{
  "best_metric": 0.23535355925559998,
  "best_model_checkpoint": "autotrain-vit-large-patch16-224/checkpoint-10234",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 10234,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01709986320109439,
      "grad_norm": 16.03294563293457,
      "learning_rate": 8.207934336525308e-07,
      "loss": 3.1283,
      "step": 25
    },
    {
      "epoch": 0.03419972640218878,
      "grad_norm": 15.965583801269531,
      "learning_rate": 1.6073871409028727e-06,
      "loss": 2.7271,
      "step": 50
    },
    {
      "epoch": 0.05129958960328317,
      "grad_norm": 10.72908878326416,
      "learning_rate": 2.4623803009575924e-06,
      "loss": 2.2885,
      "step": 75
    },
    {
      "epoch": 0.06839945280437756,
      "grad_norm": 12.793851852416992,
      "learning_rate": 3.3173734610123124e-06,
      "loss": 1.9045,
      "step": 100
    },
    {
      "epoch": 0.08549931600547196,
      "grad_norm": 11.025067329406738,
      "learning_rate": 4.172366621067032e-06,
      "loss": 1.4383,
      "step": 125
    },
    {
      "epoch": 0.10259917920656635,
      "grad_norm": 13.093396186828613,
      "learning_rate": 5.027359781121752e-06,
      "loss": 1.3066,
      "step": 150
    },
    {
      "epoch": 0.11969904240766074,
      "grad_norm": 8.261626243591309,
      "learning_rate": 5.882352941176471e-06,
      "loss": 1.1223,
      "step": 175
    },
    {
      "epoch": 0.13679890560875513,
      "grad_norm": 9.678178787231445,
      "learning_rate": 6.73734610123119e-06,
      "loss": 1.1502,
      "step": 200
    },
    {
      "epoch": 0.1538987688098495,
      "grad_norm": 13.990431785583496,
      "learning_rate": 7.592339261285911e-06,
      "loss": 0.9833,
      "step": 225
    },
    {
      "epoch": 0.17099863201094392,
      "grad_norm": 11.807621002197266,
      "learning_rate": 8.44733242134063e-06,
      "loss": 0.8433,
      "step": 250
    },
    {
      "epoch": 0.1880984952120383,
      "grad_norm": 21.150636672973633,
      "learning_rate": 9.302325581395349e-06,
      "loss": 0.8553,
      "step": 275
    },
    {
      "epoch": 0.2051983584131327,
      "grad_norm": 16.56648826599121,
      "learning_rate": 1.0157318741450068e-05,
      "loss": 1.0502,
      "step": 300
    },
    {
      "epoch": 0.22229822161422708,
      "grad_norm": 8.26397705078125,
      "learning_rate": 1.1012311901504789e-05,
      "loss": 0.8686,
      "step": 325
    },
    {
      "epoch": 0.2393980848153215,
      "grad_norm": 7.628605842590332,
      "learning_rate": 1.1867305061559508e-05,
      "loss": 0.7838,
      "step": 350
    },
    {
      "epoch": 0.25649794801641584,
      "grad_norm": 6.690649509429932,
      "learning_rate": 1.2722298221614229e-05,
      "loss": 0.7284,
      "step": 375
    },
    {
      "epoch": 0.27359781121751026,
      "grad_norm": 9.397786140441895,
      "learning_rate": 1.3577291381668946e-05,
      "loss": 0.6408,
      "step": 400
    },
    {
      "epoch": 0.29069767441860467,
      "grad_norm": 1.714457392692566,
      "learning_rate": 1.4432284541723667e-05,
      "loss": 0.7046,
      "step": 425
    },
    {
      "epoch": 0.307797537619699,
      "grad_norm": 16.014860153198242,
      "learning_rate": 1.5287277701778386e-05,
      "loss": 0.7838,
      "step": 450
    },
    {
      "epoch": 0.32489740082079344,
      "grad_norm": 10.126370429992676,
      "learning_rate": 1.6142270861833107e-05,
      "loss": 0.735,
      "step": 475
    },
    {
      "epoch": 0.34199726402188785,
      "grad_norm": 10.995970726013184,
      "learning_rate": 1.6997264021887825e-05,
      "loss": 0.7523,
      "step": 500
    },
    {
      "epoch": 0.3590971272229822,
      "grad_norm": 7.909482955932617,
      "learning_rate": 1.7852257181942546e-05,
      "loss": 0.7157,
      "step": 525
    },
    {
      "epoch": 0.3761969904240766,
      "grad_norm": 10.751182556152344,
      "learning_rate": 1.8707250341997263e-05,
      "loss": 0.7966,
      "step": 550
    },
    {
      "epoch": 0.393296853625171,
      "grad_norm": 8.872684478759766,
      "learning_rate": 1.9562243502051984e-05,
      "loss": 0.6923,
      "step": 575
    },
    {
      "epoch": 0.4103967168262654,
      "grad_norm": 12.542756080627441,
      "learning_rate": 2.0417236662106705e-05,
      "loss": 0.8022,
      "step": 600
    },
    {
      "epoch": 0.4274965800273598,
      "grad_norm": 4.314883708953857,
      "learning_rate": 2.1272229822161423e-05,
      "loss": 0.6959,
      "step": 625
    },
    {
      "epoch": 0.44459644322845415,
      "grad_norm": 10.603111267089844,
      "learning_rate": 2.2127222982216144e-05,
      "loss": 0.7282,
      "step": 650
    },
    {
      "epoch": 0.46169630642954856,
      "grad_norm": 12.411425590515137,
      "learning_rate": 2.298221614227086e-05,
      "loss": 0.8191,
      "step": 675
    },
    {
      "epoch": 0.478796169630643,
      "grad_norm": 6.108178615570068,
      "learning_rate": 2.3837209302325582e-05,
      "loss": 0.507,
      "step": 700
    },
    {
      "epoch": 0.49589603283173733,
      "grad_norm": 11.323492050170898,
      "learning_rate": 2.46922024623803e-05,
      "loss": 0.5333,
      "step": 725
    },
    {
      "epoch": 0.5129958960328317,
      "grad_norm": 9.001029014587402,
      "learning_rate": 2.554719562243502e-05,
      "loss": 0.6816,
      "step": 750
    },
    {
      "epoch": 0.5300957592339262,
      "grad_norm": 1.6873475313186646,
      "learning_rate": 2.6402188782489738e-05,
      "loss": 0.6621,
      "step": 775
    },
    {
      "epoch": 0.5471956224350205,
      "grad_norm": 9.71667194366455,
      "learning_rate": 2.7257181942544462e-05,
      "loss": 0.5873,
      "step": 800
    },
    {
      "epoch": 0.5642954856361149,
      "grad_norm": 5.7018866539001465,
      "learning_rate": 2.811217510259918e-05,
      "loss": 0.6119,
      "step": 825
    },
    {
      "epoch": 0.5813953488372093,
      "grad_norm": 6.869658946990967,
      "learning_rate": 2.8967168262653897e-05,
      "loss": 0.8036,
      "step": 850
    },
    {
      "epoch": 0.5984952120383037,
      "grad_norm": 10.419452667236328,
      "learning_rate": 2.982216142270862e-05,
      "loss": 0.6884,
      "step": 875
    },
    {
      "epoch": 0.615595075239398,
      "grad_norm": 6.166721343994141,
      "learning_rate": 3.067715458276334e-05,
      "loss": 0.652,
      "step": 900
    },
    {
      "epoch": 0.6326949384404925,
      "grad_norm": 10.317444801330566,
      "learning_rate": 3.1532147742818057e-05,
      "loss": 0.6249,
      "step": 925
    },
    {
      "epoch": 0.6497948016415869,
      "grad_norm": 11.344230651855469,
      "learning_rate": 3.2387140902872784e-05,
      "loss": 0.6797,
      "step": 950
    },
    {
      "epoch": 0.6668946648426812,
      "grad_norm": 8.010811805725098,
      "learning_rate": 3.32421340629275e-05,
      "loss": 0.8409,
      "step": 975
    },
    {
      "epoch": 0.6839945280437757,
      "grad_norm": 8.49295425415039,
      "learning_rate": 3.409712722298222e-05,
      "loss": 0.5985,
      "step": 1000
    },
    {
      "epoch": 0.70109439124487,
      "grad_norm": 14.201081275939941,
      "learning_rate": 3.4952120383036933e-05,
      "loss": 0.7377,
      "step": 1025
    },
    {
      "epoch": 0.7181942544459644,
      "grad_norm": 12.270833015441895,
      "learning_rate": 3.580711354309166e-05,
      "loss": 0.5195,
      "step": 1050
    },
    {
      "epoch": 0.7352941176470589,
      "grad_norm": 2.879152297973633,
      "learning_rate": 3.6662106703146375e-05,
      "loss": 0.6526,
      "step": 1075
    },
    {
      "epoch": 0.7523939808481532,
      "grad_norm": 8.10263442993164,
      "learning_rate": 3.7517099863201096e-05,
      "loss": 0.7119,
      "step": 1100
    },
    {
      "epoch": 0.7694938440492476,
      "grad_norm": 4.584366321563721,
      "learning_rate": 3.837209302325582e-05,
      "loss": 0.6515,
      "step": 1125
    },
    {
      "epoch": 0.786593707250342,
      "grad_norm": 6.819866180419922,
      "learning_rate": 3.922708618331054e-05,
      "loss": 0.5304,
      "step": 1150
    },
    {
      "epoch": 0.8036935704514364,
      "grad_norm": 9.715897560119629,
      "learning_rate": 4.008207934336525e-05,
      "loss": 0.5718,
      "step": 1175
    },
    {
      "epoch": 0.8207934336525308,
      "grad_norm": 6.830766677856445,
      "learning_rate": 4.093707250341998e-05,
      "loss": 0.6228,
      "step": 1200
    },
    {
      "epoch": 0.8378932968536251,
      "grad_norm": 8.660406112670898,
      "learning_rate": 4.1792065663474694e-05,
      "loss": 0.7842,
      "step": 1225
    },
    {
      "epoch": 0.8549931600547196,
      "grad_norm": 8.981478691101074,
      "learning_rate": 4.2647058823529415e-05,
      "loss": 0.7702,
      "step": 1250
    },
    {
      "epoch": 0.872093023255814,
      "grad_norm": 5.491425037384033,
      "learning_rate": 4.3502051983584136e-05,
      "loss": 0.6483,
      "step": 1275
    },
    {
      "epoch": 0.8891928864569083,
      "grad_norm": 8.423704147338867,
      "learning_rate": 4.435704514363886e-05,
      "loss": 0.6889,
      "step": 1300
    },
    {
      "epoch": 0.9062927496580028,
      "grad_norm": 11.879225730895996,
      "learning_rate": 4.521203830369357e-05,
      "loss": 0.7592,
      "step": 1325
    },
    {
      "epoch": 0.9233926128590971,
      "grad_norm": 10.297099113464355,
