{
  "best_metric": 0.3373493975903614,
  "best_model_checkpoint": "videomae-base-finetuned-numbers-augmented2/checkpoint-2816",
  "epoch": 3.2496448863636362,
  "eval_steps": 500,
  "global_step": 2816,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0035511363636363635,
      "grad_norm": 11.042091369628906,
      "learning_rate": 1.7730496453900713e-07,
      "loss": 2.6254,
      "step": 10
    },
    {
      "epoch": 0.007102272727272727,
      "grad_norm": 15.068171501159668,
      "learning_rate": 3.5460992907801425e-07,
      "loss": 2.6399,
      "step": 20
    },
    {
      "epoch": 0.01065340909090909,
      "grad_norm": 12.725520133972168,
      "learning_rate": 5.319148936170213e-07,
      "loss": 2.5084,
      "step": 30
    },
    {
      "epoch": 0.014204545454545454,
      "grad_norm": 12.825174331665039,
      "learning_rate": 7.092198581560285e-07,
      "loss": 2.5423,
      "step": 40
    },
    {
      "epoch": 0.01775568181818182,
      "grad_norm": 19.975149154663086,
      "learning_rate": 8.865248226950356e-07,
      "loss": 2.6042,
      "step": 50
    },
    {
      "epoch": 0.02130681818181818,
      "grad_norm": 13.08024787902832,
      "learning_rate": 1.0638297872340427e-06,
      "loss": 2.4857,
      "step": 60
    },
    {
      "epoch": 0.024857954545454544,
      "grad_norm": 13.69168758392334,
      "learning_rate": 1.2411347517730497e-06,
      "loss": 2.5421,
      "step": 70
    },
    {
      "epoch": 0.028409090909090908,
      "grad_norm": 21.916378021240234,
      "learning_rate": 1.418439716312057e-06,
      "loss": 2.5581,
      "step": 80
    },
    {
      "epoch": 0.03196022727272727,
      "grad_norm": 44.70994567871094,
      "learning_rate": 1.595744680851064e-06,
      "loss": 2.5354,
      "step": 90
    },
    {
      "epoch": 0.03551136363636364,
      "grad_norm": 16.238237380981445,
      "learning_rate": 1.7730496453900712e-06,
      "loss": 2.4371,
      "step": 100
    },
    {
      "epoch": 0.0390625,
      "grad_norm": 11.415323257446289,
      "learning_rate": 1.9503546099290782e-06,
      "loss": 2.4871,
      "step": 110
    },
    {
      "epoch": 0.04261363636363636,
      "grad_norm": 8.98475456237793,
      "learning_rate": 2.1276595744680853e-06,
      "loss": 2.3905,
      "step": 120
    },
    {
      "epoch": 0.04616477272727273,
      "grad_norm": 11.144891738891602,
      "learning_rate": 2.3049645390070924e-06,
      "loss": 2.4819,
      "step": 130
    },
    {
      "epoch": 0.04971590909090909,
      "grad_norm": 15.341207504272461,
      "learning_rate": 2.4822695035460995e-06,
      "loss": 2.4592,
      "step": 140
    },
    {
      "epoch": 0.053267045454545456,
      "grad_norm": 12.760394096374512,
      "learning_rate": 2.6595744680851065e-06,
      "loss": 2.4855,
      "step": 150
    },
    {
      "epoch": 0.056818181818181816,
      "grad_norm": 13.143548965454102,
      "learning_rate": 2.836879432624114e-06,
      "loss": 2.4556,
      "step": 160
    },
    {
      "epoch": 0.060369318181818184,
      "grad_norm": 11.302803993225098,
      "learning_rate": 3.0141843971631207e-06,
      "loss": 2.4462,
      "step": 170
    },
    {
      "epoch": 0.06392045454545454,
      "grad_norm": 11.462811470031738,
      "learning_rate": 3.191489361702128e-06,
      "loss": 2.3793,
      "step": 180
    },
    {
      "epoch": 0.06747159090909091,
      "grad_norm": 14.533432960510254,
      "learning_rate": 3.368794326241135e-06,
      "loss": 2.3991,
      "step": 190
    },
    {
      "epoch": 0.07102272727272728,
      "grad_norm": 16.54037094116211,
      "learning_rate": 3.5460992907801423e-06,
      "loss": 2.4178,
      "step": 200
    },
    {
      "epoch": 0.07457386363636363,
      "grad_norm": 25.168344497680664,
      "learning_rate": 3.723404255319149e-06,
      "loss": 2.3658,
      "step": 210
    },
    {
      "epoch": 0.078125,
      "grad_norm": 14.53443431854248,
      "learning_rate": 3.9007092198581565e-06,
      "loss": 2.4974,
      "step": 220
    },
    {
      "epoch": 0.08167613636363637,
      "grad_norm": 13.964099884033203,
      "learning_rate": 4.078014184397163e-06,
      "loss": 2.3467,
      "step": 230
    },
    {
      "epoch": 0.08522727272727272,
      "grad_norm": 14.819562911987305,
      "learning_rate": 4.255319148936171e-06,
      "loss": 2.3952,
      "step": 240
    },
    {
      "epoch": 0.08877840909090909,
      "grad_norm": 9.89671802520752,
      "learning_rate": 4.432624113475177e-06,
      "loss": 2.4154,
      "step": 250
    },
    {
      "epoch": 0.09232954545454546,
      "grad_norm": 16.636743545532227,
      "learning_rate": 4.609929078014185e-06,
      "loss": 2.4225,
      "step": 260
    },
    {
      "epoch": 0.09588068181818182,
      "grad_norm": 12.833456993103027,
      "learning_rate": 4.787234042553192e-06,
      "loss": 2.4442,
      "step": 270
    },
    {
      "epoch": 0.09943181818181818,
      "grad_norm": 11.386968612670898,
      "learning_rate": 4.964539007092199e-06,
      "loss": 2.3345,
      "step": 280
    },
    {
      "epoch": 0.10298295454545454,
      "grad_norm": 15.00284481048584,
      "learning_rate": 4.984214680347277e-06,
      "loss": 2.3783,
      "step": 290
    },
    {
      "epoch": 0.10653409090909091,
      "grad_norm": 12.9210786819458,
      "learning_rate": 4.964483030781374e-06,
      "loss": 2.3678,
      "step": 300
    },
    {
      "epoch": 0.11008522727272728,
      "grad_norm": 46.81148910522461,
      "learning_rate": 4.94475138121547e-06,
      "loss": 2.3202,
      "step": 310
    },
    {
      "epoch": 0.11363636363636363,
      "grad_norm": 11.623534202575684,
      "learning_rate": 4.925019731649567e-06,
      "loss": 2.4116,
      "step": 320
    },
    {
      "epoch": 0.1171875,
      "grad_norm": 12.292633056640625,
      "learning_rate": 4.905288082083662e-06,
      "loss": 2.3704,
      "step": 330
    },
    {
      "epoch": 0.12073863636363637,
      "grad_norm": 13.08430004119873,
      "learning_rate": 4.885556432517759e-06,
      "loss": 2.351,
      "step": 340
    },
    {
      "epoch": 0.12428977272727272,
      "grad_norm": 63.07197570800781,
      "learning_rate": 4.865824782951855e-06,
      "loss": 2.3944,
      "step": 350
    },
    {
      "epoch": 0.1278409090909091,
      "grad_norm": 20.782880783081055,
      "learning_rate": 4.846093133385951e-06,
      "loss": 2.3507,
