{
  "best_metric": 0.967008960660055,
  "best_model_checkpoint": "/kaggle/working/mmoe_vit_results/checkpoint-24012",
  "epoch": 8.0,
  "eval_steps": 500,
  "global_step": 32016,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.12493753123438281,
      "grad_norm": 57.01617431640625,
      "learning_rate": 5e-05,
      "loss": 16.5766,
      "step": 500
    },
    {
      "epoch": 0.24987506246876562,
      "grad_norm": 36.94090270996094,
      "learning_rate": 4.9367408906882594e-05,
      "loss": 10.9165,
      "step": 1000
    },
    {
      "epoch": 0.3748125937031484,
      "grad_norm": 30.66377067565918,
      "learning_rate": 4.8734817813765186e-05,
      "loss": 9.8167,
      "step": 1500
    },
    {
      "epoch": 0.49975012493753124,
      "grad_norm": 42.035491943359375,
      "learning_rate": 4.810222672064778e-05,
      "loss": 9.3138,
      "step": 2000
    },
    {
      "epoch": 0.624687656171914,
      "grad_norm": 51.97323989868164,
      "learning_rate": 4.746963562753037e-05,
      "loss": 8.8995,
      "step": 2500
    },
    {
      "epoch": 0.7496251874062968,
      "grad_norm": 30.12773895263672,
      "learning_rate": 4.683704453441296e-05,
      "loss": 8.6501,
      "step": 3000
    },
    {
      "epoch": 0.8745627186406797,
      "grad_norm": 35.8332405090332,
      "learning_rate": 4.6204453441295545e-05,
      "loss": 8.4094,
      "step": 3500
    },
    {
      "epoch": 0.9995002498750625,
      "grad_norm": 33.1800422668457,
      "learning_rate": 4.557186234817814e-05,
      "loss": 8.2048,
      "step": 4000
    },
    {
      "epoch": 1.0,
      "eval_category_macro_f1": 0.7056917901689866,
      "eval_category_precision": 0.959686334325201,
      "eval_category_recall": 0.9583528164843044,
      "eval_category_weighted_f1": 0.9569810583065036,
      "eval_color_macro_f1": 0.2688303275737633,
      "eval_color_precision": 0.7518445947091915,
      "eval_color_recall": 0.7785778858443219,
      "eval_color_weighted_f1": 0.7567459270340884,
      "eval_gender_macro_f1": 0.61691692266805,
      "eval_gender_precision": 0.924795901785245,
      "eval_gender_recall": 0.9251535272296518,
      "eval_gender_weighted_f1": 0.9247660825888746,
      "eval_loss": 7.85603666305542,
      "eval_material_macro_f1": 0.20719595562175594,
      "eval_material_precision": 0.5500768038697358,
      "eval_material_recall": 0.6039153340191402,
      "eval_material_weighted_f1": 0.5466956697100748,
      "eval_neck_macro_f1": 0.19113506916293588,
      "eval_neck_precision": 0.7622121166320025,
      "eval_neck_recall": 0.7717509897440693,
      "eval_neck_weighted_f1": 0.7581205711853191,
      "eval_pattern_macro_f1": 0.07869956737005358,
      "eval_pattern_precision": 0.5012835314156145,
      "eval_pattern_recall": 0.5952804015087753,
      "eval_pattern_weighted_f1": 0.5053200171793594,
      "eval_product_type_macro_f1": 0.5397272380061237,
      "eval_product_type_precision": 0.659065702055643,
      "eval_product_type_recall": 0.6470899965709654,
      "eval_product_type_weighted_f1": 0.6335857444723578,
      "eval_runtime": 319.4683,
      "eval_samples_per_second": 100.414,
      "eval_sleeve_macro_f1": 0.32806975143028533,
      "eval_sleeve_precision": 0.8325346017171937,
      "eval_sleeve_recall": 0.8443530035225537,
      "eval_sleeve_weighted_f1": 0.8322418774164811,
      "eval_steps_per_second": 1.571,
      "eval_style_macro_f1": 0.14638281117715193,
      "eval_style_precision": 0.5779767242059484,
