{
  "best_global_step": 4500,
  "best_metric": 0.9824253474141716,
  "best_model_checkpoint": "/data/semantic-router.bak/models/mmbert32k_feedback_detector_lora/checkpoint-4500",
  "epoch": 4.021447721179625,
  "eval_steps": 500,
  "global_step": 4500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.08936550491510277,
      "grad_norm": 36.226261138916016,
      "learning_rate": 1.769436997319035e-06,
      "loss": 3.3394851684570312,
      "step": 100
    },
    {
      "epoch": 0.17873100983020554,
      "grad_norm": 18.235830307006836,
      "learning_rate": 3.5567470956210903e-06,
      "loss": 1.2395378875732421,
      "step": 200
    },
    {
      "epoch": 0.2680965147453083,
      "grad_norm": 23.832252502441406,
      "learning_rate": 5.344057193923146e-06,
      "loss": 0.598966407775879,
      "step": 300
    },
    {
      "epoch": 0.3574620196604111,
      "grad_norm": 25.597322463989258,
      "learning_rate": 7.131367292225202e-06,
      "loss": 0.40449214935302735,
      "step": 400
    },
    {
      "epoch": 0.44682752457551383,
      "grad_norm": 20.845447540283203,
      "learning_rate": 8.918677390527258e-06,
      "loss": 0.316771125793457,
      "step": 500
    },
    {
      "epoch": 0.44682752457551383,
      "eval_accuracy": 0.9326633165829146,
      "eval_f1_NEED_CLARIFICATION": 0.9524752475247524,
      "eval_f1_SAT": 0.9983260796786073,
      "eval_f1_WANT_DIFFERENT": 0.8243801652892562,
      "eval_f1_WRONG_ANSWER": 0.8218905472636816,
      "eval_f1_macro": 0.8992680099390744,
      "eval_f1_weighted": 0.9322209959529292,
      "eval_loss": 0.1708805114030838,
      "eval_runtime": 7.0864,
      "eval_samples_per_second": 421.229,
      "eval_steps_per_second": 13.265,
      "step": 500
    },
    {
      "epoch": 0.5361930294906166,
      "grad_norm": 13.492064476013184,
      "learning_rate": 1.0705987488829312e-05,
      "loss": 0.21014398574829102,
      "step": 600
    },
    {
      "epoch": 0.6255585344057194,
      "grad_norm": 1.0692031383514404,
      "learning_rate": 1.2493297587131368e-05,
      "loss": 0.23827705383300782,
      "step": 700
    },
    {
      "epoch": 0.7149240393208222,
      "grad_norm": 21.394258499145508,
      "learning_rate": 1.4280607685433422e-05,
      "loss": 0.22943824768066406,
      "step": 800
    },
    {
      "epoch": 0.8042895442359249,
      "grad_norm": 9.537849426269531,
      "learning_rate": 1.6067917783735482e-05,
      "loss": 0.18561811447143556,
      "step": 900
    },
    {
      "epoch": 0.8936550491510277,
      "grad_norm": 13.74893569946289,
      "learning_rate": 1.7855227882037534e-05,
      "loss": 0.18598514556884765,
      "step": 1000
    },
    {
      "epoch": 0.8936550491510277,
      "eval_accuracy": 0.9534338358458961,
      "eval_f1_NEED_CLARIFICATION": 0.9868819374369324,
      "eval_f1_SAT": 0.9959595959595959,
      "eval_f1_WANT_DIFFERENT": 0.8679245283018868,
      "eval_f1_WRONG_ANSWER": 0.8790613718411552,
      "eval_f1_macro": 0.9324568583848927,
      "eval_f1_weighted": 0.9535818896785879,
      "eval_loss": 0.1380130499601364,
      "eval_runtime": 4.5166,
      "eval_samples_per_second": 660.891,
      "eval_steps_per_second": 20.812,