      "step": 360
    },
    {
      "epoch": 0.13139204545454544,
      "grad_norm": 11.26919174194336,
      "learning_rate": 4.8263614838200475e-06,
      "loss": 2.316,
      "step": 370
    },
    {
      "epoch": 0.13494318181818182,
      "grad_norm": 13.572308540344238,
      "learning_rate": 4.806629834254144e-06,
      "loss": 2.3005,
      "step": 380
    },
    {
      "epoch": 0.13849431818181818,
      "grad_norm": 17.582345962524414,
      "learning_rate": 4.78689818468824e-06,
      "loss": 2.2751,
      "step": 390
    },
    {
      "epoch": 0.14204545454545456,
      "grad_norm": 12.714279174804688,
      "learning_rate": 4.767166535122337e-06,
      "loss": 2.3287,
      "step": 400
    },
    {
      "epoch": 0.1455965909090909,
      "grad_norm": 11.136545181274414,
      "learning_rate": 4.747434885556433e-06,
      "loss": 2.3605,
      "step": 410
    },
    {
      "epoch": 0.14914772727272727,
      "grad_norm": 26.191139221191406,
      "learning_rate": 4.727703235990529e-06,
      "loss": 2.3669,
      "step": 420
    },
    {
      "epoch": 0.15269886363636365,
      "grad_norm": 15.782291412353516,
      "learning_rate": 4.707971586424626e-06,
      "loss": 2.3671,
      "step": 430
    },
    {
      "epoch": 0.15625,
      "grad_norm": 13.54359245300293,
      "learning_rate": 4.6882399368587215e-06,
      "loss": 2.3991,
      "step": 440
    },
    {
      "epoch": 0.15980113636363635,
      "grad_norm": 12.135380744934082,
      "learning_rate": 4.668508287292818e-06,
      "loss": 2.3314,
      "step": 450
    },
    {
      "epoch": 0.16335227272727273,
      "grad_norm": 12.856097221374512,
      "learning_rate": 4.6487766377269145e-06,
      "loss": 2.2192,
      "step": 460
    },
    {
      "epoch": 0.1669034090909091,
      "grad_norm": 15.03661823272705,
      "learning_rate": 4.62904498816101e-06,
      "loss": 2.3,
      "step": 470
    },
    {
      "epoch": 0.17045454545454544,
      "grad_norm": 11.673720359802246,
      "learning_rate": 4.609313338595107e-06,
      "loss": 2.2383,
      "step": 480
    },
    {
      "epoch": 0.17400568181818182,
      "grad_norm": 14.064128875732422,
      "learning_rate": 4.589581689029203e-06,
      "loss": 2.2481,
      "step": 490
    },
    {
      "epoch": 0.17755681818181818,
      "grad_norm": 15.709892272949219,
      "learning_rate": 4.5698500394633e-06,
      "loss": 2.3622,
      "step": 500
    },
    {
      "epoch": 0.18110795454545456,
      "grad_norm": 14.75145149230957,
      "learning_rate": 4.550118389897396e-06,
      "loss": 2.2797,
      "step": 510
    },
    {
      "epoch": 0.1846590909090909,
      "grad_norm": 12.432939529418945,
      "learning_rate": 4.530386740331492e-06,
      "loss": 2.2493,
      "step": 520
    },
    {
      "epoch": 0.18821022727272727,
      "grad_norm": 13.656457901000977,
      "learning_rate": 4.510655090765588e-06,
      "loss": 2.2605,
      "step": 530
    },
    {
      "epoch": 0.19176136363636365,
      "grad_norm": 16.014392852783203,
      "learning_rate": 4.490923441199685e-06,
      "loss": 2.2669,
      "step": 540
    },
    {
      "epoch": 0.1953125,
      "grad_norm": 17.114728927612305,
      "learning_rate": 4.4711917916337814e-06,
      "loss": 2.3496,
      "step": 550
    },
    {
      "epoch": 0.19886363636363635,
      "grad_norm": 13.276822090148926,
      "learning_rate": 4.451460142067877e-06,
      "loss": 2.1851,
      "step": 560
    },
    {
      "epoch": 0.20241477272727273,
      "grad_norm": 14.958836555480957,
      "learning_rate": 4.431728492501974e-06,
      "loss": 2.2624,
      "step": 570
    },
    {
      "epoch": 0.2059659090909091,
      "grad_norm": 17.03455924987793,
      "learning_rate": 4.41199684293607e-06,
      "loss": 2.1772,
      "step": 580
    },
    {
      "epoch": 0.20951704545454544,
      "grad_norm": 14.998839378356934,
      "learning_rate": 4.392265193370166e-06,
      "loss": 2.2558,
      "step": 590
    },
    {
      "epoch": 0.21306818181818182,
      "grad_norm": 16.826061248779297,
      "learning_rate": 4.372533543804262e-06,
      "loss": 2.2345,
      "step": 600
    },
    {
      "epoch": 0.21661931818181818,
      "grad_norm": 12.38198471069336,
      "learning_rate": 4.352801894238359e-06,
      "loss": 2.2144,
      "step": 610
    },
    {
      "epoch": 0.22017045454545456,
      "grad_norm": 12.347530364990234,
      "learning_rate": 4.3330702446724545e-06,
      "loss": 2.1875,
      "step": 620
    },
    {
      "epoch": 0.2237215909090909,
      "grad_norm": 17.95762825012207,
      "learning_rate": 4.313338595106552e-06,
      "loss": 2.1382,
      "step": 630
    },
    {
      "epoch": 0.22727272727272727,
      "grad_norm": 20.329545974731445,
      "learning_rate": 4.2936069455406475e-06,
      "loss": 2.1519,
      "step": 640
    },
    {
      "epoch": 0.23082386363636365,
      "grad_norm": 14.157989501953125,
      "learning_rate": 4.273875295974744e-06,
      "loss": 2.1994,
      "step": 650
    },
    {
      "epoch": 0.234375,
      "grad_norm": 17.65989112854004,
      "learning_rate": 4.2541436464088406e-06,
      "loss": 2.2021,
      "step": 660
    },
    {
      "epoch": 0.23792613636363635,
      "grad_norm": 19.491857528686523,
      "learning_rate": 4.234411996842936e-06,
      "loss": 2.25,
      "step": 670
    },
    {
      "epoch": 0.24147727272727273,
      "grad_norm": 14.291483879089355,
      "learning_rate": 4.214680347277033e-06,
      "loss": 2.1732,
      "step": 680
    },
    {
      "epoch": 0.2450284090909091,
      "grad_norm": 15.794211387634277,
      "learning_rate": 4.194948697711129e-06,
      "loss": 2.2288,
      "step": 690
    },
    {
      "epoch": 0.24857954545454544,
      "grad_norm": 13.537517547607422,
      "learning_rate": 4.175217048145225e-06,
      "loss": 2.2279,
      "step": 700
    },
    {
      "epoch": 0.25035511363636365,
      "eval_accuracy": 0.18244406196213425,
      "eval_f1": 0.1261780590834414,
      "eval_loss": 2.2644505500793457,
      "eval_precision": 0.25592730406989506,
      "eval_recall": 0.17920384277210025,
      "eval_runtime": 691.0928,
      "eval_samples_per_second": 0.841,
      "eval_steps_per_second": 0.211,
      "step": 705
    },
    {
      "epoch": 1.0017755681818181,
      "grad_norm": 16.53750228881836,
      "learning_rate": 4.1554853985793214e-06,
      "loss": 2.2194,
      "step": 710
    },