      "eval_style_recall": 0.6343402225755167,
      "eval_style_weighted_f1": 0.5854913490611046,
      "step": 4002
    },
    {
      "epoch": 1.1244377811094453,
      "grad_norm": 29.43435287475586,
      "learning_rate": 4.4939271255060735e-05,
      "loss": 7.7873,
      "step": 4500
    },
    {
      "epoch": 1.249375312343828,
      "grad_norm": 36.80842971801758,
      "learning_rate": 4.430668016194332e-05,
      "loss": 7.7689,
      "step": 5000
    },
    {
      "epoch": 1.3743128435782108,
      "grad_norm": 29.0100040435791,
      "learning_rate": 4.367408906882591e-05,
      "loss": 7.6509,
      "step": 5500
    },
    {
      "epoch": 1.4992503748125938,
      "grad_norm": 43.44118118286133,
      "learning_rate": 4.3041497975708504e-05,
      "loss": 7.6131,
      "step": 6000
    },
    {
      "epoch": 1.6241879060469766,
      "grad_norm": 27.290224075317383,
      "learning_rate": 4.2408906882591095e-05,
      "loss": 7.4747,
      "step": 6500
    },
    {
      "epoch": 1.7491254372813594,
      "grad_norm": 32.5240364074707,
      "learning_rate": 4.177631578947369e-05,
      "loss": 7.307,
      "step": 7000
    },
    {
      "epoch": 1.8740629685157422,
      "grad_norm": 30.75071907043457,
      "learning_rate": 4.114372469635628e-05,
      "loss": 7.2819,
      "step": 7500
    },
    {
      "epoch": 1.999000499750125,
      "grad_norm": 38.38020706176758,
      "learning_rate": 4.051113360323887e-05,
      "loss": 7.2531,
      "step": 8000
    },
    {
      "epoch": 2.0,
      "eval_category_macro_f1": 0.7602602884544463,
      "eval_category_precision": 0.9629087982936062,
      "eval_category_recall": 0.9625300040524954,
      "eval_category_weighted_f1": 0.9614384751352656,
      "eval_color_macro_f1": 0.32999615687202,
      "eval_color_precision": 0.7657516039866678,
      "eval_color_recall": 0.7893637582218898,
      "eval_color_weighted_f1": 0.7702899342050024,
      "eval_gender_macro_f1": 0.65351315015835,
      "eval_gender_precision": 0.9344584897700611,
      "eval_gender_recall": 0.9329779606596216,
      "eval_gender_weighted_f1": 0.9321715496558077,
      "eval_loss": 7.258897304534912,
      "eval_material_macro_f1": 0.2746169076986001,
      "eval_material_precision": 0.5915997907196028,
      "eval_material_recall": 0.6195953739206335,
      "eval_material_weighted_f1": 0.5668523075222497,
      "eval_neck_macro_f1": 0.20049848494224712,
      "eval_neck_precision": 0.7708207481094091,
      "eval_neck_recall": 0.7876492409364382,
      "eval_neck_weighted_f1": 0.7684817623431263,
      "eval_pattern_macro_f1": 0.11355145063472773,
      "eval_pattern_precision": 0.5245684834877903,
      "eval_pattern_recall": 0.6085289441690825,
      "eval_pattern_weighted_f1": 0.5261034858766623,
      "eval_product_type_macro_f1": 0.613166417428187,
      "eval_product_type_precision": 0.6931319121735884,
      "eval_product_type_recall": 0.6792917484959008,
      "eval_product_type_weighted_f1": 0.6700524758340307,
      "eval_runtime": 305.4934,
      "eval_samples_per_second": 105.007,
      "eval_sleeve_macro_f1": 0.3328543256219435,
      "eval_sleeve_precision": 0.8471446159191267,
      "eval_sleeve_recall": 0.8456622712678076,
      "eval_sleeve_weighted_f1": 0.8332191903962787,
      "eval_steps_per_second": 1.643,
      "eval_style_macro_f1": 0.19206324944658193,
      "eval_style_precision": 0.6055367377975819,
      "eval_style_recall": 0.6526699710090713,