      "learning_rate": 4.606703146374829e-05,
      "loss": 0.6994,
      "step": 1350
    },
    {
      "epoch": 0.9404924760601915,
      "grad_norm": 34.574771881103516,
      "learning_rate": 4.692202462380301e-05,
      "loss": 0.6254,
      "step": 1375
    },
    {
      "epoch": 0.957592339261286,
      "grad_norm": 7.829855442047119,
      "learning_rate": 4.7777017783857733e-05,
      "loss": 0.62,
      "step": 1400
    },
    {
      "epoch": 0.9746922024623803,
      "grad_norm": 10.330737113952637,
      "learning_rate": 4.863201094391245e-05,
      "loss": 0.5799,
      "step": 1425
    },
    {
      "epoch": 0.9917920656634747,
      "grad_norm": 7.56599235534668,
      "learning_rate": 4.948700410396717e-05,
      "loss": 0.6286,
      "step": 1450
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8296346876066917,
      "eval_f1_macro": 0.5237337030166633,
      "eval_f1_micro": 0.8296346876066917,
      "eval_f1_weighted": 0.818531665770605,
      "eval_loss": 0.5858680605888367,
      "eval_precision_macro": 0.6349808627313414,
      "eval_precision_micro": 0.8296346876066917,
      "eval_precision_weighted": 0.8428198229026794,
      "eval_recall_macro": 0.5158734772675513,
      "eval_recall_micro": 0.8296346876066917,
      "eval_recall_weighted": 0.8296346876066917,
      "eval_runtime": 29.6775,
      "eval_samples_per_second": 98.694,
      "eval_steps_per_second": 6.2,
      "step": 1462
    },
    {
      "epoch": 1.008891928864569,
      "grad_norm": 7.980313301086426,
      "learning_rate": 4.996200030399757e-05,
      "loss": 0.6037,
      "step": 1475
    },
    {
      "epoch": 1.0259917920656634,
      "grad_norm": 6.049495220184326,
      "learning_rate": 4.9867001063991494e-05,
      "loss": 0.6332,
      "step": 1500
    },
    {
      "epoch": 1.043091655266758,
      "grad_norm": 11.885618209838867,
      "learning_rate": 4.977200182398541e-05,
      "loss": 0.761,
      "step": 1525
    },
    {
      "epoch": 1.0601915184678523,
      "grad_norm": 14.27984619140625,
      "learning_rate": 4.9677002583979335e-05,
      "loss": 0.4568,
      "step": 1550
    },
    {
      "epoch": 1.0772913816689467,
      "grad_norm": 6.687506198883057,
      "learning_rate": 4.958200334397325e-05,
      "loss": 0.6308,
      "step": 1575
    },
    {
      "epoch": 1.094391244870041,
      "grad_norm": 6.140110492706299,
      "learning_rate": 4.948700410396717e-05,
      "loss": 0.5896,
      "step": 1600
    },
    {
      "epoch": 1.1114911080711354,
      "grad_norm": 8.485834121704102,
      "learning_rate": 4.939200486396109e-05,
      "loss": 0.7167,
      "step": 1625
    },
    {
      "epoch": 1.1285909712722297,
      "grad_norm": 8.836180686950684,
      "learning_rate": 4.929700562395501e-05,
      "loss": 0.5643,
      "step": 1650
    },
    {
      "epoch": 1.1456908344733243,
      "grad_norm": 5.937542915344238,
      "learning_rate": 4.9202006383948926e-05,
      "loss": 0.5568,
      "step": 1675
    },
    {
      "epoch": 1.1627906976744187,
      "grad_norm": 4.0352582931518555,
      "learning_rate": 4.910700714394285e-05,
      "loss": 0.7228,
      "step": 1700
    },
    {
      "epoch": 1.179890560875513,
      "grad_norm": 6.907437801361084,
      "learning_rate": 4.9012007903936766e-05,
      "loss": 0.5816,
      "step": 1725
    },
    {
      "epoch": 1.1969904240766074,
      "grad_norm": 6.376392841339111,
      "learning_rate": 4.891700866393069e-05,
      "loss": 0.5199,
      "step": 1750
    },
    {
      "epoch": 1.2140902872777017,
      "grad_norm": 9.630695343017578,
      "learning_rate": 4.882200942392461e-05,
      "loss": 0.6383,
      "step": 1775
    },
    {
      "epoch": 1.231190150478796,
      "grad_norm": 6.556370258331299,
      "learning_rate": 4.872701018391853e-05,
      "loss": 0.5977,
      "step": 1800
    },
    {
      "epoch": 1.2482900136798905,
      "grad_norm": 8.746458053588867,
      "learning_rate": 4.863201094391245e-05,
      "loss": 0.5699,
      "step": 1825
    },
    {
      "epoch": 1.265389876880985,
      "grad_norm": 6.0708537101745605,
      "learning_rate": 4.853701170390637e-05,
      "loss": 0.6479,
      "step": 1850
    },
    {
      "epoch": 1.2824897400820794,
      "grad_norm": 7.421648979187012,
      "learning_rate": 4.8442012463900295e-05,
      "loss": 0.7258,
      "step": 1875
    },
    {
      "epoch": 1.2995896032831737,
      "grad_norm": 11.180326461791992,
      "learning_rate": 4.834701322389421e-05,
      "loss": 0.5853,
      "step": 1900
    },
    {
      "epoch": 1.316689466484268,
      "grad_norm": 12.808277130126953,
      "learning_rate": 4.8252013983888135e-05,
      "loss": 0.625,
      "step": 1925
    },
    {
      "epoch": 1.3337893296853625,
      "grad_norm": 5.873989105224609,
      "learning_rate": 4.815701474388205e-05,
      "loss": 0.6044,
      "step": 1950
    },
    {
      "epoch": 1.350889192886457,
      "grad_norm": 6.302676200866699,
      "learning_rate": 4.8062015503875976e-05,
      "loss": 0.7381,
      "step": 1975
    },
    {
      "epoch": 1.3679890560875512,
      "grad_norm": 10.768912315368652,
      "learning_rate": 4.796701626386989e-05,
      "loss": 0.7416,
      "step": 2000
    },
    {
      "epoch": 1.3850889192886457,
      "grad_norm": 3.4966821670532227,
      "learning_rate": 4.7872017023863817e-05,
      "loss": 0.5816,
      "step": 2025
    },
    {
      "epoch": 1.40218878248974,
      "grad_norm": 20.856693267822266,
      "learning_rate": 4.7777017783857733e-05,
      "loss": 0.674,
      "step": 2050
    },
    {
      "epoch": 1.4192886456908345,
      "grad_norm": 3.353882074356079,
      "learning_rate": 4.768201854385165e-05,
      "loss": 0.6317,
      "step": 2075
    },
    {
      "epoch": 1.4363885088919288,
      "grad_norm": 4.5955939292907715,
      "learning_rate": 4.758701930384557e-05,
      "loss": 0.5058,
      "step": 2100
    },
    {
      "epoch": 1.4534883720930232,
      "grad_norm": 3.8296780586242676,
      "learning_rate": 4.749202006383949e-05,
      "loss": 0.55,
      "step": 2125
    },
    {
      "epoch": 1.4705882352941178,
      "grad_norm": 8.511550903320312,
      "learning_rate": 4.739702082383341e-05,
      "loss": 0.7234,
      "step": 2150
    },
    {
      "epoch": 1.487688098495212,
      "grad_norm": 7.452313423156738,
      "learning_rate": 4.730202158382733e-05,
      "loss": 0.4657,
      "step": 2175
    },
    {
      "epoch": 1.5047879616963065,
      "grad_norm": 5.583883762359619,
      "learning_rate": 4.720702234382125e-05,
      "loss": 0.5554,
      "step": 2200
    },
    {
      "epoch": 1.5218878248974008,
      "grad_norm": 3.6462392807006836,
      "learning_rate": 4.711202310381517e-05,
      "loss": 0.6182,
      "step": 2225
    },
    {
      "epoch": 1.5389876880984952,
      "grad_norm": 5.7832207679748535,
      "learning_rate": 4.701702386380909e-05,
      "loss": 0.4694,
      "step": 2250
    },
    {
      "epoch": 1.5560875512995898,
      "grad_norm": 3.335848331451416,
      "learning_rate": 4.692202462380301e-05,
      "loss": 0.6423,
      "step": 2275
    },
    {
      "epoch": 1.573187414500684,
      "grad_norm": 15.582513809204102,
      "learning_rate": 4.682702538379693e-05,
      "loss": 0.6912,
      "step": 2300
    },
    {
      "epoch": 1.5902872777017785,
      "grad_norm": 8.966438293457031,
      "learning_rate": 4.673202614379085e-05,
      "loss": 0.4998,
      "step": 2325
    },
    {
      "epoch": 1.6073871409028728,
      "grad_norm": 6.383381366729736,
      "learning_rate": 4.663702690378477e-05,
      "loss": 0.5879,
      "step": 2350
    },
    {
      "epoch": 1.6244870041039672,
      "grad_norm": 3.314211845397949,
      "learning_rate": 4.6542027663778694e-05,
      "loss": 0.5865,
      "step": 2375
    },
    {
      "epoch": 1.6415868673050615,
      "grad_norm": 7.680863380432129,
      "learning_rate": 4.644702842377261e-05,
      "loss": 0.5731,
      "step": 2400
    },
    {
      "epoch": 1.658686730506156,
      "grad_norm": 15.084454536437988,
      "learning_rate": 4.6352029183766534e-05,
      "loss": 0.6018,
      "step": 2425
    },
    {
      "epoch": 1.6757865937072505,
      "grad_norm": 4.016242027282715,
      "learning_rate": 4.625702994376046e-05,
      "loss": 0.4291,
      "step": 2450
    },
    {
      "epoch": 1.6928864569083446,
      "grad_norm": 6.622522354125977,
      "learning_rate": 4.6162030703754375e-05,
      "loss": 0.5729,
      "step": 2475
    },
    {
      "epoch": 1.7099863201094392,
      "grad_norm": 7.149597644805908,
      "learning_rate": 4.606703146374829e-05,
      "loss": 0.6435,
      "step": 2500
    },
    {
      "epoch": 1.7270861833105335,
      "grad_norm": 4.473049163818359,
      "learning_rate": 4.597203222374221e-05,
      "loss": 0.5068,
      "step": 2525
    },
    {
      "epoch": 1.744186046511628,
      "grad_norm": 8.462128639221191,
      "learning_rate": 4.587703298373613e-05,
      "loss": 0.4735,
      "step": 2550
    },
    {
      "epoch": 1.7612859097127223,
      "grad_norm": 3.4091100692749023,
      "learning_rate": 4.578203374373005e-05,
      "loss": 0.591,
      "step": 2575
    },
    {
      "epoch": 1.7783857729138166,
      "grad_norm": 4.581203937530518,
      "learning_rate": 4.568703450372397e-05,
      "loss": 0.4642,
      "step": 2600
    },
    {
      "epoch": 1.7954856361149112,
      "grad_norm": 7.462313175201416,