    {
      "epoch": 1.0053267045454546,
      "grad_norm": 14.238625526428223,
      "learning_rate": 4.135753749013418e-06,
      "loss": 2.1206,
      "step": 720
    },
    {
      "epoch": 1.0088778409090908,
      "grad_norm": 19.52684783935547,
      "learning_rate": 4.1160220994475145e-06,
      "loss": 2.0227,
      "step": 730
    },
    {
      "epoch": 1.0124289772727273,
      "grad_norm": 21.326141357421875,
      "learning_rate": 4.09629044988161e-06,
      "loss": 2.2306,
      "step": 740
    },
    {
      "epoch": 1.0159801136363635,
      "grad_norm": 18.599313735961914,
      "learning_rate": 4.076558800315707e-06,
      "loss": 2.0473,
      "step": 750
    },
    {
      "epoch": 1.01953125,
      "grad_norm": 21.073596954345703,
      "learning_rate": 4.056827150749803e-06,
      "loss": 2.0669,
      "step": 760
    },
    {
      "epoch": 1.0230823863636365,
      "grad_norm": 13.940530776977539,
      "learning_rate": 4.037095501183899e-06,
      "loss": 2.0881,
      "step": 770
    },
    {
      "epoch": 1.0266335227272727,
      "grad_norm": 23.56789207458496,
      "learning_rate": 4.017363851617996e-06,
      "loss": 2.0495,
      "step": 780
    },
    {
      "epoch": 1.0301846590909092,
      "grad_norm": 24.358179092407227,
      "learning_rate": 3.997632202052092e-06,
      "loss": 2.1131,
      "step": 790
    },
    {
      "epoch": 1.0337357954545454,
      "grad_norm": 23.59551239013672,
      "learning_rate": 3.977900552486188e-06,
      "loss": 2.2194,
      "step": 800
    },
    {
      "epoch": 1.0372869318181819,
      "grad_norm": 18.300479888916016,
      "learning_rate": 3.958168902920285e-06,
      "loss": 2.0495,
      "step": 810
    },
    {
      "epoch": 1.0408380681818181,
      "grad_norm": 14.824146270751953,
      "learning_rate": 3.9384372533543806e-06,
      "loss": 2.0368,
      "step": 820
    },
    {
      "epoch": 1.0443892045454546,
      "grad_norm": 30.449533462524414,
      "learning_rate": 3.918705603788477e-06,
      "loss": 1.9466,
      "step": 830
    },
    {
      "epoch": 1.0479403409090908,
      "grad_norm": 30.030054092407227,
      "learning_rate": 3.898973954222574e-06,
      "loss": 2.037,
      "step": 840
    },
    {
      "epoch": 1.0514914772727273,
      "grad_norm": 16.496719360351562,
      "learning_rate": 3.879242304656669e-06,
      "loss": 2.0618,
      "step": 850
    },
    {
      "epoch": 1.0550426136363635,
      "grad_norm": 30.52495574951172,
      "learning_rate": 3.859510655090766e-06,
      "loss": 1.9784,
      "step": 860
    },
    {
      "epoch": 1.05859375,
      "grad_norm": 16.27585792541504,
      "learning_rate": 3.839779005524862e-06,
      "loss": 2.0537,
      "step": 870
    },
    {
      "epoch": 1.0621448863636365,
      "grad_norm": 21.442113876342773,
      "learning_rate": 3.820047355958959e-06,
      "loss": 2.0197,
      "step": 880
    },
    {
      "epoch": 1.0656960227272727,
      "grad_norm": 23.343584060668945,
      "learning_rate": 3.8003157063930545e-06,
      "loss": 1.9195,
      "step": 890
    },
    {
      "epoch": 1.0692471590909092,
      "grad_norm": 12.5782470703125,
      "learning_rate": 3.7805840568271514e-06,
      "loss": 1.9142,
      "step": 900
    },
    {
      "epoch": 1.0727982954545454,
      "grad_norm": 14.708935737609863,
      "learning_rate": 3.7608524072612475e-06,
      "loss": 1.958,
      "step": 910
    },
    {
      "epoch": 1.0763494318181819,
      "grad_norm": 14.57020378112793,
      "learning_rate": 3.7411207576953436e-06,
      "loss": 1.9806,
      "step": 920
    },
    {
      "epoch": 1.0799005681818181,
      "grad_norm": 12.808838844299316,
      "learning_rate": 3.72138910812944e-06,
      "loss": 2.0507,
      "step": 930
    },
    {
      "epoch": 1.0834517045454546,
      "grad_norm": 20.186933517456055,
      "learning_rate": 3.7016574585635362e-06,
      "loss": 2.1607,
      "step": 940
    },
    {
      "epoch": 1.0870028409090908,
      "grad_norm": 17.26938819885254,
      "learning_rate": 3.6819258089976323e-06,
      "loss": 1.9388,
      "step": 950
    },
    {
      "epoch": 1.0905539772727273,
      "grad_norm": 18.56201934814453,
      "learning_rate": 3.662194159431729e-06,
      "loss": 1.984,
      "step": 960
    },
    {
      "epoch": 1.0941051136363635,
      "grad_norm": 13.663825035095215,
      "learning_rate": 3.642462509865825e-06,
      "loss": 1.9401,
      "step": 970
    },
    {
      "epoch": 1.09765625,
      "grad_norm": 18.881681442260742,
      "learning_rate": 3.622730860299921e-06,
      "loss": 1.914,
      "step": 980
    },
    {
      "epoch": 1.1012073863636365,
      "grad_norm": 25.957923889160156,
      "learning_rate": 3.602999210734018e-06,
      "loss": 1.9745,
      "step": 990
    },
    {
      "epoch": 1.1047585227272727,
      "grad_norm": 33.24240493774414,
      "learning_rate": 3.583267561168114e-06,
      "loss": 1.9489,
      "step": 1000
    },
    {
      "epoch": 1.1083096590909092,
      "grad_norm": 16.282686233520508,
      "learning_rate": 3.56353591160221e-06,
      "loss": 1.9327,
      "step": 1010
    },
    {
      "epoch": 1.1118607954545454,
      "grad_norm": 27.778982162475586,
      "learning_rate": 3.5438042620363066e-06,
      "loss": 1.801,
      "step": 1020
    },
    {
      "epoch": 1.1154119318181819,
      "grad_norm": 20.463930130004883,
      "learning_rate": 3.5240726124704027e-06,
      "loss": 1.9934,
      "step": 1030
    },
    {
      "epoch": 1.1189630681818181,
      "grad_norm": 26.343746185302734,
      "learning_rate": 3.504340962904499e-06,
      "loss": 1.915,
      "step": 1040
    },
    {
      "epoch": 1.1225142045454546,
      "grad_norm": 21.86429214477539,
      "learning_rate": 3.4846093133385953e-06,
      "loss": 1.9066,
      "step": 1050
    },
    {
      "epoch": 1.1260653409090908,
      "grad_norm": 21.5794620513916,
      "learning_rate": 3.4648776637726914e-06,
      "loss": 1.66,
      "step": 1060
    },
    {
      "epoch": 1.1296164772727273,
      "grad_norm": 14.89384937286377,
      "learning_rate": 3.4451460142067875e-06,
      "loss": 1.7305,
      "step": 1070
    },
    {
      "epoch": 1.1331676136363638,
      "grad_norm": 19.86824607849121,
      "learning_rate": 3.4254143646408845e-06,
      "loss": 2.0088,
      "step": 1080
    },
    {
      "epoch": 1.13671875,
      "grad_norm": 19.656740188598633,
