825,
      "step": 3350
    },
    {
      "epoch": 0.43005247664149493,
      "grad_norm": 1.0516977310180664,
      "learning_rate": 1.9036784224497538e-05,
      "loss": 0.5117,
      "step": 3360
    },
    {
      "epoch": 0.4313323947267375,
      "grad_norm": 0.9433191418647766,
      "learning_rate": 1.9027303754266212e-05,
      "loss": 0.484,
      "step": 3370
    },
    {
      "epoch": 0.43261231281198004,
      "grad_norm": 2.7105135917663574,
      "learning_rate": 1.901782328403489e-05,
      "loss": 0.4986,
      "step": 3380
    },
    {
      "epoch": 0.43389223089722256,
      "grad_norm": 0.8293517827987671,
      "learning_rate": 1.9008342813803567e-05,
      "loss": 0.4797,
      "step": 3390
    },
    {
      "epoch": 0.43517214898246515,
      "grad_norm": 1.3104497194290161,
      "learning_rate": 1.899886234357224e-05,
      "loss": 0.4971,
      "step": 3400
    },
    {
      "epoch": 0.43645206706770767,
      "grad_norm": 2.3978216648101807,
      "learning_rate": 1.898938187334092e-05,
      "loss": 0.5012,
      "step": 3410
    },
    {
      "epoch": 0.4377319851529502,
      "grad_norm": 0.838647186756134,
      "learning_rate": 1.8979901403109594e-05,
      "loss": 0.5042,
      "step": 3420
    },
    {
      "epoch": 0.4390119032381928,
      "grad_norm": 0.7617325782775879,
      "learning_rate": 1.897042093287827e-05,
      "loss": 0.5094,
      "step": 3430
    },
    {
      "epoch": 0.4402918213234353,
      "grad_norm": 0.8923324942588806,
      "learning_rate": 1.896094046264695e-05,
      "loss": 0.487,
      "step": 3440
    },
    {
      "epoch": 0.44157173940867783,
      "grad_norm": 1.7616841793060303,
      "learning_rate": 1.8951459992415627e-05,
      "loss": 0.4929,
      "step": 3450
    },
    {
      "epoch": 0.4428516574939204,
      "grad_norm": 0.7590444087982178,
      "learning_rate": 1.89419795221843e-05,
      "loss": 0.4844,
      "step": 3460
    },
    {
      "epoch": 0.44413157557916294,
      "grad_norm": 0.8714098930358887,
      "learning_rate": 1.893249905195298e-05,
      "loss": 0.4736,
      "step": 3470
    },
    {
      "epoch": 0.44541149366440547,
      "grad_norm": 1.3062705993652344,
      "learning_rate": 1.8923018581721656e-05,
      "loss": 0.501,
      "step": 3480
    },
    {
      "epoch": 0.44669141174964805,
      "grad_norm": 1.3274645805358887,
      "learning_rate": 1.891353811149033e-05,
      "loss": 0.506,
      "step": 3490
    },
    {
      "epoch": 0.4479713298348906,
      "grad_norm": 3.321751832962036,
      "learning_rate": 1.8904057641259008e-05,
      "loss": 0.5244,
      "step": 3500
    },
    {
      "epoch": 0.4492512479201331,
      "grad_norm": 0.8489649891853333,
      "learning_rate": 1.8894577171027683e-05,
      "loss": 0.5003,
      "step": 3510
    },
    {
      "epoch": 0.4505311660053757,
      "grad_norm": 1.0105081796646118,
      "learning_rate": 1.888509670079636e-05,
      "loss": 0.4748,
      "step": 3520
    },
    {
      "epoch": 0.4518110840906182,
      "grad_norm": 1.452003836631775,
      "learning_rate": 1.8875616230565038e-05,
      "loss": 0.4842,
      "step": 3530
    },
    {
      "epoch": 0.45309100217586074,
      "grad_norm": 0.8805971741676331,
      "learning_rate": 1.8866135760333716e-05,
      "loss": 0.4923,
      "step": 3540
    },
    {
      "epoch": 0.4543709202611033,
      "grad_norm": 2.563197135925293,
      "learning_rate": 1.885665529010239e-05,
      "loss": 0.5021,
      "step": 3550
    },
    {
      "epoch": 0.45565083834634584,
      "grad_norm": 0.7883086204528809,
      "learning_rate": 1.8847174819871067e-05,
      "loss": 0.4891,
      "step": 3560
    },
    {
      "epoch": 0.45693075643158837,
      "grad_norm": 1.5375005006790161,
      "learning_rate": 1.8837694349639745e-05,
      "loss": 0.486,
      "step": 3570
    },
    {
      "epoch": 0.4582106745168309,
      "grad_norm": 1.0281457901000977,
      "learning_rate": 1.8828213879408423e-05,