      "learning_rate": 4.559203526371789e-05,
      "loss": 0.5036,
      "step": 2625
    },
    {
      "epoch": 1.8125854993160053,
      "grad_norm": 7.391847610473633,
      "learning_rate": 4.549703602371181e-05,
      "loss": 0.5039,
      "step": 2650
    },
    {
      "epoch": 1.8296853625171,
      "grad_norm": 16.374174118041992,
      "learning_rate": 4.540203678370573e-05,
      "loss": 0.6136,
      "step": 2675
    },
    {
      "epoch": 1.8467852257181943,
      "grad_norm": 7.047588348388672,
      "learning_rate": 4.5307037543699654e-05,
      "loss": 0.4051,
      "step": 2700
    },
    {
      "epoch": 1.8638850889192886,
      "grad_norm": 4.181390285491943,
      "learning_rate": 4.521203830369357e-05,
      "loss": 0.7065,
      "step": 2725
    },
    {
      "epoch": 1.8809849521203832,
      "grad_norm": 2.601778984069824,
      "learning_rate": 4.5117039063687494e-05,
      "loss": 0.5517,
      "step": 2750
    },
    {
      "epoch": 1.8980848153214773,
      "grad_norm": 2.9935567378997803,
      "learning_rate": 4.502203982368141e-05,
      "loss": 0.3861,
      "step": 2775
    },
    {
      "epoch": 1.915184678522572,
      "grad_norm": 7.431847095489502,
      "learning_rate": 4.4927040583675335e-05,
      "loss": 0.5537,
      "step": 2800
    },
    {
      "epoch": 1.9322845417236663,
      "grad_norm": 7.30590295791626,
      "learning_rate": 4.483204134366925e-05,
      "loss": 0.5082,
      "step": 2825
    },
    {
      "epoch": 1.9493844049247606,
      "grad_norm": 6.986799240112305,
      "learning_rate": 4.4737042103663176e-05,
      "loss": 0.5727,
      "step": 2850
    },
    {
      "epoch": 1.966484268125855,
      "grad_norm": 4.823264122009277,
      "learning_rate": 4.464204286365709e-05,
      "loss": 0.5773,
      "step": 2875
    },
    {
      "epoch": 1.9835841313269493,
      "grad_norm": 0.19885371625423431,
      "learning_rate": 4.4547043623651016e-05,
      "loss": 0.5056,
      "step": 2900
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9095254353021509,
      "eval_f1_macro": 0.7148556374573719,
      "eval_f1_micro": 0.9095254353021509,
      "eval_f1_weighted": 0.906235846022486,
      "eval_loss": 0.2884175181388855,
      "eval_precision_macro": 0.779691628927798,
      "eval_precision_micro": 0.9095254353021509,
      "eval_precision_weighted": 0.9118121732094211,
      "eval_recall_macro": 0.6904984139815875,
      "eval_recall_micro": 0.9095254353021509,
      "eval_recall_weighted": 0.9095254353021509,
      "eval_runtime": 29.8929,
      "eval_samples_per_second": 97.983,
      "eval_steps_per_second": 6.155,
      "step": 2924
    },
    {
      "epoch": 2.000683994528044,
      "grad_norm": 8.045063972473145,
      "learning_rate": 4.445204438364493e-05,
      "loss": 0.5348,
      "step": 2925
    },
    {
      "epoch": 2.017783857729138,
      "grad_norm": 7.1167521476745605,
      "learning_rate": 4.435704514363886e-05,
      "loss": 0.4205,
      "step": 2950
    },
    {
      "epoch": 2.0348837209302326,
      "grad_norm": 7.508376121520996,
      "learning_rate": 4.4262045903632774e-05,
      "loss": 0.5367,
      "step": 2975
    },
    {
      "epoch": 2.0519835841313268,
      "grad_norm": 10.526673316955566,
      "learning_rate": 4.416704666362669e-05,
      "loss": 0.4326,
      "step": 3000
    },
    {
      "epoch": 2.0690834473324213,
      "grad_norm": 7.46666145324707,
      "learning_rate": 4.407204742362061e-05,
      "loss": 0.6013,
      "step": 3025
    },
    {
      "epoch": 2.086183310533516,
      "grad_norm": 3.7393107414245605,
      "learning_rate": 4.397704818361453e-05,
      "loss": 0.5544,
      "step": 3050
    },
    {
      "epoch": 2.10328317373461,
      "grad_norm": 6.285150051116943,
      "learning_rate": 4.3882048943608455e-05,
      "loss": 0.5717,
      "step": 3075
    },
    {
      "epoch": 2.1203830369357046,
      "grad_norm": 8.156254768371582,
      "learning_rate": 4.378704970360237e-05,
      "loss": 0.5008,
      "step": 3100
    },
    {
      "epoch": 2.1374829001367988,
      "grad_norm": 10.280779838562012,
      "learning_rate": 4.3692050463596295e-05,
      "loss": 0.3871,
      "step": 3125
    },
    {
      "epoch": 2.1545827633378933,
      "grad_norm": 11.541882514953613,
      "learning_rate": 4.359705122359021e-05,
      "loss": 0.4684,
      "step": 3150
    },
    {
      "epoch": 2.1716826265389875,
      "grad_norm": 12.31938648223877,
      "learning_rate": 4.3502051983584136e-05,
      "loss": 0.5055,
      "step": 3175
    },
    {
      "epoch": 2.188782489740082,
      "grad_norm": 6.534464359283447,
      "learning_rate": 4.340705274357805e-05,
      "loss": 0.682,
      "step": 3200
    },
    {
      "epoch": 2.2058823529411766,
      "grad_norm": 6.284212112426758,
      "learning_rate": 4.3312053503571976e-05,
      "loss": 0.5189,
      "step": 3225
    },
    {
      "epoch": 2.2229822161422708,
      "grad_norm": 4.796578884124756,
      "learning_rate": 4.321705426356589e-05,
      "loss": 0.5516,
      "step": 3250
    },
    {
      "epoch": 2.2400820793433653,
      "grad_norm": 0.08828147500753403,
      "learning_rate": 4.312205502355982e-05,
      "loss": 0.4325,
      "step": 3275
    },
    {
      "epoch": 2.2571819425444595,
      "grad_norm": 3.709010362625122,
      "learning_rate": 4.3027055783553734e-05,
      "loss": 0.524,
      "step": 3300
    },
    {
      "epoch": 2.274281805745554,
      "grad_norm": 6.48209810256958,
      "learning_rate": 4.293205654354766e-05,
      "loss": 0.4304,
      "step": 3325
    },
    {
      "epoch": 2.2913816689466486,
      "grad_norm": 7.461411952972412,
      "learning_rate": 4.2837057303541574e-05,
      "loss": 0.5894,
      "step": 3350
    },
    {
      "epoch": 2.3084815321477428,
      "grad_norm": 6.687474727630615,
      "learning_rate": 4.27420580635355e-05,
      "loss": 0.4504,
      "step": 3375
    },
    {
      "epoch": 2.3255813953488373,
      "grad_norm": 4.4408745765686035,
      "learning_rate": 4.2647058823529415e-05,
      "loss": 0.5091,
      "step": 3400
    },
    {
      "epoch": 2.3426812585499315,
      "grad_norm": 7.8160552978515625,
      "learning_rate": 4.255205958352333e-05,
      "loss": 0.3612,
      "step": 3425
    },
    {
      "epoch": 2.359781121751026,
      "grad_norm": 4.452754497528076,
      "learning_rate": 4.2457060343517255e-05,
      "loss": 0.6184,
      "step": 3450
    },
    {
      "epoch": 2.37688098495212,
      "grad_norm": 12.292354583740234,
      "learning_rate": 4.236206110351117e-05,
      "loss": 0.5868,
      "step": 3475
    },
    {
      "epoch": 2.3939808481532148,
      "grad_norm": 3.688055992126465,
      "learning_rate": 4.226706186350509e-05,
      "loss": 0.5325,
      "step": 3500
    },
    {
      "epoch": 2.4110807113543093,
      "grad_norm": 5.77033805847168,
      "learning_rate": 4.217206262349901e-05,
      "loss": 0.3577,
      "step": 3525
    },
    {
      "epoch": 2.4281805745554035,
      "grad_norm": 4.816909313201904,
      "learning_rate": 4.207706338349293e-05,
      "loss": 0.4681,
      "step": 3550
    },
    {
      "epoch": 2.445280437756498,
      "grad_norm": 8.695076942443848,
      "learning_rate": 4.198206414348685e-05,
      "loss": 0.4151,
      "step": 3575
    },
    {
      "epoch": 2.462380300957592,
      "grad_norm": 10.278604507446289,
      "learning_rate": 4.188706490348077e-05,
      "loss": 0.4165,
      "step": 3600
    },
    {
      "epoch": 2.4794801641586868,
      "grad_norm": 17.01686668395996,
      "learning_rate": 4.1792065663474694e-05,
      "loss": 0.5273,
      "step": 3625
    },
    {
      "epoch": 2.496580027359781,
      "grad_norm": 7.072406768798828,
      "learning_rate": 4.170086639306886e-05,
      "loss": 0.4383,
      "step": 3650
    },
    {
      "epoch": 2.5136798905608755,
      "grad_norm": 5.283012866973877,
      "learning_rate": 4.1605867153062774e-05,
      "loss": 0.578,
      "step": 3675
    },
    {
      "epoch": 2.53077975376197,
      "grad_norm": 3.6656084060668945,
      "learning_rate": 4.15108679130567e-05,
      "loss": 0.491,
      "step": 3700
    },
    {
      "epoch": 2.547879616963064,
      "grad_norm": 9.410102844238281,
      "learning_rate": 4.1415868673050615e-05,
      "loss": 0.4382,
      "step": 3725
    },
    {
      "epoch": 2.5649794801641588,
      "grad_norm": 8.41618824005127,
      "learning_rate": 4.132086943304454e-05,
      "loss": 0.5658,
      "step": 3750
    },
    {
      "epoch": 2.582079343365253,
      "grad_norm": 1.3863807916641235,
      "learning_rate": 4.1225870193038455e-05,
      "loss": 0.5228,
      "step": 3775
    },
    {
      "epoch": 2.5991792065663475,
      "grad_norm": 3.929469585418701,
      "learning_rate": 4.113087095303238e-05,
      "loss": 0.4993,
      "step": 3800
    },
    {
      "epoch": 2.616279069767442,
      "grad_norm": 4.706065654754639,
      "learning_rate": 4.1035871713026296e-05,
      "loss": 0.4045,
      "step": 3825
    },
    {
      "epoch": 2.633378932968536,
      "grad_norm": 2.805471658706665,
      "learning_rate": 4.094087247302022e-05,
      "loss": 0.4283,
      "step": 3850
    },
    {
      "epoch": 2.650478796169631,
      "grad_norm": 8.297080993652344,
      "learning_rate": 4.0845873233014136e-05,
      "loss": 0.4499,
      "step": 3875
    },
    {
      "epoch": 2.667578659370725,