      "learning_rate": 3.4056827150749806e-06,
      "loss": 2.007,
      "step": 1090
    },
    {
      "epoch": 1.1402698863636362,
      "grad_norm": 19.889986038208008,
      "learning_rate": 3.3859510655090767e-06,
      "loss": 1.9091,
      "step": 1100
    },
    {
      "epoch": 1.1438210227272727,
      "grad_norm": 34.58203887939453,
      "learning_rate": 3.366219415943173e-06,
      "loss": 1.982,
      "step": 1110
    },
    {
      "epoch": 1.1473721590909092,
      "grad_norm": 14.81122875213623,
      "learning_rate": 3.3464877663772693e-06,
      "loss": 1.916,
      "step": 1120
    },
    {
      "epoch": 1.1509232954545454,
      "grad_norm": 25.33427619934082,
      "learning_rate": 3.3267561168113654e-06,
      "loss": 1.9497,
      "step": 1130
    },
    {
      "epoch": 1.1544744318181819,
      "grad_norm": 18.72292709350586,
      "learning_rate": 3.3070244672454623e-06,
      "loss": 1.7237,
      "step": 1140
    },
    {
      "epoch": 1.1580255681818181,
      "grad_norm": 20.77664566040039,
      "learning_rate": 3.2872928176795584e-06,
      "loss": 1.8997,
      "step": 1150
    },
    {
      "epoch": 1.1615767045454546,
      "grad_norm": 23.85498809814453,
      "learning_rate": 3.2675611681136545e-06,
      "loss": 1.8602,
      "step": 1160
    },
    {
      "epoch": 1.1651278409090908,
      "grad_norm": 22.836532592773438,
      "learning_rate": 3.247829518547751e-06,
      "loss": 1.8654,
      "step": 1170
    },
    {
      "epoch": 1.1686789772727273,
      "grad_norm": 35.54890441894531,
      "learning_rate": 3.228097868981847e-06,
      "loss": 1.6553,
      "step": 1180
    },
    {
      "epoch": 1.1722301136363638,
      "grad_norm": 19.193073272705078,
      "learning_rate": 3.208366219415943e-06,
      "loss": 1.8181,
      "step": 1190
    },
    {
      "epoch": 1.17578125,
      "grad_norm": 18.36795425415039,
      "learning_rate": 3.1886345698500397e-06,
      "loss": 1.796,
      "step": 1200
    },
    {
      "epoch": 1.1793323863636362,
      "grad_norm": 28.257104873657227,
      "learning_rate": 3.1689029202841358e-06,
      "loss": 1.9526,
      "step": 1210
    },
    {
      "epoch": 1.1828835227272727,
      "grad_norm": 21.073034286499023,
      "learning_rate": 3.149171270718232e-06,
      "loss": 1.8972,
      "step": 1220
    },
    {
      "epoch": 1.1864346590909092,
      "grad_norm": 30.784828186035156,
      "learning_rate": 3.129439621152329e-06,
      "loss": 1.8178,
      "step": 1230
    },
    {
      "epoch": 1.1899857954545454,
      "grad_norm": 33.99855041503906,
      "learning_rate": 3.109707971586425e-06,
      "loss": 1.7748,
      "step": 1240
    },
    {
      "epoch": 1.1935369318181819,
      "grad_norm": 19.716716766357422,
      "learning_rate": 3.089976322020521e-06,
      "loss": 1.7614,
      "step": 1250
    },
    {
      "epoch": 1.1970880681818181,
      "grad_norm": 17.041454315185547,
      "learning_rate": 3.0702446724546175e-06,
      "loss": 1.7971,
      "step": 1260
    },
    {
      "epoch": 1.2006392045454546,
      "grad_norm": 32.49284744262695,
      "learning_rate": 3.0505130228887136e-06,
      "loss": 1.8446,
      "step": 1270
    },
    {
      "epoch": 1.2041903409090908,
      "grad_norm": 18.60474395751953,
      "learning_rate": 3.03078137332281e-06,
      "loss": 1.6414,
      "step": 1280
    },
    {
      "epoch": 1.2077414772727273,
      "grad_norm": 16.46165657043457,
      "learning_rate": 3.0110497237569062e-06,
      "loss": 1.9618,
      "step": 1290
    },
    {
      "epoch": 1.2112926136363638,
      "grad_norm": 23.773658752441406,
      "learning_rate": 2.9913180741910023e-06,
      "loss": 1.69,
      "step": 1300
    },
    {
      "epoch": 1.21484375,
      "grad_norm": 26.765705108642578,
      "learning_rate": 2.9715864246250992e-06,
      "loss": 1.8167,
      "step": 1310
    },
    {
      "epoch": 1.2183948863636362,
      "grad_norm": 28.61939811706543,
      "learning_rate": 2.9518547750591953e-06,
      "loss": 1.7087,
      "step": 1320
    },
    {
      "epoch": 1.2219460227272727,
      "grad_norm": 12.1895112991333,
      "learning_rate": 2.9321231254932914e-06,
      "loss": 1.6608,
      "step": 1330
    },
    {
      "epoch": 1.2254971590909092,
      "grad_norm": 36.88252258300781,
      "learning_rate": 2.912391475927388e-06,
      "loss": 1.8907,
      "step": 1340
    },
    {
      "epoch": 1.2290482954545454,
      "grad_norm": 12.692719459533691,
      "learning_rate": 2.892659826361484e-06,
      "loss": 1.7991,
      "step": 1350
    },
    {
      "epoch": 1.2325994318181819,
      "grad_norm": 17.56617546081543,
      "learning_rate": 2.87292817679558e-06,
      "loss": 1.7126,
      "step": 1360
    },
    {
      "epoch": 1.2361505681818181,
      "grad_norm": 26.137487411499023,
      "learning_rate": 2.853196527229677e-06,
      "loss": 1.6474,
      "step": 1370
    },
    {
      "epoch": 1.2397017045454546,
      "grad_norm": 18.798303604125977,
      "learning_rate": 2.833464877663773e-06,
      "loss": 1.5864,
      "step": 1380
    },
    {
      "epoch": 1.2432528409090908,
      "grad_norm": 29.38694953918457,
      "learning_rate": 2.8137332280978693e-06,
      "loss": 1.8485,
      "step": 1390
    },
    {
      "epoch": 1.2468039772727273,
      "grad_norm": 25.83952522277832,
      "learning_rate": 2.7940015785319658e-06,
      "loss": 1.7024,
      "step": 1400
    },
    {
      "epoch": 1.25,
      "eval_accuracy": 0.31669535283993117,
      "eval_f1": 0.2827610289664265,
      "eval_loss": 2.046189308166504,
      "eval_precision": 0.3353755833663839,
      "eval_recall": 0.31518980464596447,
      "eval_runtime": 747.7036,
      "eval_samples_per_second": 0.777,
      "eval_steps_per_second": 0.195,
      "step": 1409
    },
    {
      "epoch": 2.0003551136363638,
      "grad_norm": 20.87928009033203,
      "learning_rate": 2.774269928966062e-06,
      "loss": 1.6345,
      "step": 1410
    },
    {
      "epoch": 2.00390625,
      "grad_norm": 28.964290618896484,
      "learning_rate": 2.754538279400158e-06,
      "loss": 1.7143,
      "step": 1420
    },
    {
      "epoch": 2.0074573863636362,
      "grad_norm": 17.409208297729492,
      "learning_rate": 2.7348066298342545e-06,
      "loss": 1.6937,
      "step": 1430
    },
    {
      "epoch": 2.011008522727273,
      "grad_norm": 24.774322509765625,