      "eval_style_weighted_f1": 0.6042691229527709,
      "step": 8004
    },
    {
      "epoch": 2.1239380309845077,
      "grad_norm": 34.519927978515625,
      "learning_rate": 3.9878542510121455e-05,
      "loss": 6.7502,
      "step": 8500
    },
    {
      "epoch": 2.2488755622188905,
      "grad_norm": 34.661766052246094,
      "learning_rate": 3.924595141700405e-05,
      "loss": 6.8356,
      "step": 9000
    },
    {
      "epoch": 2.3738130934532733,
      "grad_norm": 33.54723358154297,
      "learning_rate": 3.8613360323886645e-05,
      "loss": 6.6994,
      "step": 9500
    },
    {
      "epoch": 2.498750624687656,
      "grad_norm": 36.0948486328125,
      "learning_rate": 3.798076923076923e-05,
      "loss": 6.6972,
      "step": 10000
    },
    {
      "epoch": 2.623688155922039,
      "grad_norm": 36.17250442504883,
      "learning_rate": 3.734817813765182e-05,
      "loss": 6.6196,
      "step": 10500
    },
    {
      "epoch": 2.7486256871564216,
      "grad_norm": 34.5025634765625,
      "learning_rate": 3.671558704453441e-05,
      "loss": 6.6215,
      "step": 11000
    },
    {
      "epoch": 2.873563218390805,
      "grad_norm": 33.60887908935547,
      "learning_rate": 3.6082995951417005e-05,
      "loss": 6.5135,
      "step": 11500
    },
    {
      "epoch": 2.9985007496251876,
      "grad_norm": 34.10890197753906,
      "learning_rate": 3.5450404858299596e-05,
      "loss": 6.5109,
      "step": 12000
    },
    {
      "epoch": 3.0,
      "eval_category_macro_f1": 0.787512350264236,
      "eval_category_precision": 0.9635144196428028,
      "eval_category_recall": 0.9634028492159981,
      "eval_category_weighted_f1": 0.9624150257086617,
      "eval_color_macro_f1": 0.34627790250554974,
      "eval_color_precision": 0.7784139120798733,
      "eval_color_recall": 0.7970634994856448,
      "eval_color_weighted_f1": 0.7762200635616149,
      "eval_gender_macro_f1": 0.6754785001926855,
      "eval_gender_precision": 0.9226440250759032,
      "eval_gender_recall": 0.9195112067084386,
      "eval_gender_weighted_f1": 0.9188428393913575,
      "eval_loss": 6.8828935623168945,
      "eval_material_macro_f1": 0.34439436651770383,
      "eval_material_precision": 0.6044803466129843,
      "eval_material_recall": 0.6432557124598647,
      "eval_material_weighted_f1": 0.5987268283398588,
      "eval_neck_macro_f1": 0.2636374805840866,
      "eval_neck_precision": 0.7859040450058413,
      "eval_neck_recall": 0.7974375759842888,
      "eval_neck_weighted_f1": 0.785483280031206,
      "eval_pattern_macro_f1": 0.15190625172777147,
      "eval_pattern_precision": 0.5791524062798472,
      "eval_pattern_recall": 0.6267340004364226,
      "eval_pattern_weighted_f1": 0.5539934146060564,
      "eval_product_type_macro_f1": 0.6638223544457101,
      "eval_product_type_precision": 0.7147695152999568,
      "eval_product_type_recall": 0.7016428192898782,
      "eval_product_type_weighted_f1": 0.6952318435900471,
      "eval_runtime": 300.8363,
      "eval_samples_per_second": 106.633,
      "eval_sleeve_macro_f1": 0.3475305353320075,
      "eval_sleeve_precision": 0.8334833047421953,
      "eval_sleeve_recall": 0.8485925371738521,
      "eval_sleeve_weighted_f1": 0.8360923435643329,
      "eval_steps_per_second": 1.669,
      "eval_style_macro_f1": 0.22495804935114228,
      "eval_style_precision": 0.622297675917607,
      "eval_style_recall": 0.6627388634309049,
      "eval_style_weighted_f1": 0.6259413091730935,