      "loss": 0.5057,
      "step": 3580
    },
    {
      "epoch": 0.4594905926020735,
      "grad_norm": 1.281043291091919,
      "learning_rate": 1.8818733409177097e-05,
      "loss": 0.4998,
      "step": 3590
    },
    {
      "epoch": 0.460770510687316,
      "grad_norm": 0.9816468954086304,
      "learning_rate": 1.880925293894577e-05,
      "loss": 0.4944,
      "step": 3600
    },
    {
      "epoch": 0.46205042877255853,
      "grad_norm": 0.9405584335327148,
      "learning_rate": 1.879977246871445e-05,
      "loss": 0.5036,
      "step": 3610
    },
    {
      "epoch": 0.4633303468578011,
      "grad_norm": 1.3431020975112915,
      "learning_rate": 1.8790291998483127e-05,
      "loss": 0.5004,
      "step": 3620
    },
    {
      "epoch": 0.46461026494304364,
      "grad_norm": 1.1362318992614746,
      "learning_rate": 1.8780811528251804e-05,
      "loss": 0.494,
      "step": 3630
    },
    {
      "epoch": 0.46589018302828616,
      "grad_norm": 1.2532789707183838,
      "learning_rate": 1.877133105802048e-05,
      "loss": 0.4748,
      "step": 3640
    },
    {
      "epoch": 0.46717010111352875,
      "grad_norm": 0.729250431060791,
      "learning_rate": 1.8761850587789156e-05,
      "loss": 0.5086,
      "step": 3650
    },
    {
      "epoch": 0.4684500191987713,
      "grad_norm": 0.6976643800735474,
      "learning_rate": 1.875237011755783e-05,
      "loss": 0.4781,
      "step": 3660
    },
    {
      "epoch": 0.4697299372840138,
      "grad_norm": 0.9878288507461548,
      "learning_rate": 1.8742889647326508e-05,
      "loss": 0.4809,
      "step": 3670
    },
    {
      "epoch": 0.4710098553692564,
      "grad_norm": 0.9907483458518982,
      "learning_rate": 1.8733409177095186e-05,
      "loss": 0.4914,
      "step": 3680
    },
    {
      "epoch": 0.4722897734544989,
      "grad_norm": 1.046175479888916,
      "learning_rate": 1.872392870686386e-05,
      "loss": 0.5048,
      "step": 3690
    },
    {
      "epoch": 0.47356969153974143,
      "grad_norm": 1.8168991804122925,
      "learning_rate": 1.8714448236632538e-05,
      "loss": 0.4528,
      "step": 3700
    },
    {
      "epoch": 0.474849609624984,
      "grad_norm": 2.9127612113952637,
      "learning_rate": 1.8704967766401216e-05,
      "loss": 0.5226,
      "step": 3710
    },
    {
      "epoch": 0.47612952771022654,
      "grad_norm": 1.4959512948989868,
      "learning_rate": 1.8695487296169893e-05,
      "loss": 0.4727,
      "step": 3720
    },
    {
      "epoch": 0.47740944579546907,
      "grad_norm": 0.6817557215690613,
      "learning_rate": 1.8686006825938568e-05,
      "loss": 0.468,
      "step": 3730
    },
    {
      "epoch": 0.47868936388071165,
      "grad_norm": 1.5274559259414673,
      "learning_rate": 1.8676526355707245e-05,
      "loss": 0.5158,
      "step": 3740
    },
    {
      "epoch": 0.4799692819659542,
      "grad_norm": 1.8979755640029907,
      "learning_rate": 1.866704588547592e-05,
      "loss": 0.4628,
      "step": 3750
    },
    {
      "epoch": 0.4812492000511967,
      "grad_norm": 0.7936620712280273,
      "learning_rate": 1.8657565415244597e-05,
      "loss": 0.5159,
      "step": 3760
    },
    {
      "epoch": 0.4825291181364393,
      "grad_norm": 1.2421464920043945,
      "learning_rate": 1.8648084945013275e-05,
      "loss": 0.495,
      "step": 3770
    },
    {
      "epoch": 0.4838090362216818,
      "grad_norm": 0.8110264539718628,
      "learning_rate": 1.8638604474781953e-05,
      "loss": 0.514,
      "step": 3780
    },
    {
      "epoch": 0.48508895430692434,
      "grad_norm": 1.168480634689331,
      "learning_rate": 1.8629124004550627e-05,
      "loss": 0.4842,
      "step": 3790
    },
    {
      "epoch": 0.4863688723921669,
      "grad_norm": 1.4188495874404907,
      "learning_rate": 1.86196435343193e-05,
      "loss": 0.4855,
      "step": 3800
    },
    {
      "epoch": 0.48764879047740944,
      "grad_norm": 1.9594807624816895,
      "learning_rate": 1.8610163064087982e-05,