      "learning_rate": 2.7150749802683506e-06,
      "loss": 1.6225,
      "step": 1440
    },
    {
      "epoch": 2.014559659090909,
      "grad_norm": 33.20758819580078,
      "learning_rate": 2.6953433307024466e-06,
      "loss": 1.6123,
      "step": 1450
    },
    {
      "epoch": 2.0181107954545454,
      "grad_norm": 24.023727416992188,
      "learning_rate": 2.6756116811365436e-06,
      "loss": 1.6963,
      "step": 1460
    },
    {
      "epoch": 2.0216619318181817,
      "grad_norm": 26.691741943359375,
      "learning_rate": 2.6558800315706397e-06,
      "loss": 1.7662,
      "step": 1470
    },
    {
      "epoch": 2.0252130681818183,
      "grad_norm": 31.09050178527832,
      "learning_rate": 2.6361483820047358e-06,
      "loss": 1.6466,
      "step": 1480
    },
    {
      "epoch": 2.0287642045454546,
      "grad_norm": 34.407222747802734,
      "learning_rate": 2.6164167324388323e-06,
      "loss": 1.577,
      "step": 1490
    },
    {
      "epoch": 2.032315340909091,
      "grad_norm": 21.151485443115234,
      "learning_rate": 2.5966850828729284e-06,
      "loss": 1.4755,
      "step": 1500
    },
    {
      "epoch": 2.035866477272727,
      "grad_norm": 33.150962829589844,
      "learning_rate": 2.5769534333070245e-06,
      "loss": 1.7999,
      "step": 1510
    },
    {
      "epoch": 2.0394176136363638,
      "grad_norm": 37.68964385986328,
      "learning_rate": 2.557221783741121e-06,
      "loss": 1.7355,
      "step": 1520
    },
    {
      "epoch": 2.04296875,
      "grad_norm": 21.464839935302734,
      "learning_rate": 2.537490134175217e-06,
      "loss": 1.5964,
      "step": 1530
    },
    {
      "epoch": 2.0465198863636362,
      "grad_norm": 33.7103157043457,
      "learning_rate": 2.517758484609313e-06,
      "loss": 1.6579,
      "step": 1540
    },
    {
      "epoch": 2.050071022727273,
      "grad_norm": 34.679019927978516,
      "learning_rate": 2.4980268350434097e-06,
      "loss": 1.5309,
      "step": 1550
    },
    {
      "epoch": 2.053622159090909,
      "grad_norm": 25.11505889892578,
      "learning_rate": 2.478295185477506e-06,
      "loss": 1.605,
      "step": 1560
    },
    {
      "epoch": 2.0571732954545454,
      "grad_norm": 14.84094524383545,
      "learning_rate": 2.4585635359116027e-06,
      "loss": 1.5956,
      "step": 1570
    },
    {
      "epoch": 2.0607244318181817,
      "grad_norm": 22.252281188964844,
      "learning_rate": 2.438831886345699e-06,
      "loss": 1.5511,
      "step": 1580
    },
    {
      "epoch": 2.0642755681818183,
      "grad_norm": 50.657798767089844,
      "learning_rate": 2.419100236779795e-06,
      "loss": 1.5066,
      "step": 1590
    },
    {
      "epoch": 2.0678267045454546,
      "grad_norm": 32.89131546020508,
      "learning_rate": 2.3993685872138914e-06,
      "loss": 1.6128,
      "step": 1600
    },
    {
      "epoch": 2.071377840909091,
      "grad_norm": 32.423648834228516,
      "learning_rate": 2.3796369376479875e-06,
      "loss": 1.6446,
      "step": 1610
    },
    {
      "epoch": 2.074928977272727,
      "grad_norm": 30.093395233154297,
      "learning_rate": 2.359905288082084e-06,
      "loss": 1.5277,
      "step": 1620
    },
    {
      "epoch": 2.0784801136363638,
      "grad_norm": 40.93949508666992,
      "learning_rate": 2.34017363851618e-06,
      "loss": 1.5847,
      "step": 1630
    },
    {
      "epoch": 2.08203125,
      "grad_norm": 21.456939697265625,
      "learning_rate": 2.320441988950276e-06,
      "loss": 1.5051,
      "step": 1640
    },
    {
      "epoch": 2.0855823863636362,
      "grad_norm": 13.460980415344238,
      "learning_rate": 2.3007103393843727e-06,
      "loss": 1.6683,
      "step": 1650
    },
    {
      "epoch": 2.089133522727273,
      "grad_norm": 25.33333969116211,
      "learning_rate": 2.2809786898184692e-06,
      "loss": 1.5145,
      "step": 1660
    },
    {
      "epoch": 2.092684659090909,
      "grad_norm": 35.350223541259766,
      "learning_rate": 2.2612470402525653e-06,
      "loss": 1.5161,
      "step": 1670
    },
    {
      "epoch": 2.0962357954545454,
      "grad_norm": 11.545449256896973,
      "learning_rate": 2.2415153906866614e-06,
      "loss": 1.5556,
      "step": 1680
    },
    {
      "epoch": 2.0997869318181817,
      "grad_norm": 24.93936538696289,
      "learning_rate": 2.221783741120758e-06,
      "loss": 1.6311,
      "step": 1690
    },
    {
      "epoch": 2.1033380681818183,
      "grad_norm": 17.35971450805664,
      "learning_rate": 2.202052091554854e-06,
      "loss": 1.6324,
      "step": 1700
    },
    {
      "epoch": 2.1068892045454546,
      "grad_norm": 30.070161819458008,
      "learning_rate": 2.1823204419889505e-06,
      "loss": 1.6366,
      "step": 1710
    },
    {
      "epoch": 2.110440340909091,
      "grad_norm": 30.212112426757812,
      "learning_rate": 2.1625887924230466e-06,
      "loss": 1.5943,
      "step": 1720
    },
    {
      "epoch": 2.113991477272727,
      "grad_norm": 19.462881088256836,
      "learning_rate": 2.1428571428571427e-06,
      "loss": 1.5983,
      "step": 1730
    },
    {
      "epoch": 2.1175426136363638,
      "grad_norm": 17.48008155822754,
      "learning_rate": 2.1231254932912392e-06,
      "loss": 1.5725,
      "step": 1740
    },
    {
      "epoch": 2.12109375,
      "grad_norm": 18.313573837280273,
      "learning_rate": 2.1033938437253358e-06,
      "loss": 1.6039,
      "step": 1750
    },
    {
      "epoch": 2.1246448863636362,
      "grad_norm": 12.866426467895508,
      "learning_rate": 2.083662194159432e-06,
      "loss": 1.4405,
      "step": 1760
    },
    {
      "epoch": 2.128196022727273,
      "grad_norm": 25.331958770751953,
      "learning_rate": 2.063930544593528e-06,
      "loss": 1.728,
      "step": 1770
    },
    {
      "epoch": 2.131747159090909,
      "grad_norm": 31.10738182067871,
      "learning_rate": 2.0441988950276245e-06,
      "loss": 1.6007,
      "step": 1780
    },
    {
      "epoch": 2.1352982954545454,
      "grad_norm": 27.759248733520508,
      "learning_rate": 2.0244672454617206e-06,
      "loss": 1.5672,
      "step": 1790
    },
    {
      "epoch": 2.1388494318181817,
      "grad_norm": 16.519384384155273,
      "learning_rate": 2.004735595895817e-06,
      "loss": 1.3412,
      "step": 1800
    },
    {
      "epoch": 2.1424005681818183,
      "grad_norm": 19.062610626220703,
      "learning_rate": 1.9850039463299136e-06,