      "step": 12006
    },
    {
      "epoch": 3.1234382808595704,
      "grad_norm": 37.55683517456055,
      "learning_rate": 3.481781376518219e-05,
      "loss": 6.0245,
      "step": 12500
    },
    {
      "epoch": 3.248375812093953,
      "grad_norm": 31.214113235473633,
      "learning_rate": 3.418522267206478e-05,
      "loss": 6.0258,
      "step": 13000
    },
    {
      "epoch": 3.373313343328336,
      "grad_norm": 42.683753967285156,
      "learning_rate": 3.355263157894737e-05,
      "loss": 6.0101,
      "step": 13500
    },
    {
      "epoch": 3.4982508745627188,
      "grad_norm": 33.178009033203125,
      "learning_rate": 3.292004048582996e-05,
      "loss": 6.0655,
      "step": 14000
    },
    {
      "epoch": 3.6231884057971016,
      "grad_norm": 33.86723709106445,
      "learning_rate": 3.2287449392712554e-05,
      "loss": 6.0188,
      "step": 14500
    },
    {
      "epoch": 3.7481259370314843,
      "grad_norm": 49.44440460205078,
      "learning_rate": 3.165485829959514e-05,
      "loss": 6.0043,
      "step": 15000
    },
    {
      "epoch": 3.873063468265867,
      "grad_norm": 26.736482620239258,
      "learning_rate": 3.102226720647773e-05,
      "loss": 5.873,
      "step": 15500
    },
    {
      "epoch": 3.99800099950025,
      "grad_norm": 33.90946578979492,
      "learning_rate": 3.0389676113360326e-05,
      "loss": 5.7982,
      "step": 16000
    },
    {
      "epoch": 4.0,
      "eval_category_macro_f1": 0.7955569681071116,
      "eval_category_precision": 0.9633120393270671,
      "eval_category_recall": 0.9633405031328907,
      "eval_category_weighted_f1": 0.962656221421562,
      "eval_color_macro_f1": 0.39246945574897335,
      "eval_color_precision": 0.7946069660631935,
      "eval_color_recall": 0.8061036815362075,
      "eval_color_weighted_f1": 0.7926817060026948,
      "eval_gender_macro_f1": 0.7037573639471852,
      "eval_gender_precision": 0.9465872847377934,
      "eval_gender_recall": 0.9456342155304093,
      "eval_gender_weighted_f1": 0.9454295788332787,
      "eval_loss": 6.5368194580078125,
      "eval_material_macro_f1": 0.38069446247038957,
      "eval_material_precision": 0.6231463458041611,
      "eval_material_recall": 0.6512671841391564,
      "eval_material_weighted_f1": 0.6147786520859259,
      "eval_neck_macro_f1": 0.27672490278496564,
      "eval_neck_precision": 0.7914577242024867,
      "eval_neck_recall": 0.8040774338352193,
      "eval_neck_weighted_f1": 0.792155013664771,
      "eval_pattern_macro_f1": 0.19187411544892444,
      "eval_pattern_precision": 0.586085709618732,
      "eval_pattern_recall": 0.6352130677390193,
      "eval_pattern_weighted_f1": 0.5795812233806185,
      "eval_product_type_macro_f1": 0.7027178432304306,
      "eval_product_type_precision": 0.7296781379497822,
      "eval_product_type_recall": 0.7236198135852115,
      "eval_product_type_weighted_f1": 0.7179279185541192,
      "eval_runtime": 302.0245,
      "eval_samples_per_second": 106.213,
      "eval_sleeve_macro_f1": 0.3677025507155895,
      "eval_sleeve_precision": 0.8508002629167147,
      "eval_sleeve_recall": 0.8566663549362511,
      "eval_sleeve_weighted_f1": 0.8455598759505015,
      "eval_steps_per_second": 1.662,
      "eval_style_macro_f1": 0.2618223911719054,
      "eval_style_precision": 0.6421653958993869,
      "eval_style_recall": 0.6738988123071168,
      "eval_style_weighted_f1": 0.6433547708269393,
      "step": 16008