      "grad_norm": 2.637622594833374,
      "learning_rate": 4.075087399300806e-05,
      "loss": 0.5087,
      "step": 3900
    },
    {
      "epoch": 2.6846785225718195,
      "grad_norm": 2.461876392364502,
      "learning_rate": 4.065587475300198e-05,
      "loss": 0.5104,
      "step": 3925
    },
    {
      "epoch": 2.701778385772914,
      "grad_norm": 6.3255486488342285,
      "learning_rate": 4.05608755129959e-05,
      "loss": 0.5861,
      "step": 3950
    },
    {
      "epoch": 2.718878248974008,
      "grad_norm": 11.981715202331543,
      "learning_rate": 4.046587627298982e-05,
      "loss": 0.5593,
      "step": 3975
    },
    {
      "epoch": 2.7359781121751023,
      "grad_norm": 7.206056594848633,
      "learning_rate": 4.037087703298374e-05,
      "loss": 0.4665,
      "step": 4000
    },
    {
      "epoch": 2.753077975376197,
      "grad_norm": 1.9749879837036133,
      "learning_rate": 4.027587779297766e-05,
      "loss": 0.4984,
      "step": 4025
    },
    {
      "epoch": 2.7701778385772915,
      "grad_norm": 4.363369464874268,
      "learning_rate": 4.018087855297158e-05,
      "loss": 0.5785,
      "step": 4050
    },
    {
      "epoch": 2.7872777017783856,
      "grad_norm": 6.443023681640625,
      "learning_rate": 4.00858793129655e-05,
      "loss": 0.5287,
      "step": 4075
    },
    {
      "epoch": 2.80437756497948,
      "grad_norm": 6.106603622436523,
      "learning_rate": 3.999088007295942e-05,
      "loss": 0.6016,
      "step": 4100
    },
    {
      "epoch": 2.8214774281805743,
      "grad_norm": 9.416887283325195,
      "learning_rate": 3.989588083295334e-05,
      "loss": 0.5532,
      "step": 4125
    },
    {
      "epoch": 2.838577291381669,
      "grad_norm": 2.5579991340637207,
      "learning_rate": 3.9800881592947256e-05,
      "loss": 0.4579,
      "step": 4150
    },
    {
      "epoch": 2.8556771545827635,
      "grad_norm": 10.23747444152832,
      "learning_rate": 3.970588235294117e-05,
      "loss": 0.5402,
      "step": 4175
    },
    {
      "epoch": 2.8727770177838576,
      "grad_norm": 9.435909271240234,
      "learning_rate": 3.96108831129351e-05,
      "loss": 0.4845,
      "step": 4200
    },
    {
      "epoch": 2.889876880984952,
      "grad_norm": 3.5227017402648926,
      "learning_rate": 3.9515883872929014e-05,
      "loss": 0.3969,
      "step": 4225
    },
    {
      "epoch": 2.9069767441860463,
      "grad_norm": 2.499969482421875,
      "learning_rate": 3.942088463292294e-05,
      "loss": 0.4392,
      "step": 4250
    },
    {
      "epoch": 2.924076607387141,
      "grad_norm": 8.345962524414062,
      "learning_rate": 3.9325885392916854e-05,
      "loss": 0.4584,
      "step": 4275
    },
    {
      "epoch": 2.9411764705882355,
      "grad_norm": 3.583355188369751,
      "learning_rate": 3.923088615291078e-05,
      "loss": 0.3816,
      "step": 4300
    },
    {
      "epoch": 2.9582763337893296,
      "grad_norm": 1.7532494068145752,
      "learning_rate": 3.91358869129047e-05,
      "loss": 0.3497,
      "step": 4325
    },
    {
      "epoch": 2.975376196990424,
      "grad_norm": 9.220611572265625,
      "learning_rate": 3.904088767289862e-05,
      "loss": 0.5747,
      "step": 4350
    },
    {
      "epoch": 2.9924760601915183,
      "grad_norm": 1.715482234954834,
      "learning_rate": 3.894588843289254e-05,
      "loss": 0.4701,
      "step": 4375
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9071355411403209,
      "eval_f1_macro": 0.7615092412094518,
      "eval_f1_micro": 0.9071355411403209,
      "eval_f1_weighted": 0.9053250303728207,
      "eval_loss": 0.2766139805316925,
      "eval_precision_macro": 0.8223788212097681,
      "eval_precision_micro": 0.9071355411403209,
      "eval_precision_weighted": 0.9095949572268466,
      "eval_recall_macro": 0.7322063659665187,
      "eval_recall_micro": 0.9071355411403209,
      "eval_recall_weighted": 0.9071355411403209,
      "eval_runtime": 29.8386,
      "eval_samples_per_second": 98.161,
      "eval_steps_per_second": 6.167,
      "step": 4386
    },
    {
      "epoch": 3.009575923392613,
      "grad_norm": 4.551536560058594,
      "learning_rate": 3.885088919288646e-05,
      "loss": 0.4776,
      "step": 4400
    },
    {
      "epoch": 3.026675786593707,
      "grad_norm": 10.111888885498047,
      "learning_rate": 3.875588995288038e-05,
      "loss": 0.5966,
      "step": 4425
    },
    {
      "epoch": 3.0437756497948016,
      "grad_norm": 13.998083114624023,
      "learning_rate": 3.86608907128743e-05,
      "loss": 0.4699,
      "step": 4450
    },
    {
      "epoch": 3.060875512995896,
      "grad_norm": 10.541669845581055,
      "learning_rate": 3.856589147286822e-05,
      "loss": 0.5154,
      "step": 4475
    },
    {
      "epoch": 3.0779753761969904,
      "grad_norm": 3.1978766918182373,
      "learning_rate": 3.847089223286214e-05,
      "loss": 0.4304,
      "step": 4500
    },
    {
      "epoch": 3.095075239398085,
      "grad_norm": 5.206043720245361,
      "learning_rate": 3.8375892992856064e-05,
      "loss": 0.3223,
      "step": 4525
    },
    {
      "epoch": 3.112175102599179,
      "grad_norm": 9.216475486755371,
      "learning_rate": 3.828089375284998e-05,
      "loss": 0.6146,
      "step": 4550
    },
    {
      "epoch": 3.1292749658002736,
      "grad_norm": 4.771245002746582,
      "learning_rate": 3.81858945128439e-05,
      "loss": 0.3407,
      "step": 4575
    },
    {
      "epoch": 3.146374829001368,
      "grad_norm": 8.766341209411621,
      "learning_rate": 3.8090895272837814e-05,
      "loss": 0.4866,
      "step": 4600
    },
    {
      "epoch": 3.1634746922024624,
      "grad_norm": 6.949918270111084,
      "learning_rate": 3.799589603283174e-05,
      "loss": 0.4899,
      "step": 4625
    },
    {
      "epoch": 3.180574555403557,
      "grad_norm": 1.0002543926239014,
      "learning_rate": 3.7900896792825655e-05,
      "loss": 0.481,
      "step": 4650
    },
    {
      "epoch": 3.197674418604651,
      "grad_norm": 9.768813133239746,
      "learning_rate": 3.780589755281958e-05,
      "loss": 0.4338,
      "step": 4675
    },
    {
      "epoch": 3.2147742818057456,
      "grad_norm": 4.305062770843506,
      "learning_rate": 3.7710898312813495e-05,
      "loss": 0.3783,
      "step": 4700
    },
    {
      "epoch": 3.23187414500684,
      "grad_norm": 0.7493396997451782,
      "learning_rate": 3.761589907280742e-05,
      "loss": 0.5001,
      "step": 4725
    },
    {
      "epoch": 3.2489740082079344,
      "grad_norm": 5.481766700744629,
      "learning_rate": 3.7520899832801336e-05,
      "loss": 0.4894,
      "step": 4750
    },
    {
      "epoch": 3.266073871409029,
      "grad_norm": 3.8942747116088867,
      "learning_rate": 3.742590059279526e-05,
      "loss": 0.4761,
      "step": 4775
    },
    {
      "epoch": 3.283173734610123,
      "grad_norm": 8.91220760345459,
      "learning_rate": 3.7330901352789176e-05,
      "loss": 0.5526,
      "step": 4800
    },
    {
      "epoch": 3.3002735978112177,
      "grad_norm": 1.765817642211914,
      "learning_rate": 3.72359021127831e-05,
      "loss": 0.5665,
      "step": 4825
    },
    {
      "epoch": 3.317373461012312,
      "grad_norm": 6.553985118865967,
      "learning_rate": 3.714090287277702e-05,
      "loss": 0.287,
      "step": 4850
    },
    {
      "epoch": 3.3344733242134064,
      "grad_norm": 5.839998245239258,
      "learning_rate": 3.704590363277094e-05,
      "loss": 0.4927,
      "step": 4875
    },
    {
      "epoch": 3.3515731874145005,
      "grad_norm": 8.003843307495117,
      "learning_rate": 3.6950904392764864e-05,
      "loss": 0.552,
      "step": 4900
    },
    {
      "epoch": 3.368673050615595,
      "grad_norm": 3.3592958450317383,
      "learning_rate": 3.685590515275878e-05,
      "loss": 0.5703,
      "step": 4925
    },
    {
      "epoch": 3.3857729138166897,
      "grad_norm": 4.747976779937744,
      "learning_rate": 3.6760905912752705e-05,
      "loss": 0.5218,
      "step": 4950
    },
    {
      "epoch": 3.402872777017784,
      "grad_norm": 4.351706504821777,
      "learning_rate": 3.666590667274662e-05,
      "loss": 0.4671,
      "step": 4975
    },
    {
      "epoch": 3.4199726402188784,
      "grad_norm": 5.302130222320557,
      "learning_rate": 3.6570907432740545e-05,
      "loss": 0.3913,
      "step": 5000
    },
    {
      "epoch": 3.4370725034199725,
      "grad_norm": 1.5275038480758667,
      "learning_rate": 3.647590819273446e-05,
      "loss": 0.3149,
      "step": 5025
    },
    {
      "epoch": 3.454172366621067,
      "grad_norm": 0.8008119463920593,
      "learning_rate": 3.638090895272838e-05,
      "loss": 0.5232,
      "step": 5050
    },
    {
      "epoch": 3.471272229822161,
      "grad_norm": 6.361542701721191,
      "learning_rate": 3.6285909712722296e-05,
      "loss": 0.4149,
      "step": 5075
    },
    {
      "epoch": 3.488372093023256,
      "grad_norm": 0.8227640390396118,
      "learning_rate": 3.619091047271622e-05,
      "loss": 0.5698,
      "step": 5100
    },
    {
      "epoch": 3.5054719562243504,
      "grad_norm": 4.734671115875244,
      "learning_rate": 3.609591123271014e-05,
      "loss": 0.4835,
      "step": 5125
    },
    {
      "epoch": 3.5225718194254445,
      "grad_norm": 6.9522576332092285,
      "learning_rate": 3.600091199270406e-05,
      "loss": 0.5387,
      "step": 5150
    },
    {
      "epoch": 3.539671682626539,