      "loss": 1.6176,
      "step": 1810
    },
    {
      "epoch": 2.1459517045454546,
      "grad_norm": 48.56722640991211,
      "learning_rate": 1.9652722967640097e-06,
      "loss": 1.6479,
      "step": 1820
    },
    {
      "epoch": 2.149502840909091,
      "grad_norm": 22.719276428222656,
      "learning_rate": 1.9455406471981058e-06,
      "loss": 1.4365,
      "step": 1830
    },
    {
      "epoch": 2.153053977272727,
      "grad_norm": 12.384623527526855,
      "learning_rate": 1.9258089976322023e-06,
      "loss": 1.4842,
      "step": 1840
    },
    {
      "epoch": 2.1566051136363638,
      "grad_norm": 29.254623413085938,
      "learning_rate": 1.9060773480662986e-06,
      "loss": 1.4222,
      "step": 1850
    },
    {
      "epoch": 2.16015625,
      "grad_norm": 32.79444885253906,
      "learning_rate": 1.8863456985003947e-06,
      "loss": 1.6447,
      "step": 1860
    },
    {
      "epoch": 2.1637073863636362,
      "grad_norm": 21.711156845092773,
      "learning_rate": 1.8666140489344912e-06,
      "loss": 1.4027,
      "step": 1870
    },
    {
      "epoch": 2.167258522727273,
      "grad_norm": 29.14606285095215,
      "learning_rate": 1.8468823993685875e-06,
      "loss": 1.5075,
      "step": 1880
    },
    {
      "epoch": 2.170809659090909,
      "grad_norm": 35.72248077392578,
      "learning_rate": 1.8271507498026836e-06,
      "loss": 1.6284,
      "step": 1890
    },
    {
      "epoch": 2.1743607954545454,
      "grad_norm": 28.522951126098633,
      "learning_rate": 1.8074191002367799e-06,
      "loss": 1.5546,
      "step": 1900
    },
    {
      "epoch": 2.1779119318181817,
      "grad_norm": 14.657508850097656,
      "learning_rate": 1.7876874506708764e-06,
      "loss": 1.5282,
      "step": 1910
    },
    {
      "epoch": 2.1814630681818183,
      "grad_norm": 23.2537899017334,
      "learning_rate": 1.7679558011049725e-06,
      "loss": 1.4651,
      "step": 1920
    },
    {
      "epoch": 2.1850142045454546,
      "grad_norm": 21.244699478149414,
      "learning_rate": 1.7482241515390688e-06,
      "loss": 1.4957,
      "step": 1930
    },
    {
      "epoch": 2.188565340909091,
      "grad_norm": 42.69889831542969,
      "learning_rate": 1.7284925019731651e-06,
      "loss": 1.6205,
      "step": 1940
    },
    {
      "epoch": 2.192116477272727,
      "grad_norm": 32.654991149902344,
      "learning_rate": 1.7087608524072614e-06,
      "loss": 1.5725,
      "step": 1950
    },
    {
      "epoch": 2.1956676136363638,
      "grad_norm": 60.657169342041016,
      "learning_rate": 1.6890292028413577e-06,
      "loss": 1.5037,
      "step": 1960
    },
    {
      "epoch": 2.19921875,
      "grad_norm": 39.7910041809082,
      "learning_rate": 1.669297553275454e-06,
      "loss": 1.4801,
      "step": 1970
    },
    {
      "epoch": 2.2027698863636362,
      "grad_norm": 28.17043113708496,
      "learning_rate": 1.6495659037095501e-06,
      "loss": 1.4794,
      "step": 1980
    },
    {
      "epoch": 2.206321022727273,
      "grad_norm": 22.981891632080078,
      "learning_rate": 1.6298342541436466e-06,
      "loss": 1.5583,
      "step": 1990
    },
    {
      "epoch": 2.209872159090909,
      "grad_norm": 24.155668258666992,
      "learning_rate": 1.610102604577743e-06,
      "loss": 1.4282,
      "step": 2000
    },
    {
      "epoch": 2.2134232954545454,
      "grad_norm": 54.06626892089844,
      "learning_rate": 1.590370955011839e-06,
      "loss": 1.5627,
      "step": 2010
    },
    {
      "epoch": 2.2169744318181817,
      "grad_norm": 22.495920181274414,
      "learning_rate": 1.5706393054459353e-06,
      "loss": 1.4818,
      "step": 2020
    },
    {
      "epoch": 2.2205255681818183,
      "grad_norm": 36.265045166015625,
      "learning_rate": 1.5509076558800318e-06,
      "loss": 1.5929,
      "step": 2030
    },
    {
      "epoch": 2.2240767045454546,
      "grad_norm": 15.528562545776367,
      "learning_rate": 1.531176006314128e-06,
      "loss": 1.4649,
      "step": 2040
    },
    {
      "epoch": 2.227627840909091,
      "grad_norm": 26.963422775268555,
      "learning_rate": 1.5114443567482242e-06,
      "loss": 1.4863,
      "step": 2050
    },
    {
      "epoch": 2.231178977272727,
      "grad_norm": 16.62386131286621,
      "learning_rate": 1.4917127071823205e-06,
      "loss": 1.2891,
      "step": 2060
    },
    {
      "epoch": 2.2347301136363638,
      "grad_norm": 22.612010955810547,
      "learning_rate": 1.4719810576164168e-06,
      "loss": 1.4939,
      "step": 2070
    },
    {
      "epoch": 2.23828125,
      "grad_norm": 59.00920104980469,
      "learning_rate": 1.4522494080505131e-06,
      "loss": 1.471,
      "step": 2080
    },
    {
      "epoch": 2.2418323863636362,
      "grad_norm": 22.010278701782227,
      "learning_rate": 1.4325177584846095e-06,
      "loss": 1.471,
      "step": 2090
    },
    {
      "epoch": 2.245383522727273,
      "grad_norm": 25.769332885742188,
      "learning_rate": 1.4127861089187055e-06,
      "loss": 1.3952,
      "step": 2100
    },
    {
      "epoch": 2.248934659090909,
      "grad_norm": 31.15215301513672,
      "learning_rate": 1.393054459352802e-06,
      "loss": 1.3164,
      "step": 2110
    },
    {
      "epoch": 2.25,
      "eval_accuracy": 0.3080895008605852,
      "eval_f1": 0.25675363952610347,
      "eval_loss": 1.9758973121643066,
      "eval_precision": 0.30216996841958826,
      "eval_recall": 0.3084819823166105,
      "eval_runtime": 716.7915,
      "eval_samples_per_second": 0.811,
      "eval_steps_per_second": 0.204,
      "step": 2113
    },
    {
      "epoch": 3.0024857954545454,
      "grad_norm": 25.143760681152344,
      "learning_rate": 1.3733228097868984e-06,
      "loss": 1.4265,
      "step": 2120
    },
    {
      "epoch": 3.0060369318181817,
      "grad_norm": 33.473609924316406,
      "learning_rate": 1.3535911602209945e-06,
      "loss": 1.3055,
      "step": 2130
    },
    {
      "epoch": 3.0095880681818183,
      "grad_norm": 15.528383255004883,
      "learning_rate": 1.3338595106550908e-06,
      "loss": 1.332,
      "step": 2140
    },
    {
      "epoch": 3.0131392045454546,
      "grad_norm": 17.708980560302734,
      "learning_rate": 1.3141278610891873e-06,
      "loss": 1.3542,
      "step": 2150
    },
    {
      "epoch": 3.016690340909091,
      "grad_norm": 90.82689666748047,
      "learning_rate": 1.2943962115232834e-06,