    },
    {
      "epoch": 4.122938530734633,
      "grad_norm": 40.41415023803711,
      "learning_rate": 2.9757085020242914e-05,
      "loss": 5.3993,
      "step": 16500
    },
    {
      "epoch": 4.2478760619690155,
      "grad_norm": 35.530174255371094,
      "learning_rate": 2.9124493927125506e-05,
      "loss": 5.3668,
      "step": 17000
    },
    {
      "epoch": 4.372813593203398,
      "grad_norm": 37.02978515625,
      "learning_rate": 2.84919028340081e-05,
      "loss": 5.4427,
      "step": 17500
    },
    {
      "epoch": 4.497751124437781,
      "grad_norm": 31.028322219848633,
      "learning_rate": 2.7859311740890692e-05,
      "loss": 5.4199,
      "step": 18000
    },
    {
      "epoch": 4.622688655672164,
      "grad_norm": 39.566871643066406,
      "learning_rate": 2.722672064777328e-05,
      "loss": 5.3339,
      "step": 18500
    },
    {
      "epoch": 4.747626186906547,
      "grad_norm": 29.115066528320312,
      "learning_rate": 2.6594129554655872e-05,
      "loss": 5.3469,
      "step": 19000
    },
    {
      "epoch": 4.872563718140929,
      "grad_norm": 27.22176742553711,
      "learning_rate": 2.5961538461538464e-05,
      "loss": 5.3245,
      "step": 19500
    },
    {
      "epoch": 4.997501249375312,
      "grad_norm": 30.805620193481445,
      "learning_rate": 2.5328947368421052e-05,
      "loss": 5.3437,
      "step": 20000
    },
    {
      "epoch": 5.0,
      "eval_category_macro_f1": 0.8085353223473223,
      "eval_category_precision": 0.9670311680024095,
      "eval_category_recall": 0.9672059602855451,
      "eval_category_weighted_f1": 0.9663742766459245,
      "eval_color_macro_f1": 0.4238649051693472,
      "eval_color_precision": 0.7981474301781062,
      "eval_color_recall": 0.8128994045949063,
      "eval_color_weighted_f1": 0.8002174907155432,
      "eval_gender_macro_f1": 0.7196717213524463,
      "eval_gender_precision": 0.9467508045608919,
      "eval_gender_recall": 0.9464447146108046,
      "eval_gender_weighted_f1": 0.9462666634376086,
      "eval_loss": 6.324069499969482,
      "eval_material_macro_f1": 0.40655027085392736,
      "eval_material_precision": 0.6409986821944779,
      "eval_material_recall": 0.6640169581346052,
      "eval_material_weighted_f1": 0.6282823649788557,
      "eval_neck_macro_f1": 0.28417588679207595,
      "eval_neck_precision": 0.8012300473656296,
      "eval_neck_recall": 0.8069141806166028,
      "eval_neck_weighted_f1": 0.7976287506977606,
      "eval_pattern_macro_f1": 0.2187785306978239,
      "eval_pattern_precision": 0.6129304703362072,
      "eval_pattern_recall": 0.6472770348202874,
      "eval_pattern_weighted_f1": 0.5881639092284188,
      "eval_product_type_macro_f1": 0.7169662707860948,
      "eval_product_type_precision": 0.739852886082883,
      "eval_product_type_recall": 0.7304778827270176,
      "eval_product_type_weighted_f1": 0.7280465570285056,
      "eval_runtime": 300.8243,
      "eval_samples_per_second": 106.637,
      "eval_sleeve_macro_f1": 0.39617711979641707,
      "eval_sleeve_precision": 0.8573303931972993,
      "eval_sleeve_recall": 0.858069141806166,
      "eval_sleeve_weighted_f1": 0.8493273999182868,
      "eval_steps_per_second": 1.669,
      "eval_style_macro_f1": 0.2753572234324515,
      "eval_style_precision": 0.6415522634195322,
      "eval_style_recall": 0.6796658249945448,
      "eval_style_weighted_f1": 0.644725501006127,
      "step": 20010
    },
    {
      "epoch": 5.122438780609695,