      "grad_norm": 3.4246246814727783,
      "learning_rate": 3.590591275269798e-05,
      "loss": 0.4044,
      "step": 5175
    },
    {
      "epoch": 3.556771545827633,
      "grad_norm": 6.183599472045898,
      "learning_rate": 3.58109135126919e-05,
      "loss": 0.4785,
      "step": 5200
    },
    {
      "epoch": 3.573871409028728,
      "grad_norm": 3.738954782485962,
      "learning_rate": 3.571591427268582e-05,
      "loss": 0.5516,
      "step": 5225
    },
    {
      "epoch": 3.5909712722298224,
      "grad_norm": 7.439993381500244,
      "learning_rate": 3.562091503267974e-05,
      "loss": 0.523,
      "step": 5250
    },
    {
      "epoch": 3.6080711354309165,
      "grad_norm": 4.419105052947998,
      "learning_rate": 3.552591579267366e-05,
      "loss": 0.4118,
      "step": 5275
    },
    {
      "epoch": 3.625170998632011,
      "grad_norm": 16.105833053588867,
      "learning_rate": 3.543091655266758e-05,
      "loss": 0.5083,
      "step": 5300
    },
    {
      "epoch": 3.6422708618331052,
      "grad_norm": 5.370678424835205,
      "learning_rate": 3.53359173126615e-05,
      "loss": 0.4519,
      "step": 5325
    },
    {
      "epoch": 3.6593707250342,
      "grad_norm": 11.25657844543457,
      "learning_rate": 3.524091807265542e-05,
      "loss": 0.3801,
      "step": 5350
    },
    {
      "epoch": 3.6764705882352944,
      "grad_norm": 3.6341259479522705,
      "learning_rate": 3.514591883264934e-05,
      "loss": 0.3978,
      "step": 5375
    },
    {
      "epoch": 3.6935704514363885,
      "grad_norm": 0.18588456511497498,
      "learning_rate": 3.505091959264326e-05,
      "loss": 0.4341,
      "step": 5400
    },
    {
      "epoch": 3.7106703146374826,
      "grad_norm": 4.884586811065674,
      "learning_rate": 3.495592035263718e-05,
      "loss": 0.5095,
      "step": 5425
    },
    {
      "epoch": 3.7277701778385772,
      "grad_norm": 5.724035739898682,
      "learning_rate": 3.4860921112631104e-05,
      "loss": 0.4625,
      "step": 5450
    },
    {
      "epoch": 3.744870041039672,
      "grad_norm": 5.873854637145996,
      "learning_rate": 3.476592187262502e-05,
      "loss": 0.4019,
      "step": 5475
    },
    {
      "epoch": 3.761969904240766,
      "grad_norm": 3.843010902404785,
      "learning_rate": 3.467092263261894e-05,
      "loss": 0.5835,
      "step": 5500
    },
    {
      "epoch": 3.7790697674418605,
      "grad_norm": 5.200079441070557,
      "learning_rate": 3.457592339261286e-05,
      "loss": 0.4441,
      "step": 5525
    },
    {
      "epoch": 3.7961696306429547,
      "grad_norm": 7.437492847442627,
      "learning_rate": 3.448092415260678e-05,
      "loss": 0.3881,
      "step": 5550
    },
    {
      "epoch": 3.8132694938440492,
      "grad_norm": 5.660449504852295,
      "learning_rate": 3.43859249126007e-05,
      "loss": 0.446,
      "step": 5575
    },
    {
      "epoch": 3.830369357045144,
      "grad_norm": 2.2128264904022217,
      "learning_rate": 3.429092567259462e-05,
      "loss": 0.4257,
      "step": 5600
    },
    {
      "epoch": 3.847469220246238,
      "grad_norm": 5.218535423278809,
      "learning_rate": 3.419592643258854e-05,
      "loss": 0.4946,
      "step": 5625
    },
    {
      "epoch": 3.8645690834473325,
      "grad_norm": 1.4989601373672485,
      "learning_rate": 3.410092719258246e-05,
      "loss": 0.3866,
      "step": 5650
    },
    {
      "epoch": 3.8816689466484267,
      "grad_norm": 11.225085258483887,
      "learning_rate": 3.400592795257638e-05,
      "loss": 0.4764,
      "step": 5675
    },
    {
      "epoch": 3.8987688098495212,
      "grad_norm": 5.653949737548828,
      "learning_rate": 3.39109287125703e-05,
      "loss": 0.4362,
      "step": 5700
    },
    {
      "epoch": 3.915868673050616,
      "grad_norm": 10.87972354888916,
      "learning_rate": 3.381592947256422e-05,
      "loss": 0.4708,
      "step": 5725
    },
    {
      "epoch": 3.93296853625171,
      "grad_norm": 5.110008239746094,
      "learning_rate": 3.372093023255814e-05,
      "loss": 0.4373,
      "step": 5750
    },
    {
      "epoch": 3.9500683994528045,
      "grad_norm": 4.611896991729736,
      "learning_rate": 3.3625930992552064e-05,
      "loss": 0.4808,
      "step": 5775
    },
    {
      "epoch": 3.9671682626538987,
      "grad_norm": 3.967752695083618,
      "learning_rate": 3.353093175254598e-05,
      "loss": 0.3827,
      "step": 5800
    },
    {
      "epoch": 3.9842681258549932,
      "grad_norm": 7.738473892211914,
      "learning_rate": 3.3435932512539904e-05,
      "loss": 0.4564,
      "step": 5825
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8968931375896211,
      "eval_f1_macro": 0.6842607301838518,
      "eval_f1_micro": 0.8968931375896211,
      "eval_f1_weighted": 0.8922998232908407,
      "eval_loss": 0.3079198896884918,
      "eval_precision_macro": 0.7711077815931238,
      "eval_precision_micro": 0.8968931375896211,
      "eval_precision_weighted": 0.9026293299812516,
      "eval_recall_macro": 0.6595431638738259,
      "eval_recall_micro": 0.8968931375896211,
      "eval_recall_weighted": 0.8968931375896211,
      "eval_runtime": 29.6904,
      "eval_samples_per_second": 98.651,
      "eval_steps_per_second": 6.197,
      "step": 5848
    },
    {
      "epoch": 4.001367989056088,
      "grad_norm": 12.561144828796387,
      "learning_rate": 3.334093327253382e-05,
      "loss": 0.285,
      "step": 5850
    },
    {
      "epoch": 4.0184678522571815,
      "grad_norm": 4.646189212799072,
      "learning_rate": 3.3245934032527745e-05,
      "loss": 0.4118,
      "step": 5875
    },
    {
      "epoch": 4.035567715458276,
      "grad_norm": 3.1821000576019287,
      "learning_rate": 3.315093479252166e-05,
      "loss": 0.5941,
      "step": 5900
    },
    {
      "epoch": 4.052667578659371,
      "grad_norm": 6.912432670593262,
      "learning_rate": 3.3055935552515586e-05,
      "loss": 0.3519,
      "step": 5925
    },
    {
      "epoch": 4.069767441860465,
      "grad_norm": 2.416905641555786,
      "learning_rate": 3.29609363125095e-05,
      "loss": 0.4982,
      "step": 5950
    },
    {
      "epoch": 4.08686730506156,
      "grad_norm": 8.833620071411133,
      "learning_rate": 3.286593707250342e-05,
      "loss": 0.3608,
      "step": 5975
    },
    {
      "epoch": 4.1039671682626535,
      "grad_norm": 0.06249883025884628,
      "learning_rate": 3.2770937832497336e-05,
      "loss": 0.4883,
      "step": 6000
    },
    {
      "epoch": 4.121067031463748,
      "grad_norm": 2.5257112979888916,
      "learning_rate": 3.267593859249126e-05,
      "loss": 0.3371,
      "step": 6025
    },
    {
      "epoch": 4.138166894664843,
      "grad_norm": 5.706219673156738,
      "learning_rate": 3.258093935248518e-05,
      "loss": 0.3998,
      "step": 6050
    },
    {
      "epoch": 4.155266757865937,
      "grad_norm": 9.878058433532715,
      "learning_rate": 3.24859401124791e-05,
      "loss": 0.5082,
      "step": 6075
    },
    {
      "epoch": 4.172366621067032,
      "grad_norm": 0.6408124566078186,
      "learning_rate": 3.239094087247302e-05,
      "loss": 0.2915,
      "step": 6100
    },
    {
      "epoch": 4.1894664842681255,
      "grad_norm": 4.155546188354492,
      "learning_rate": 3.229594163246694e-05,
      "loss": 0.421,
      "step": 6125
    },
    {
      "epoch": 4.20656634746922,
      "grad_norm": 2.4020016193389893,
      "learning_rate": 3.2200942392460865e-05,
      "loss": 0.3445,
      "step": 6150
    },
    {
      "epoch": 4.223666210670315,
      "grad_norm": 9.44356632232666,
      "learning_rate": 3.210594315245478e-05,
      "loss": 0.478,
      "step": 6175
    },
    {
      "epoch": 4.240766073871409,
      "grad_norm": 6.012132167816162,
      "learning_rate": 3.2010943912448705e-05,
      "loss": 0.4779,
      "step": 6200
    },
    {
      "epoch": 4.257865937072504,
      "grad_norm": 5.681918621063232,
      "learning_rate": 3.191594467244262e-05,
      "loss": 0.5116,
      "step": 6225
    },
    {
      "epoch": 4.2749658002735975,
      "grad_norm": 7.404222011566162,
      "learning_rate": 3.1820945432436546e-05,
      "loss": 0.3494,
      "step": 6250
    },
    {
      "epoch": 4.292065663474692,
      "grad_norm": 14.846484184265137,
      "learning_rate": 3.172594619243046e-05,
      "loss": 0.5149,
      "step": 6275
    },
    {
      "epoch": 4.309165526675787,
      "grad_norm": 1.3198221921920776,
      "learning_rate": 3.1630946952424386e-05,
      "loss": 0.4053,
      "step": 6300
    },
    {
      "epoch": 4.326265389876881,
      "grad_norm": 3.9458107948303223,
      "learning_rate": 3.15359477124183e-05,
      "loss": 0.4499,
      "step": 6325
    },
    {
      "epoch": 4.343365253077975,
      "grad_norm": 4.272012233734131,
      "learning_rate": 3.144094847241223e-05,
      "loss": 0.4298,
      "step": 6350
    },
    {
      "epoch": 4.3604651162790695,
      "grad_norm": 8.574115753173828,
      "learning_rate": 3.1345949232406144e-05,
      "loss": 0.3282,
      "step": 6375
    },
    {
      "epoch": 4.377564979480164,
      "grad_norm": 10.030281066894531,
      "learning_rate": 3.125094999240006e-05,
      "loss": 0.486,
      "step": 6400
    },
    {
      "epoch": 4.394664842681259,
      "grad_norm": 8.065329551696777,
      "learning_rate": 3.115595075239398e-05,
      "loss": 0.332,
      "step": 6425
    },
    {
      "epoch": 4.411764705882353,