      "loss": 0.486,
      "step": 3810
    },
    {
      "epoch": 0.48892870856265197,
      "grad_norm": 0.8144770264625549,
      "learning_rate": 1.8600682593856656e-05,
      "loss": 0.4939,
      "step": 3820
    },
    {
      "epoch": 0.49020862664789455,
      "grad_norm": 1.4995648860931396,
      "learning_rate": 1.8591202123625334e-05,
      "loss": 0.4912,
      "step": 3830
    },
    {
      "epoch": 0.4914885447331371,
      "grad_norm": 1.1382790803909302,
      "learning_rate": 1.858172165339401e-05,
      "loss": 0.4634,
      "step": 3840
    },
    {
      "epoch": 0.4927684628183796,
      "grad_norm": 0.8508588075637817,
      "learning_rate": 1.8572241183162686e-05,
      "loss": 0.5236,
      "step": 3850
    },
    {
      "epoch": 0.4940483809036222,
      "grad_norm": 1.2297135591506958,
      "learning_rate": 1.8562760712931364e-05,
      "loss": 0.4806,
      "step": 3860
    },
    {
      "epoch": 0.4953282989888647,
      "grad_norm": 0.863645076751709,
      "learning_rate": 1.855328024270004e-05,
      "loss": 0.4713,
      "step": 3870
    },
    {
      "epoch": 0.49660821707410724,
      "grad_norm": 1.6313464641571045,
      "learning_rate": 1.8543799772468716e-05,
      "loss": 0.4734,
      "step": 3880
    },
    {
      "epoch": 0.4978881351593498,
      "grad_norm": 3.317040205001831,
      "learning_rate": 1.8534319302237393e-05,
      "loss": 0.495,
      "step": 3890
    },
    {
      "epoch": 0.49916805324459235,
      "grad_norm": 0.8402191996574402,
      "learning_rate": 1.8524838832006068e-05,
      "loss": 0.4916,
      "step": 3900
    },
    {
      "epoch": 0.5004479713298349,
      "grad_norm": 2.1389431953430176,
      "learning_rate": 1.8515358361774745e-05,
      "loss": 0.4868,
      "step": 3910
    },
    {
      "epoch": 0.5017278894150774,
      "grad_norm": 1.6004793643951416,
      "learning_rate": 1.8505877891543423e-05,
      "loss": 0.5038,
      "step": 3920
    },
    {
      "epoch": 0.5030078075003199,
      "grad_norm": 1.4133943319320679,
      "learning_rate": 1.8496397421312097e-05,
      "loss": 0.5067,
      "step": 3930
    },
    {
      "epoch": 0.5042877255855626,
      "grad_norm": 2.345825433731079,
      "learning_rate": 1.8486916951080775e-05,
      "loss": 0.4888,
      "step": 3940
    },
    {
      "epoch": 0.5055676436708051,
      "grad_norm": 0.8149171471595764,
      "learning_rate": 1.8477436480849453e-05,
      "loss": 0.4847,
      "step": 3950
    },
    {
      "epoch": 0.5068475617560476,
      "grad_norm": 0.8076463937759399,
      "learning_rate": 1.846795601061813e-05,
      "loss": 0.4786,
      "step": 3960
    },
    {
      "epoch": 0.5081274798412901,
      "grad_norm": 0.7464845180511475,
      "learning_rate": 1.8458475540386805e-05,
      "loss": 0.5039,
      "step": 3970
    },
    {
      "epoch": 0.5094073979265327,
      "grad_norm": 0.9748256802558899,
      "learning_rate": 1.8448995070155482e-05,
      "loss": 0.4732,
      "step": 3980
    },
    {
      "epoch": 0.5106873160117752,
      "grad_norm": 1.2856113910675049,
      "learning_rate": 1.8439514599924156e-05,
      "loss": 0.4914,
      "step": 3990
    },
    {
      "epoch": 0.5119672340970178,
      "grad_norm": 0.9492047429084778,
      "learning_rate": 1.8430034129692834e-05,
      "loss": 0.5132,
      "step": 4000
    },
    {
      "epoch": 0.5119672340970178,
      "eval_accuracy": 0.79022,
      "eval_loss": 0.44629669189453125,
      "eval_macro_f1": 0.7899137546751205,
      "eval_precision": 0.7975042104934613,
      "eval_recall": 0.7936668066341269,
      "eval_runtime": 705.8506,
      "eval_samples_per_second": 141.673,
      "eval_steps_per_second": 1.108,
      "step": 4000
    }
  ],
  "logging_steps": 10,
  "max_steps": 23439,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 200,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 6.735643017216e+16,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}