      "grad_norm": 32.560482025146484,
      "learning_rate": 2.4696356275303644e-05,
      "loss": 4.7955,
      "step": 20500
    },
    {
      "epoch": 5.247376311844078,
      "grad_norm": 31.55764389038086,
      "learning_rate": 2.4063765182186235e-05,
      "loss": 4.7692,
      "step": 21000
    },
    {
      "epoch": 5.3723138430784605,
      "grad_norm": 31.34699821472168,
      "learning_rate": 2.3431174089068827e-05,
      "loss": 4.8501,
      "step": 21500
    },
    {
      "epoch": 5.497251374312843,
      "grad_norm": 35.56719207763672,
      "learning_rate": 2.279858299595142e-05,
      "loss": 4.8357,
      "step": 22000
    },
    {
      "epoch": 5.622188905547226,
      "grad_norm": 27.214921951293945,
      "learning_rate": 2.216599190283401e-05,
      "loss": 4.7804,
      "step": 22500
    },
    {
      "epoch": 5.747126436781609,
      "grad_norm": 47.55564880371094,
      "learning_rate": 2.15334008097166e-05,
      "loss": 4.8164,
      "step": 23000
    },
    {
      "epoch": 5.872063968015992,
      "grad_norm": 27.112457275390625,
      "learning_rate": 2.090080971659919e-05,
      "loss": 4.8067,
      "step": 23500
    },
    {
      "epoch": 5.997001499250375,
      "grad_norm": 37.71267318725586,
      "learning_rate": 2.0268218623481782e-05,
      "loss": 4.7547,
      "step": 24000
    },
    {
      "epoch": 6.0,
      "eval_category_macro_f1": 0.8067492828396051,
      "eval_category_precision": 0.9675417493566353,
      "eval_category_recall": 0.967767075033511,
      "eval_category_weighted_f1": 0.967008960660055,
      "eval_color_macro_f1": 0.46170642529262285,
      "eval_color_precision": 0.808572630866145,
      "eval_color_recall": 0.8167025156644534,
      "eval_color_weighted_f1": 0.8052156787210335,
      "eval_gender_macro_f1": 0.7253099260026424,
      "eval_gender_precision": 0.9516655675023045,
      "eval_gender_recall": 0.9511206708438542,
      "eval_gender_weighted_f1": 0.9509797608509165,
      "eval_loss": 6.1924147605896,
      "eval_material_macro_f1": 0.4418062098856238,
      "eval_material_precision": 0.6532578684941821,
      "eval_material_recall": 0.6747716574706194,
      "eval_material_weighted_f1": 0.6468422079358449,
      "eval_neck_macro_f1": 0.30263560541401846,
      "eval_neck_precision": 0.8066569574286967,
      "eval_neck_recall": 0.812556501137816,
      "eval_neck_weighted_f1": 0.8053312212408406,
      "eval_pattern_macro_f1": 0.24683911824062632,
      "eval_pattern_precision": 0.6203592848139506,
      "eval_pattern_recall": 0.6539792387543253,
      "eval_pattern_weighted_f1": 0.6034023280202176,
      "eval_product_type_macro_f1": 0.7286701507504901,
      "eval_product_type_precision": 0.7477061999565924,
      "eval_product_type_recall": 0.7434146949717884,
      "eval_product_type_weighted_f1": 0.7404621958524741,
      "eval_runtime": 301.5697,
      "eval_samples_per_second": 106.373,
      "eval_sleeve_macro_f1": 0.4149438597919067,
      "eval_sleeve_precision": 0.8619098226901072,
      "eval_sleeve_recall": 0.8607500233797811,
      "eval_sleeve_weighted_f1": 0.8529440106549326,
      "eval_steps_per_second": 1.665,
      "eval_style_macro_f1": 0.28852599697346254,
      "eval_style_precision": 0.6561689572939597,
      "eval_style_recall": 0.6877396427569438,
      "eval_style_weighted_f1": 0.6590960855736121,
      "step": 24012
    },
    {
      "epoch": 6.121939030484757,
      "grad_norm": 31.159828186035156,