      "grad_norm": 10.601835250854492,
      "learning_rate": 3.10609515123879e-05,
      "loss": 0.4674,
      "step": 6450
    },
    {
      "epoch": 4.428864569083447,
      "grad_norm": 4.035892486572266,
      "learning_rate": 3.096595227238182e-05,
      "loss": 0.5078,
      "step": 6475
    },
    {
      "epoch": 4.4459644322845415,
      "grad_norm": 6.8672685623168945,
      "learning_rate": 3.087095303237574e-05,
      "loss": 0.3255,
      "step": 6500
    },
    {
      "epoch": 4.463064295485636,
      "grad_norm": 4.926382541656494,
      "learning_rate": 3.077595379236966e-05,
      "loss": 0.4204,
      "step": 6525
    },
    {
      "epoch": 4.480164158686731,
      "grad_norm": 4.504785060882568,
      "learning_rate": 3.068095455236358e-05,
      "loss": 0.4257,
      "step": 6550
    },
    {
      "epoch": 4.497264021887825,
      "grad_norm": 10.662145614624023,
      "learning_rate": 3.05859553123575e-05,
      "loss": 0.4991,
      "step": 6575
    },
    {
      "epoch": 4.514363885088919,
      "grad_norm": 4.892814636230469,
      "learning_rate": 3.0490956072351423e-05,
      "loss": 0.3588,
      "step": 6600
    },
    {
      "epoch": 4.5314637482900135,
      "grad_norm": 2.461637258529663,
      "learning_rate": 3.039595683234534e-05,
      "loss": 0.3509,
      "step": 6625
    },
    {
      "epoch": 4.548563611491108,
      "grad_norm": 0.9665390253067017,
      "learning_rate": 3.0300957592339263e-05,
      "loss": 0.4587,
      "step": 6650
    },
    {
      "epoch": 4.565663474692203,
      "grad_norm": 5.2920002937316895,
      "learning_rate": 3.020595835233318e-05,
      "loss": 0.3853,
      "step": 6675
    },
    {
      "epoch": 4.582763337893297,
      "grad_norm": 6.542552471160889,
      "learning_rate": 3.0110959112327104e-05,
      "loss": 0.475,
      "step": 6700
    },
    {
      "epoch": 4.599863201094391,
      "grad_norm": 3.7444801330566406,
      "learning_rate": 3.0015959872321024e-05,
      "loss": 0.3638,
      "step": 6725
    },
    {
      "epoch": 4.6169630642954855,
      "grad_norm": 6.542828559875488,
      "learning_rate": 2.992096063231494e-05,
      "loss": 0.4892,
      "step": 6750
    },
    {
      "epoch": 4.63406292749658,
      "grad_norm": 1.4131304025650024,
      "learning_rate": 2.9825961392308865e-05,
      "loss": 0.4617,
      "step": 6775
    },
    {
      "epoch": 4.651162790697675,
      "grad_norm": 2.6974217891693115,
      "learning_rate": 2.973096215230278e-05,
      "loss": 0.3848,
      "step": 6800
    },
    {
      "epoch": 4.668262653898768,
      "grad_norm": 5.639090061187744,
      "learning_rate": 2.9635962912296705e-05,
      "loss": 0.3584,
      "step": 6825
    },
    {
      "epoch": 4.685362517099863,
      "grad_norm": 11.76826000213623,
      "learning_rate": 2.9540963672290622e-05,
      "loss": 0.4498,
      "step": 6850
    },
    {
      "epoch": 4.7024623803009575,
      "grad_norm": 3.208690643310547,
      "learning_rate": 2.9445964432284546e-05,
      "loss": 0.3597,
      "step": 6875
    },
    {
      "epoch": 4.719562243502052,
      "grad_norm": 5.359434604644775,
      "learning_rate": 2.9350965192278463e-05,
      "loss": 0.3621,
      "step": 6900
    },
    {
      "epoch": 4.736662106703147,
      "grad_norm": 6.385953426361084,
      "learning_rate": 2.9255965952272386e-05,
      "loss": 0.4097,
      "step": 6925
    },
    {
      "epoch": 4.75376196990424,
      "grad_norm": 10.398621559143066,
      "learning_rate": 2.9160966712266303e-05,
      "loss": 0.3647,
      "step": 6950
    },
    {
      "epoch": 4.770861833105335,
      "grad_norm": 2.8369972705841064,
      "learning_rate": 2.9065967472260224e-05,
      "loss": 0.3396,
      "step": 6975
    },
    {
      "epoch": 4.7879616963064295,
      "grad_norm": 6.925849437713623,
      "learning_rate": 2.897096823225414e-05,
      "loss": 0.3689,
      "step": 7000
    },
    {
      "epoch": 4.805061559507524,
      "grad_norm": 3.137173652648926,
      "learning_rate": 2.8875968992248064e-05,
      "loss": 0.5454,
      "step": 7025
    },
    {
      "epoch": 4.822161422708619,
      "grad_norm": 12.329541206359863,
      "learning_rate": 2.878096975224198e-05,
      "loss": 0.417,
      "step": 7050
    },
    {
      "epoch": 4.839261285909712,
      "grad_norm": 1.9504649639129639,
      "learning_rate": 2.8685970512235905e-05,
      "loss": 0.3552,
      "step": 7075
    },
    {
      "epoch": 4.856361149110807,
      "grad_norm": 6.352153778076172,
      "learning_rate": 2.859097127222982e-05,
      "loss": 0.4427,
      "step": 7100
    },
    {
      "epoch": 4.8734610123119015,
      "grad_norm": 5.652337074279785,
      "learning_rate": 2.8495972032223745e-05,
      "loss": 0.4454,
      "step": 7125
    },
    {
      "epoch": 4.890560875512996,
      "grad_norm": 4.495980262756348,
      "learning_rate": 2.8400972792217662e-05,
      "loss": 0.4396,
      "step": 7150
    },
    {
      "epoch": 4.907660738714091,
      "grad_norm": 9.805567741394043,
      "learning_rate": 2.8305973552211586e-05,
      "loss": 0.4436,
      "step": 7175
    },
    {
      "epoch": 4.924760601915184,
      "grad_norm": 2.151742696762085,
      "learning_rate": 2.8210974312205503e-05,
      "loss": 0.4238,
      "step": 7200
    },
    {
      "epoch": 4.941860465116279,
      "grad_norm": 4.97649621963501,
      "learning_rate": 2.8115975072199423e-05,
      "loss": 0.4693,
      "step": 7225
    },
    {
      "epoch": 4.9589603283173735,
      "grad_norm": 2.5066914558410645,
      "learning_rate": 2.802097583219334e-05,
      "loss": 0.4568,
      "step": 7250
    },
    {
      "epoch": 4.976060191518468,
      "grad_norm": 13.676875114440918,
      "learning_rate": 2.7925976592187263e-05,
      "loss": 0.3527,
      "step": 7275
    },
    {
      "epoch": 4.993160054719562,
      "grad_norm": 3.240598678588867,
      "learning_rate": 2.783097735218118e-05,
      "loss": 0.4515,
      "step": 7300
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9166951177876408,
      "eval_f1_macro": 0.7653415866627981,
      "eval_f1_micro": 0.9166951177876408,
      "eval_f1_weighted": 0.9141778991500009,
      "eval_loss": 0.24330385029315948,
      "eval_precision_macro": 0.794330339771912,
      "eval_precision_micro": 0.9166951177876408,
      "eval_precision_weighted": 0.9162095802352392,
      "eval_recall_macro": 0.7496536933673316,
      "eval_recall_micro": 0.9166951177876408,
      "eval_recall_weighted": 0.9166951177876408,
      "eval_runtime": 29.505,
      "eval_samples_per_second": 99.271,
      "eval_steps_per_second": 6.236,
      "step": 7310
    },
    {
      "epoch": 5.010259917920656,
      "grad_norm": 3.2240583896636963,
      "learning_rate": 2.7735978112175104e-05,
      "loss": 0.406,
      "step": 7325
    },
    {
      "epoch": 5.027359781121751,
      "grad_norm": 1.1182893514633179,
      "learning_rate": 2.7640978872169028e-05,
      "loss": 0.4151,
      "step": 7350
    },
    {
      "epoch": 5.0444596443228455,
      "grad_norm": 3.5205605030059814,
      "learning_rate": 2.7545979632162945e-05,
      "loss": 0.4349,
      "step": 7375
    },
    {
      "epoch": 5.06155950752394,
      "grad_norm": 5.898115634918213,
      "learning_rate": 2.7450980392156865e-05,
      "loss": 0.4127,
      "step": 7400
    },
    {
      "epoch": 5.078659370725034,
      "grad_norm": 2.958627700805664,
      "learning_rate": 2.7355981152150785e-05,
      "loss": 0.4479,
      "step": 7425
    },
    {
      "epoch": 5.095759233926128,
      "grad_norm": 4.672809600830078,
      "learning_rate": 2.7260981912144705e-05,
      "loss": 0.4976,
      "step": 7450
    },
    {
      "epoch": 5.112859097127223,
      "grad_norm": 8.304091453552246,
      "learning_rate": 2.7165982672138622e-05,
      "loss": 0.4065,
      "step": 7475
    },
    {
      "epoch": 5.1299589603283176,
      "grad_norm": 2.276463270187378,
      "learning_rate": 2.7070983432132546e-05,
      "loss": 0.3508,
      "step": 7500
    },
    {
      "epoch": 5.147058823529412,
      "grad_norm": 0.10079041868448257,
      "learning_rate": 2.6975984192126463e-05,
      "loss": 0.4274,
      "step": 7525
    },
    {
      "epoch": 5.164158686730506,
      "grad_norm": 5.3365936279296875,
      "learning_rate": 2.6880984952120387e-05,
      "loss": 0.2555,
      "step": 7550
    },
    {
      "epoch": 5.1812585499316,
      "grad_norm": 0.33459940552711487,
      "learning_rate": 2.6785985712114303e-05,
      "loss": 0.4309,
      "step": 7575
    },
    {
      "epoch": 5.198358413132695,
      "grad_norm": 6.256367206573486,
      "learning_rate": 2.6690986472108227e-05,
      "loss": 0.3263,
      "step": 7600
    },
    {
      "epoch": 5.2154582763337896,
      "grad_norm": 2.592665672302246,
      "learning_rate": 2.6595987232102144e-05,
      "loss": 0.4089,
      "step": 7625
    },
    {
      "epoch": 5.232558139534884,
      "grad_norm": 5.69423246383667,
      "learning_rate": 2.6500987992096064e-05,
      "loss": 0.4288,
      "step": 7650
    },
    {
      "epoch": 5.249658002735978,
      "grad_norm": 7.463606834411621,
      "learning_rate": 2.6405988752089985e-05,
      "loss": 0.35,
      "step": 7675
    },
    {
      "epoch": 5.266757865937072,
      "grad_norm": 4.466919898986816,
      "learning_rate": 2.6310989512083905e-05,
      "loss": 0.4364,
      "step": 7700
    },
    {
      "epoch": 5.283857729138167,