      "loss": 1.551,
      "step": 2160
    },
    {
      "epoch": 3.020241477272727,
      "grad_norm": 45.7540397644043,
      "learning_rate": 1.2746645619573797e-06,
      "loss": 1.4612,
      "step": 2170
    },
    {
      "epoch": 3.0237926136363638,
      "grad_norm": 16.101266860961914,
      "learning_rate": 1.254932912391476e-06,
      "loss": 1.3709,
      "step": 2180
    },
    {
      "epoch": 3.02734375,
      "grad_norm": 13.82054615020752,
      "learning_rate": 1.2352012628255723e-06,
      "loss": 1.3762,
      "step": 2190
    },
    {
      "epoch": 3.0308948863636362,
      "grad_norm": 33.30424880981445,
      "learning_rate": 1.2154696132596686e-06,
      "loss": 1.4682,
      "step": 2200
    },
    {
      "epoch": 3.034446022727273,
      "grad_norm": 30.845436096191406,
      "learning_rate": 1.1957379636937649e-06,
      "loss": 1.4023,
      "step": 2210
    },
    {
      "epoch": 3.037997159090909,
      "grad_norm": 16.949199676513672,
      "learning_rate": 1.1760063141278612e-06,
      "loss": 1.3019,
      "step": 2220
    },
    {
      "epoch": 3.0415482954545454,
      "grad_norm": 19.528390884399414,
      "learning_rate": 1.1562746645619575e-06,
      "loss": 1.4701,
      "step": 2230
    },
    {
      "epoch": 3.0450994318181817,
      "grad_norm": 53.9114990234375,
      "learning_rate": 1.1365430149960538e-06,
      "loss": 1.3315,
      "step": 2240
    },
    {
      "epoch": 3.0486505681818183,
      "grad_norm": 32.42067337036133,
      "learning_rate": 1.11681136543015e-06,
      "loss": 1.5252,
      "step": 2250
    },
    {
      "epoch": 3.0522017045454546,
      "grad_norm": 16.656024932861328,
      "learning_rate": 1.0970797158642462e-06,
      "loss": 1.5387,
      "step": 2260
    },
    {
      "epoch": 3.055752840909091,
      "grad_norm": 22.413053512573242,
      "learning_rate": 1.0773480662983427e-06,
      "loss": 1.3136,
      "step": 2270
    },
    {
      "epoch": 3.059303977272727,
      "grad_norm": 11.095012664794922,
      "learning_rate": 1.0576164167324388e-06,
      "loss": 1.4976,
      "step": 2280
    },
    {
      "epoch": 3.0628551136363638,
      "grad_norm": 18.41863250732422,
      "learning_rate": 1.0378847671665353e-06,
      "loss": 1.383,
      "step": 2290
    },
    {
      "epoch": 3.06640625,
      "grad_norm": 26.215972900390625,
      "learning_rate": 1.0181531176006316e-06,
      "loss": 1.5059,
      "step": 2300
    },
    {
      "epoch": 3.0699573863636362,
      "grad_norm": 28.589454650878906,
      "learning_rate": 9.984214680347277e-07,
      "loss": 1.4089,
      "step": 2310
    },
    {
      "epoch": 3.073508522727273,
      "grad_norm": 50.552528381347656,
      "learning_rate": 9.786898184688242e-07,
      "loss": 1.4042,
      "step": 2320
    },
    {
      "epoch": 3.077059659090909,
      "grad_norm": 13.100271224975586,
      "learning_rate": 9.589581689029203e-07,
      "loss": 1.2665,
      "step": 2330
    },
    {
      "epoch": 3.0806107954545454,
      "grad_norm": 24.274110794067383,
      "learning_rate": 9.392265193370166e-07,
      "loss": 1.3727,
      "step": 2340
    },
    {
      "epoch": 3.0841619318181817,
      "grad_norm": 37.16689682006836,
      "learning_rate": 9.19494869771113e-07,
      "loss": 1.4553,
      "step": 2350
    },
    {
      "epoch": 3.0877130681818183,
      "grad_norm": 26.41681480407715,
      "learning_rate": 8.997632202052092e-07,
      "loss": 1.4915,
      "step": 2360
    },
    {
      "epoch": 3.0912642045454546,
      "grad_norm": 39.80120849609375,
      "learning_rate": 8.800315706393054e-07,
      "loss": 1.5277,
      "step": 2370
    },
    {
      "epoch": 3.094815340909091,
      "grad_norm": 20.090126037597656,
      "learning_rate": 8.602999210734018e-07,
      "loss": 1.3832,
      "step": 2380
    },
    {
      "epoch": 3.098366477272727,
      "grad_norm": 35.57830047607422,
      "learning_rate": 8.405682715074981e-07,
      "loss": 1.358,
      "step": 2390
    },
    {
      "epoch": 3.1019176136363638,
      "grad_norm": 15.675562858581543,
      "learning_rate": 8.208366219415943e-07,
      "loss": 1.3206,
      "step": 2400
    },
    {
      "epoch": 3.10546875,
      "grad_norm": 40.23310470581055,
      "learning_rate": 8.011049723756907e-07,
      "loss": 1.3537,
      "step": 2410
    },
    {
      "epoch": 3.1090198863636362,
      "grad_norm": 26.265541076660156,
      "learning_rate": 7.813733228097869e-07,
      "loss": 1.3955,
      "step": 2420
    },
    {
      "epoch": 3.112571022727273,
      "grad_norm": 25.306949615478516,
      "learning_rate": 7.616416732438831e-07,
      "loss": 1.3381,
      "step": 2430
    },
    {
      "epoch": 3.116122159090909,
      "grad_norm": 18.622577667236328,
      "learning_rate": 7.419100236779796e-07,
      "loss": 1.3047,
      "step": 2440
    },
    {
      "epoch": 3.1196732954545454,
      "grad_norm": 14.70700454711914,
      "learning_rate": 7.221783741120759e-07,
      "loss": 1.3387,
      "step": 2450
    },
    {
      "epoch": 3.1232244318181817,
      "grad_norm": 30.96035385131836,
      "learning_rate": 7.024467245461721e-07,
      "loss": 1.3884,
      "step": 2460
    },
    {
      "epoch": 3.1267755681818183,
      "grad_norm": 12.719250679016113,
      "learning_rate": 6.827150749802685e-07,
      "loss": 1.4549,
      "step": 2470
    },
    {
      "epoch": 3.1303267045454546,
      "grad_norm": 16.468534469604492,
      "learning_rate": 6.629834254143647e-07,
      "loss": 1.1182,
      "step": 2480
    },
    {
      "epoch": 3.133877840909091,
      "grad_norm": 21.350452423095703,
      "learning_rate": 6.432517758484611e-07,
      "loss": 1.318,
      "step": 2490
    },
    {
      "epoch": 3.137428977272727,
      "grad_norm": 33.729400634765625,
      "learning_rate": 6.235201262825573e-07,
      "loss": 1.3263,
      "step": 2500
    },
    {
      "epoch": 3.1409801136363638,
      "grad_norm": 42.90283966064453,
      "learning_rate": 6.037884767166536e-07,
      "loss": 1.3509,
      "step": 2510
    },
    {
      "epoch": 3.14453125,
      "grad_norm": 45.704078674316406,
      "learning_rate": 5.840568271507499e-07,
      "loss": 1.5069,
      "step": 2520
    },
    {
      "epoch": 3.1480823863636362,
      "grad_norm": 41.951148986816406,
      "learning_rate": 5.643251775848462e-07,
      "loss": 1.3066,
      "step": 2530
    },
    {