      "learning_rate": 1.9635627530364373e-05,
      "loss": 4.2872,
      "step": 24500
    },
    {
      "epoch": 6.246876561719141,
      "grad_norm": 36.643314361572266,
      "learning_rate": 1.9003036437246965e-05,
      "loss": 4.2826,
      "step": 25000
    },
    {
      "epoch": 6.371814092953524,
      "grad_norm": 31.81490707397461,
      "learning_rate": 1.8370445344129557e-05,
      "loss": 4.2782,
      "step": 25500
    },
    {
      "epoch": 6.496751624187906,
      "grad_norm": 41.419471740722656,
      "learning_rate": 1.7737854251012145e-05,
      "loss": 4.3151,
      "step": 26000
    },
    {
      "epoch": 6.621689155422289,
      "grad_norm": 36.339202880859375,
      "learning_rate": 1.7105263157894737e-05,
      "loss": 4.2912,
      "step": 26500
    },
    {
      "epoch": 6.746626686656672,
      "grad_norm": 35.872161865234375,
      "learning_rate": 1.6472672064777328e-05,
      "loss": 4.321,
      "step": 27000
    },
    {
      "epoch": 6.871564217891055,
      "grad_norm": 31.54119873046875,
      "learning_rate": 1.584008097165992e-05,
      "loss": 4.2605,
      "step": 27500
    },
    {
      "epoch": 6.9965017491254375,
      "grad_norm": 29.37108039855957,
      "learning_rate": 1.5207489878542511e-05,
      "loss": 4.1995,
      "step": 28000
    },
    {
      "epoch": 7.0,
      "eval_category_macro_f1": 0.8019924698000542,
      "eval_category_precision": 0.9653353385671121,
      "eval_category_recall": 0.9655849621247545,
      "eval_category_weighted_f1": 0.9651467292153788,
      "eval_color_macro_f1": 0.4778177910015613,
      "eval_color_precision": 0.8132908708208361,
      "eval_color_recall": 0.821846067520808,
      "eval_color_weighted_f1": 0.8108879503190392,
      "eval_gender_macro_f1": 0.7304886482949747,
      "eval_gender_precision": 0.9575237181327425,
      "eval_gender_recall": 0.9568253374481748,
      "eval_gender_weighted_f1": 0.9565473974994894,
      "eval_loss": 6.094598770141602,
      "eval_material_macro_f1": 0.4445061914881263,
      "eval_material_precision": 0.659882151181023,
      "eval_material_recall": 0.6815050344462109,
      "eval_material_weighted_f1": 0.6551634350927689,
      "eval_neck_macro_f1": 0.3312326690246815,
      "eval_neck_precision": 0.8066317752184025,
      "eval_neck_recall": 0.8143333645063748,
      "eval_neck_weighted_f1": 0.8067876189718791,
      "eval_pattern_macro_f1": 0.26541974190021195,
      "eval_pattern_precision": 0.6297518110690976,
      "eval_pattern_recall": 0.6604632313974874,
      "eval_pattern_weighted_f1": 0.6108385812217497,
      "eval_product_type_macro_f1": 0.7399005420683133,
      "eval_product_type_precision": 0.7515415781015994,
      "eval_product_type_recall": 0.7487141120359113,
      "eval_product_type_weighted_f1": 0.7451143261096842,
      "eval_runtime": 299.9165,
      "eval_samples_per_second": 106.96,
      "eval_sleeve_macro_f1": 0.43326486489655175,
      "eval_sleeve_precision": 0.8584343914083422,
      "eval_sleeve_recall": 0.8612799650861934,
      "eval_sleeve_weighted_f1": 0.8542297807421783,
      "eval_steps_per_second": 1.674,
      "eval_style_macro_f1": 0.3162728761585937,
      "eval_style_precision": 0.6621445575945633,
      "eval_style_recall": 0.6907946008292029,
      "eval_style_weighted_f1": 0.6607505081198878,
      "step": 28014
    },
    {
      "epoch": 7.12143928035982,
      "grad_norm": 43.210941314697266,
      "learning_rate": 1.4574898785425101e-05,