      "grad_norm": 5.60076904296875,
      "learning_rate": 2.621599027207782e-05,
      "loss": 0.2817,
      "step": 7725
    },
    {
      "epoch": 5.300957592339262,
      "grad_norm": 1.8708001375198364,
      "learning_rate": 2.6120991032071745e-05,
      "loss": 0.3567,
      "step": 7750
    },
    {
      "epoch": 5.318057455540355,
      "grad_norm": 4.970040798187256,
      "learning_rate": 2.6025991792065662e-05,
      "loss": 0.2724,
      "step": 7775
    },
    {
      "epoch": 5.33515731874145,
      "grad_norm": 1.4727894067764282,
      "learning_rate": 2.5930992552059586e-05,
      "loss": 0.3978,
      "step": 7800
    },
    {
      "epoch": 5.352257181942544,
      "grad_norm": 4.697504043579102,
      "learning_rate": 2.5835993312053503e-05,
      "loss": 0.447,
      "step": 7825
    },
    {
      "epoch": 5.369357045143639,
      "grad_norm": 2.3965351581573486,
      "learning_rate": 2.5740994072047426e-05,
      "loss": 0.3596,
      "step": 7850
    },
    {
      "epoch": 5.386456908344734,
      "grad_norm": 8.727534294128418,
      "learning_rate": 2.5645994832041343e-05,
      "loss": 0.3787,
      "step": 7875
    },
    {
      "epoch": 5.403556771545827,
      "grad_norm": 5.903003692626953,
      "learning_rate": 2.5550995592035264e-05,
      "loss": 0.3899,
      "step": 7900
    },
    {
      "epoch": 5.420656634746922,
      "grad_norm": 4.570006847381592,
      "learning_rate": 2.5455996352029187e-05,
      "loss": 0.4558,
      "step": 7925
    },
    {
      "epoch": 5.437756497948016,
      "grad_norm": 0.9773057699203491,
      "learning_rate": 2.5360997112023104e-05,
      "loss": 0.3345,
      "step": 7950
    },
    {
      "epoch": 5.454856361149111,
      "grad_norm": 3.641374349594116,
      "learning_rate": 2.5265997872017028e-05,
      "loss": 0.4155,
      "step": 7975
    },
    {
      "epoch": 5.471956224350206,
      "grad_norm": 5.382412433624268,
      "learning_rate": 2.5170998632010945e-05,
      "loss": 0.4111,
      "step": 8000
    },
    {
      "epoch": 5.489056087551299,
      "grad_norm": 2.235764980316162,
      "learning_rate": 2.507599939200487e-05,
      "loss": 0.3047,
      "step": 8025
    },
    {
      "epoch": 5.506155950752394,
      "grad_norm": 3.8063552379608154,
      "learning_rate": 2.4981000151998785e-05,
      "loss": 0.4702,
      "step": 8050
    },
    {
      "epoch": 5.523255813953488,
      "grad_norm": 3.573058843612671,
      "learning_rate": 2.4886000911992706e-05,
      "loss": 0.4619,
      "step": 8075
    },
    {
      "epoch": 5.540355677154583,
      "grad_norm": 7.515729904174805,
      "learning_rate": 2.4791001671986626e-05,
      "loss": 0.3179,
      "step": 8100
    },
    {
      "epoch": 5.557455540355678,
      "grad_norm": 1.7556642293930054,
      "learning_rate": 2.4696002431980546e-05,
      "loss": 0.3382,
      "step": 8125
    },
    {
      "epoch": 5.574555403556771,
      "grad_norm": 2.0541181564331055,
      "learning_rate": 2.4601003191974463e-05,
      "loss": 0.4592,
      "step": 8150
    },
    {
      "epoch": 5.591655266757866,
      "grad_norm": 2.7318685054779053,
      "learning_rate": 2.4506003951968383e-05,
      "loss": 0.4169,
      "step": 8175
    },
    {
      "epoch": 5.60875512995896,
      "grad_norm": 8.095924377441406,
      "learning_rate": 2.4411004711962304e-05,
      "loss": 0.4333,
      "step": 8200
    },
    {
      "epoch": 5.625854993160055,
      "grad_norm": 2.2528128623962402,
      "learning_rate": 2.4316005471956224e-05,
      "loss": 0.402,
      "step": 8225
    },
    {
      "epoch": 5.642954856361149,
      "grad_norm": 10.526260375976562,
      "learning_rate": 2.4221006231950147e-05,
      "loss": 0.3477,
      "step": 8250
    },
    {
      "epoch": 5.660054719562243,
      "grad_norm": 5.1542487144470215,
      "learning_rate": 2.4126006991944068e-05,
      "loss": 0.3982,
      "step": 8275
    },
    {
      "epoch": 5.677154582763338,
      "grad_norm": 5.238847255706787,
      "learning_rate": 2.4031007751937988e-05,
      "loss": 0.416,
      "step": 8300
    },
    {
      "epoch": 5.694254445964432,
      "grad_norm": 8.261098861694336,
      "learning_rate": 2.3936008511931908e-05,
      "loss": 0.2776,
      "step": 8325
    },
    {
      "epoch": 5.711354309165527,
      "grad_norm": 6.928776741027832,
      "learning_rate": 2.3841009271925825e-05,
      "loss": 0.3845,
      "step": 8350
    },
    {
      "epoch": 5.728454172366621,
      "grad_norm": 9.97043514251709,
      "learning_rate": 2.3746010031919745e-05,
      "loss": 0.4559,
      "step": 8375
    },
    {
      "epoch": 5.745554035567715,
      "grad_norm": 1.063759446144104,
      "learning_rate": 2.3651010791913666e-05,
      "loss": 0.2716,
      "step": 8400
    },
    {
      "epoch": 5.76265389876881,
      "grad_norm": 2.098356246948242,
      "learning_rate": 2.3556011551907586e-05,
      "loss": 0.3414,
      "step": 8425
    },
    {
      "epoch": 5.779753761969904,
      "grad_norm": 5.218485355377197,
      "learning_rate": 2.3461012311901506e-05,
      "loss": 0.3264,
      "step": 8450
    },
    {
      "epoch": 5.796853625170999,
      "grad_norm": 5.245093822479248,
      "learning_rate": 2.3366013071895427e-05,
      "loss": 0.3575,
      "step": 8475
    },
    {
      "epoch": 5.813953488372093,
      "grad_norm": 2.249113082885742,
      "learning_rate": 2.3271013831889347e-05,
      "loss": 0.3042,
      "step": 8500
    },
    {
      "epoch": 5.831053351573187,
      "grad_norm": 8.331680297851562,
      "learning_rate": 2.3176014591883267e-05,
      "loss": 0.349,
      "step": 8525
    },
    {
      "epoch": 5.848153214774282,
      "grad_norm": 5.0042805671691895,
      "learning_rate": 2.3081015351877187e-05,
      "loss": 0.4322,
      "step": 8550
    },
    {
      "epoch": 5.865253077975376,
      "grad_norm": 15.829472541809082,
      "learning_rate": 2.2986016111871104e-05,
      "loss": 0.5301,
      "step": 8575
    },
    {
      "epoch": 5.882352941176471,
      "grad_norm": 5.099222660064697,
      "learning_rate": 2.2891016871865025e-05,
      "loss": 0.4318,
      "step": 8600
    },
    {
      "epoch": 5.899452804377565,
      "grad_norm": 3.355231285095215,
      "learning_rate": 2.2796017631858945e-05,
      "loss": 0.3114,
      "step": 8625
    },
    {
      "epoch": 5.916552667578659,
      "grad_norm": 5.962894916534424,
      "learning_rate": 2.2701018391852865e-05,
      "loss": 0.389,
      "step": 8650
    },
    {
      "epoch": 5.933652530779754,
      "grad_norm": 0.5913941860198975,
      "learning_rate": 2.2606019151846785e-05,
      "loss": 0.3845,
      "step": 8675
    },
    {
      "epoch": 5.950752393980848,
      "grad_norm": 2.7873880863189697,
      "learning_rate": 2.2511019911840706e-05,
      "loss": 0.3045,
      "step": 8700
    },
    {
      "epoch": 5.967852257181942,
      "grad_norm": 6.107980728149414,
      "learning_rate": 2.2416020671834626e-05,
      "loss": 0.4016,
      "step": 8725
    },
    {
      "epoch": 5.984952120383037,
      "grad_norm": 2.5966835021972656,
      "learning_rate": 2.2321021431828546e-05,
      "loss": 0.3413,
      "step": 8750
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9095254353021509,
      "eval_f1_macro": 0.787176394659271,
      "eval_f1_micro": 0.9095254353021509,
      "eval_f1_weighted": 0.9072901390204282,
      "eval_loss": 0.29091742634773254,
      "eval_precision_macro": 0.8953297645243259,
      "eval_precision_micro": 0.9095254353021509,
      "eval_precision_weighted": 0.9174594106183492,
      "eval_recall_macro": 0.7531104129954299,
      "eval_recall_micro": 0.9095254353021509,
      "eval_recall_weighted": 0.9095254353021509,
      "eval_runtime": 29.8878,
      "eval_samples_per_second": 98.0,
      "eval_steps_per_second": 6.156,
      "step": 8772
    },
    {
      "epoch": 6.002051983584131,
      "grad_norm": 6.595887660980225,
      "learning_rate": 2.2226022191822466e-05,
      "loss": 0.3606,
      "step": 8775
    },
    {
      "epoch": 6.019151846785226,
      "grad_norm": 6.390087127685547,
      "learning_rate": 2.2131022951816387e-05,
      "loss": 0.4026,
      "step": 8800
    },
    {
      "epoch": 6.03625170998632,
      "grad_norm": 6.15175199508667,
      "learning_rate": 2.2036023711810304e-05,
      "loss": 0.3281,
      "step": 8825
    },
    {
      "epoch": 6.053351573187414,
      "grad_norm": 18.778564453125,
      "learning_rate": 2.1941024471804227e-05,
      "loss": 0.4304,
      "step": 8850
    },
    {
      "epoch": 6.070451436388509,
      "grad_norm": 9.964593887329102,
      "learning_rate": 2.1846025231798148e-05,
      "loss": 0.2677,
      "step": 8875
    },
    {
      "epoch": 6.087551299589603,
      "grad_norm": 10.952362060546875,
      "learning_rate": 2.1751025991792068e-05,
      "loss": 0.5143,
      "step": 8900
    },
    {
      "epoch": 6.104651162790698,
      "grad_norm": 2.978898048400879,
      "learning_rate": 2.1656026751785988e-05,
      "loss": 0.3024,
      "step": 8925
    },
    {
      "epoch": 6.121751025991792,
      "grad_norm": 3.3607687950134277,
      "learning_rate": 2.156102751177991e-05,
      "loss": 0.2684,
      "step": 8950
    },
    {
      "epoch": 6.138850889192886,
      "grad_norm": 1.9624247550964355,
      "learning_rate": 2.146602827177383e-05,
      "loss": 0.3062,
      "step": 8975
    },
    {