      "epoch": 3.151633522727273,
      "grad_norm": 12.798999786376953,
      "learning_rate": 5.445935280189425e-07,
      "loss": 1.3401,
      "step": 2540
    },
    {
      "epoch": 3.155184659090909,
      "grad_norm": 21.00889778137207,
      "learning_rate": 5.248618784530387e-07,
      "loss": 1.2965,
      "step": 2550
    },
    {
      "epoch": 3.1587357954545454,
      "grad_norm": 17.473670959472656,
      "learning_rate": 5.05130228887135e-07,
      "loss": 1.2885,
      "step": 2560
    },
    {
      "epoch": 3.1622869318181817,
      "grad_norm": 28.33868980407715,
      "learning_rate": 4.853985793212313e-07,
      "loss": 1.3802,
      "step": 2570
    },
    {
      "epoch": 3.1658380681818183,
      "grad_norm": 37.80378723144531,
      "learning_rate": 4.6566692975532754e-07,
      "loss": 1.3751,
      "step": 2580
    },
    {
      "epoch": 3.1693892045454546,
      "grad_norm": 37.315486907958984,
      "learning_rate": 4.4593528018942384e-07,
      "loss": 1.4067,
      "step": 2590
    },
    {
      "epoch": 3.172940340909091,
      "grad_norm": 26.541744232177734,
      "learning_rate": 4.2620363062352015e-07,
      "loss": 1.3521,
      "step": 2600
    },
    {
      "epoch": 3.176491477272727,
      "grad_norm": 13.96971321105957,
      "learning_rate": 4.064719810576165e-07,
      "loss": 1.4583,
      "step": 2610
    },
    {
      "epoch": 3.1800426136363638,
      "grad_norm": 19.711917877197266,
      "learning_rate": 3.867403314917127e-07,
      "loss": 1.3062,
      "step": 2620
    },
    {
      "epoch": 3.18359375,
      "grad_norm": 28.168289184570312,
      "learning_rate": 3.6700868192580906e-07,
      "loss": 1.4479,
      "step": 2630
    },
    {
      "epoch": 3.1871448863636362,
      "grad_norm": 16.958044052124023,
      "learning_rate": 3.4727703235990536e-07,
      "loss": 1.2349,
      "step": 2640
    },
    {
      "epoch": 3.190696022727273,
      "grad_norm": 14.27488899230957,
      "learning_rate": 3.2754538279400156e-07,
      "loss": 1.3124,
      "step": 2650
    },
    {
      "epoch": 3.194247159090909,
      "grad_norm": 39.53180694580078,
      "learning_rate": 3.078137332280979e-07,
      "loss": 1.3704,
      "step": 2660
    },
    {
      "epoch": 3.1977982954545454,
      "grad_norm": 30.55574607849121,
      "learning_rate": 2.8808208366219417e-07,
      "loss": 1.4764,
      "step": 2670
    },
    {
      "epoch": 3.2013494318181817,
      "grad_norm": 32.31404113769531,
      "learning_rate": 2.6835043409629047e-07,
      "loss": 1.3712,
      "step": 2680
    },
    {
      "epoch": 3.2049005681818183,
      "grad_norm": 17.91683578491211,
      "learning_rate": 2.486187845303868e-07,
      "loss": 1.3872,
      "step": 2690
    },
    {
      "epoch": 3.2084517045454546,
      "grad_norm": 27.588768005371094,
      "learning_rate": 2.2888713496448305e-07,
      "loss": 1.3406,
      "step": 2700
    },
    {
      "epoch": 3.212002840909091,
      "grad_norm": 16.939321517944336,
      "learning_rate": 2.0915548539857933e-07,
      "loss": 1.4795,
      "step": 2710
    },
    {
      "epoch": 3.215553977272727,
      "grad_norm": 30.011600494384766,
      "learning_rate": 1.8942383583267563e-07,
      "loss": 1.2874,
      "step": 2720
    },
    {
      "epoch": 3.2191051136363638,
      "grad_norm": 38.46843719482422,
      "learning_rate": 1.696921862667719e-07,
      "loss": 1.4501,
      "step": 2730
    },
    {
      "epoch": 3.22265625,
      "grad_norm": 28.307205200195312,
      "learning_rate": 1.4996053670086821e-07,
      "loss": 1.3977,
      "step": 2740
    },
    {
      "epoch": 3.2262073863636362,
      "grad_norm": 36.628013610839844,
      "learning_rate": 1.302288871349645e-07,
      "loss": 1.2878,
      "step": 2750
    },
    {
      "epoch": 3.229758522727273,
      "grad_norm": 38.92404556274414,
      "learning_rate": 1.1049723756906078e-07,
      "loss": 1.3168,
      "step": 2760
    },
    {
      "epoch": 3.233309659090909,
      "grad_norm": 30.143110275268555,
      "learning_rate": 9.076558800315707e-08,
      "loss": 1.3171,
      "step": 2770
    },
    {
      "epoch": 3.2368607954545454,
      "grad_norm": 21.9947509765625,
      "learning_rate": 7.103393843725336e-08,
      "loss": 1.3034,
      "step": 2780
    },
    {
      "epoch": 3.2404119318181817,
      "grad_norm": 29.996990203857422,
      "learning_rate": 5.1302288871349646e-08,
      "loss": 1.3983,
      "step": 2790
    },
    {
      "epoch": 3.2439630681818183,
      "grad_norm": 67.70350646972656,
      "learning_rate": 3.1570639305445937e-08,
      "loss": 1.5066,
      "step": 2800
    },
    {
      "epoch": 3.2475142045454546,
      "grad_norm": 15.010125160217285,
      "learning_rate": 1.1838989739542227e-08,
      "loss": 1.3877,
      "step": 2810
    },
    {
      "epoch": 3.2496448863636362,
      "eval_accuracy": 0.3373493975903614,
      "eval_f1": 0.2839461670482319,
      "eval_loss": 1.9640586376190186,
      "eval_precision": 0.30306634843723573,
      "eval_recall": 0.3366896237317991,
      "eval_runtime": 724.6553,
      "eval_samples_per_second": 0.802,
      "eval_steps_per_second": 0.201,
      "step": 2816
    },
    {
      "epoch": 3.2496448863636362,
      "step": 2816,
      "total_flos": 1.4033059590817677e+19,
      "train_loss": 1.800246679986065,
      "train_runtime": 18585.0507,
      "train_samples_per_second": 0.606,
      "train_steps_per_second": 0.152
    },
    {
      "epoch": 3.2496448863636362,
      "eval_accuracy": 0.3268698060941828,
      "eval_f1": 0.2716015567288163,
      "eval_loss": 1.972208857536316,
      "eval_precision": 0.3970093449590282,
      "eval_recall": 0.32766441258833473,
      "eval_runtime": 907.789,
      "eval_samples_per_second": 0.795,
      "eval_steps_per_second": 0.199,
      "step": 2816
    },
    {
      "epoch": 3.2496448863636362,
      "eval_accuracy": 0.3268698060941828,
      "eval_f1": 0.2716015567288163,
      "eval_loss": 1.972209095954895,
      "eval_precision": 0.3970093449590282,
      "eval_recall": 0.32766441258833473,
      "eval_runtime": 909.7137,
      "eval_samples_per_second": 0.794,
      "eval_steps_per_second": 0.199,
      "step": 2816
    }
  ],
  "logging_steps": 10,
  "max_steps": 2816,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 500,
  "total_flos": 1.4033059590817677e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}