      "loss": 3.7497,
      "step": 28500
    },
    {
      "epoch": 7.246376811594203,
      "grad_norm": 34.3553352355957,
      "learning_rate": 1.3942307692307693e-05,
      "loss": 3.8133,
      "step": 29000
    },
    {
      "epoch": 7.371314342828586,
      "grad_norm": 38.11077880859375,
      "learning_rate": 1.3309716599190283e-05,
      "loss": 3.8536,
      "step": 29500
    },
    {
      "epoch": 7.496251874062969,
      "grad_norm": 34.092018127441406,
      "learning_rate": 1.2677125506072876e-05,
      "loss": 3.8057,
      "step": 30000
    },
    {
      "epoch": 7.6211894052973514,
      "grad_norm": 31.957683563232422,
      "learning_rate": 1.2044534412955466e-05,
      "loss": 3.813,
      "step": 30500
    },
    {
      "epoch": 7.746126936531734,
      "grad_norm": 34.03321075439453,
      "learning_rate": 1.1411943319838058e-05,
      "loss": 3.7924,
      "step": 31000
    },
    {
      "epoch": 7.871064467766117,
      "grad_norm": 38.06154251098633,
      "learning_rate": 1.077935222672065e-05,
      "loss": 3.8135,
      "step": 31500
    },
    {
      "epoch": 7.9960019990005,
      "grad_norm": 45.78370666503906,
      "learning_rate": 1.014676113360324e-05,
      "loss": 3.7866,
      "step": 32000
    },
    {
      "epoch": 8.0,
      "eval_category_macro_f1": 0.8146478162196177,
      "eval_category_precision": 0.9667355952687459,
      "eval_category_recall": 0.9672059602855451,
      "eval_category_weighted_f1": 0.9666613335359282,
      "eval_color_macro_f1": 0.494972223269362,
      "eval_color_precision": 0.8179178024945392,
      "eval_color_recall": 0.8253686212163721,
      "eval_color_weighted_f1": 0.8171143847468179,
      "eval_gender_macro_f1": 0.726822473815732,
      "eval_gender_precision": 0.9595388610516349,
      "eval_gender_recall": 0.9593503538140217,
      "eval_gender_weighted_f1": 0.9590467785701566,
      "eval_loss": 6.086089134216309,
      "eval_material_macro_f1": 0.4653083418810649,
      "eval_material_precision": 0.6660165608707495,
      "eval_material_recall": 0.6859316063468313,
      "eval_material_weighted_f1": 0.6641222350586232,
      "eval_neck_macro_f1": 0.3385488552152788,
      "eval_neck_precision": 0.8103884255493873,
      "eval_neck_recall": 0.8168272078306681,
      "eval_neck_weighted_f1": 0.8100463958001771,
      "eval_pattern_macro_f1": 0.29354348124717355,
      "eval_pattern_precision": 0.6421689432624941,
      "eval_pattern_recall": 0.6699398360298014,
      "eval_pattern_weighted_f1": 0.6249061713173589,
      "eval_product_type_macro_f1": 0.750138460482341,
      "eval_product_type_precision": 0.7549316716640776,
      "eval_product_type_recall": 0.7525795691885657,
      "eval_product_type_weighted_f1": 0.7506829441074545,
      "eval_runtime": 301.8256,
      "eval_samples_per_second": 106.283,
      "eval_sleeve_macro_f1": 0.4663248985167306,
      "eval_sleeve_precision": 0.8591320313081251,
      "eval_sleeve_recall": 0.86252688674834,
      "eval_sleeve_weighted_f1": 0.8568151400744953,
      "eval_steps_per_second": 1.663,
      "eval_style_macro_f1": 0.3329861421512565,
      "eval_style_precision": 0.6707667240667725,
      "eval_style_recall": 0.6972162473892578,
      "eval_style_weighted_f1": 0.6732692090638879,
      "step": 32016
    }
  ],
  "logging_steps": 500,
  "max_steps": 40020,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 2
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}