      "epoch": 6.155950752393981,
      "grad_norm": 1.7253330945968628,
      "learning_rate": 2.137102903176775e-05,
      "loss": 0.3242,
      "step": 9000
    },
    {
      "epoch": 6.173050615595075,
      "grad_norm": 2.6368486881256104,
      "learning_rate": 2.1276029791761666e-05,
      "loss": 0.3678,
      "step": 9025
    },
    {
      "epoch": 6.19015047879617,
      "grad_norm": 7.951999664306641,
      "learning_rate": 2.1181030551755586e-05,
      "loss": 0.3242,
      "step": 9050
    },
    {
      "epoch": 6.207250341997264,
      "grad_norm": 13.932353019714355,
      "learning_rate": 2.1086031311749506e-05,
      "loss": 0.244,
      "step": 9075
    },
    {
      "epoch": 6.224350205198358,
      "grad_norm": 0.4225591719150543,
      "learning_rate": 2.0991032071743427e-05,
      "loss": 0.503,
      "step": 9100
    },
    {
      "epoch": 6.241450068399453,
      "grad_norm": 5.91692590713501,
      "learning_rate": 2.0896032831737347e-05,
      "loss": 0.3721,
      "step": 9125
    },
    {
      "epoch": 6.258549931600547,
      "grad_norm": 7.2567009925842285,
      "learning_rate": 2.0801033591731267e-05,
      "loss": 0.3032,
      "step": 9150
    },
    {
      "epoch": 6.275649794801642,
      "grad_norm": 12.250884056091309,
      "learning_rate": 2.0706034351725188e-05,
      "loss": 0.4468,
      "step": 9175
    },
    {
      "epoch": 6.292749658002736,
      "grad_norm": 5.396862030029297,
      "learning_rate": 2.0611035111719108e-05,
      "loss": 0.357,
      "step": 9200
    },
    {
      "epoch": 6.30984952120383,
      "grad_norm": 0.8270015120506287,
      "learning_rate": 2.0516035871713028e-05,
      "loss": 0.2127,
      "step": 9225
    },
    {
      "epoch": 6.326949384404925,
      "grad_norm": 6.996820449829102,
      "learning_rate": 2.042103663170695e-05,
      "loss": 0.4843,
      "step": 9250
    },
    {
      "epoch": 6.344049247606019,
      "grad_norm": 3.144995927810669,
      "learning_rate": 2.0326037391700865e-05,
      "loss": 0.4409,
      "step": 9275
    },
    {
      "epoch": 6.361149110807114,
      "grad_norm": 10.23337459564209,
      "learning_rate": 2.0231038151694785e-05,
      "loss": 0.406,
      "step": 9300
    },
    {
      "epoch": 6.378248974008208,
      "grad_norm": 3.1300835609436035,
      "learning_rate": 2.0136038911688706e-05,
      "loss": 0.2881,
      "step": 9325
    },
    {
      "epoch": 6.395348837209302,
      "grad_norm": 3.322756290435791,
      "learning_rate": 2.0041039671682626e-05,
      "loss": 0.2679,
      "step": 9350
    },
    {
      "epoch": 6.412448700410397,
      "grad_norm": 7.402932643890381,
      "learning_rate": 1.9946040431676546e-05,
      "loss": 0.3229,
      "step": 9375
    },
    {
      "epoch": 6.429548563611491,
      "grad_norm": 2.389960289001465,
      "learning_rate": 1.9851041191670467e-05,
      "loss": 0.3176,
      "step": 9400
    },
    {
      "epoch": 6.446648426812586,
      "grad_norm": 4.060113906860352,
      "learning_rate": 1.9756041951664387e-05,
      "loss": 0.4683,
      "step": 9425
    },
    {
      "epoch": 6.46374829001368,
      "grad_norm": 3.9129064083099365,
      "learning_rate": 1.966104271165831e-05,
      "loss": 0.3247,
      "step": 9450
    },
    {
      "epoch": 6.480848153214774,
      "grad_norm": 7.171699523925781,
      "learning_rate": 1.9566043471652227e-05,
      "loss": 0.2698,
      "step": 9475
    },
    {
      "epoch": 6.497948016415869,
      "grad_norm": 5.123276233673096,
      "learning_rate": 1.9471044231646148e-05,
      "loss": 0.3883,
      "step": 9500
    },
    {
      "epoch": 6.515047879616963,
      "grad_norm": 1.7513083219528198,
      "learning_rate": 1.9376044991640068e-05,
      "loss": 0.3488,
      "step": 9525
    },
    {
      "epoch": 6.532147742818058,
      "grad_norm": 3.334690809249878,
      "learning_rate": 1.9281045751633988e-05,
      "loss": 0.4843,
      "step": 9550
    },
    {
      "epoch": 6.549247606019152,
      "grad_norm": 2.4918205738067627,
      "learning_rate": 1.918604651162791e-05,
      "loss": 0.4017,
      "step": 9575
    },
    {
      "epoch": 6.566347469220246,
      "grad_norm": 3.277615785598755,
      "learning_rate": 1.909104727162183e-05,
      "loss": 0.3807,
      "step": 9600
    },
    {
      "epoch": 6.583447332421341,
      "grad_norm": 0.09262839704751968,
      "learning_rate": 1.899604803161575e-05,
      "loss": 0.3623,
      "step": 9625
    },
    {
      "epoch": 6.600547195622435,
      "grad_norm": 9.153847694396973,
      "learning_rate": 1.890104879160967e-05,
      "loss": 0.3903,
      "step": 9650
    },
    {
      "epoch": 6.617647058823529,
      "grad_norm": 4.24110221862793,
      "learning_rate": 1.880604955160359e-05,
      "loss": 0.3078,
      "step": 9675
    },
    {
      "epoch": 6.634746922024624,
      "grad_norm": 5.625593185424805,
      "learning_rate": 1.871105031159751e-05,
      "loss": 0.3915,
      "step": 9700
    },
    {
      "epoch": 6.651846785225718,
      "grad_norm": 6.439937591552734,
      "learning_rate": 1.8616051071591427e-05,
      "loss": 0.4783,
      "step": 9725
    },
    {
      "epoch": 6.668946648426813,
      "grad_norm": 6.7055559158325195,
      "learning_rate": 1.8521051831585347e-05,
      "loss": 0.3364,
      "step": 9750
    },
    {
      "epoch": 6.686046511627907,
      "grad_norm": 2.9428536891937256,
      "learning_rate": 1.8426052591579267e-05,
      "loss": 0.3409,
      "step": 9775
    },
    {
      "epoch": 6.703146374829001,
      "grad_norm": 3.669163942337036,
      "learning_rate": 1.8331053351573188e-05,
      "loss": 0.2637,
      "step": 9800
    },
    {
      "epoch": 6.720246238030096,
      "grad_norm": 2.797004461288452,
      "learning_rate": 1.8236054111567108e-05,
      "loss": 0.3721,
      "step": 9825
    },
    {
      "epoch": 6.73734610123119,
      "grad_norm": 1.3925397396087646,
      "learning_rate": 1.8141054871561028e-05,
      "loss": 0.3227,
      "step": 9850
    },
    {
      "epoch": 6.754445964432285,
      "grad_norm": 9.197000503540039,
      "learning_rate": 1.804605563155495e-05,
      "loss": 0.3789,
      "step": 9875
    },
    {
      "epoch": 6.771545827633379,
      "grad_norm": 7.823335647583008,
      "learning_rate": 1.795105639154887e-05,
      "loss": 0.2718,
      "step": 9900
    },
    {
      "epoch": 6.788645690834473,
      "grad_norm": 1.3368386030197144,
      "learning_rate": 1.785605715154279e-05,
      "loss": 0.3345,
      "step": 9925
    },
    {
      "epoch": 6.805745554035568,
      "grad_norm": 4.986422061920166,
      "learning_rate": 1.776105791153671e-05,
      "loss": 0.3514,
      "step": 9950
    },
    {
      "epoch": 6.822845417236662,
      "grad_norm": 3.647663116455078,
      "learning_rate": 1.7666058671530626e-05,
      "loss": 0.3476,
      "step": 9975
    },
    {
      "epoch": 6.839945280437757,
      "grad_norm": 10.99355697631836,
      "learning_rate": 1.7571059431524546e-05,
      "loss": 0.4088,
      "step": 10000
    },
    {
      "epoch": 6.857045143638851,
      "grad_norm": 6.52988862991333,
      "learning_rate": 1.7476060191518467e-05,
      "loss": 0.3857,
      "step": 10025
    },
    {
      "epoch": 6.874145006839945,
      "grad_norm": 4.713931560516357,
      "learning_rate": 1.738106095151239e-05,
      "loss": 0.5173,
      "step": 10050
    },
    {
      "epoch": 6.89124487004104,
      "grad_norm": 1.7083477973937988,
      "learning_rate": 1.728606171150631e-05,
      "loss": 0.2722,
      "step": 10075
    },
    {
      "epoch": 6.908344733242134,
      "grad_norm": 2.890033483505249,
      "learning_rate": 1.719106247150023e-05,
      "loss": 0.346,
      "step": 10100
    },
    {
      "epoch": 6.925444596443229,
      "grad_norm": 2.318939208984375,
      "learning_rate": 1.709606323149415e-05,
      "loss": 0.3563,
      "step": 10125
    },
    {
      "epoch": 6.942544459644322,
      "grad_norm": 1.522857666015625,
      "learning_rate": 1.700106399148807e-05,
      "loss": 0.4441,
      "step": 10150
    },
    {
      "epoch": 6.959644322845417,
      "grad_norm": 4.418458938598633,
      "learning_rate": 1.690986472108223e-05,
      "loss": 0.253,
      "step": 10175
    },
    {
      "epoch": 6.976744186046512,
      "grad_norm": 7.665693759918213,
      "learning_rate": 1.6814865481076152e-05,
      "loss": 0.329,
      "step": 10200
    },
    {
      "epoch": 6.993844049247606,
      "grad_norm": 2.388054609298706,
      "learning_rate": 1.6719866241070072e-05,
      "loss": 0.3777,
      "step": 10225
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9214749061113008,
      "eval_f1_macro": 0.833050479179196,
      "eval_f1_micro": 0.9214749061113008,
      "eval_f1_weighted": 0.9214057890123231,
      "eval_loss": 0.23535355925559998,
      "eval_precision_macro": 0.8898753259090769,
      "eval_precision_micro": 0.9214749061113008,
      "eval_precision_weighted": 0.9262844759199463,
      "eval_recall_macro": 0.7987766214501736,
      "eval_recall_micro": 0.9214749061113008,
      "eval_recall_weighted": 0.9214749061113008,
      "eval_runtime": 30.0564,
      "eval_samples_per_second": 97.45,
      "eval_steps_per_second": 6.122,
      "step": 10234
    }
  ],
  "logging_steps": 25,
  "max_steps": 14620,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 2.242292315610641e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}