      "step": 1000
    },
    {
      "epoch": 0.9830205540661304,
      "grad_norm": 6.975687026977539,
      "learning_rate": 1.964253798033959e-05,
      "loss": 0.15200217247009276,
      "step": 1100
    },
    {
      "epoch": 1.0723860589812333,
      "grad_norm": 0.12164141982793808,
      "learning_rate": 1.984112799126204e-05,
      "loss": 0.12448848724365234,
      "step": 1200
    },
    {
      "epoch": 1.161751563896336,
      "grad_norm": 17.040109634399414,
      "learning_rate": 1.964253798033959e-05,
      "loss": 0.1404490852355957,
      "step": 1300
    },
    {
      "epoch": 1.2511170688114388,
      "grad_norm": 3.555215358734131,
      "learning_rate": 1.944394796941714e-05,
      "loss": 0.12125076293945312,
      "step": 1400
    },
    {
      "epoch": 1.3404825737265416,
      "grad_norm": 2.8291819095611572,
      "learning_rate": 1.924535795849469e-05,
      "loss": 0.09241329193115234,
      "step": 1500
    },
    {
      "epoch": 1.3404825737265416,
      "eval_accuracy": 0.9768844221105528,
      "eval_f1_NEED_CLARIFICATION": 0.9910269192422732,
      "eval_f1_SAT": 1.0,
      "eval_f1_WANT_DIFFERENT": 0.9334677419354839,
      "eval_f1_WRONG_ANSWER": 0.9365558912386707,
      "eval_f1_macro": 0.965262638104107,
      "eval_f1_weighted": 0.9768184841217358,
      "eval_loss": 0.07608116418123245,
      "eval_runtime": 4.3221,
      "eval_samples_per_second": 690.644,
      "eval_steps_per_second": 21.749,
      "step": 1500
    },
    {
      "epoch": 1.4298480786416443,
      "grad_norm": 1.8353885412216187,
      "learning_rate": 1.9046767947572236e-05,
      "loss": 0.1183387851715088,
      "step": 1600
    },
    {
      "epoch": 1.519213583556747,
      "grad_norm": 0.9442969560623169,
      "learning_rate": 1.8848177936649786e-05,
      "loss": 0.13099024772644044,
      "step": 1700
    },
    {
      "epoch": 1.6085790884718498,
      "grad_norm": 38.03871536254883,
      "learning_rate": 1.8649587925727336e-05,
      "loss": 0.11420077323913574,
      "step": 1800
    },
    {
      "epoch": 1.6979445933869526,
      "grad_norm": 6.626965522766113,
      "learning_rate": 1.8450997914804885e-05,
      "loss": 0.07563745021820069,
      "step": 1900
    },
    {
      "epoch": 1.7873100983020556,
      "grad_norm": 17.789060592651367,
      "learning_rate": 1.8252407903882435e-05,
      "loss": 0.11484715461730957,
      "step": 2000
    },
    {
      "epoch": 1.7873100983020556,
      "eval_accuracy": 0.9782244556113903,
      "eval_f1_NEED_CLARIFICATION": 0.9979959919839679,
      "eval_f1_SAT": 1.0,
      "eval_f1_WANT_DIFFERENT": 0.9319371727748691,
      "eval_f1_WRONG_ANSWER": 0.9391304347826087,
      "eval_f1_macro": 0.9672658998853614,
      "eval_f1_weighted": 0.9781553341948542,
      "eval_loss": 0.0778149887919426,
      "eval_runtime": 4.3058,
      "eval_samples_per_second": 693.246,
      "eval_steps_per_second": 21.831,
      "step": 2000
    },
    {
      "epoch": 1.876675603217158,
      "grad_norm": 0.08597684651613235,
      "learning_rate": 1.8053817892959985e-05,
      "loss": 0.1267116928100586,
      "step": 2100
    },
    {
      "epoch": 1.966041108132261,
      "grad_norm": 0.21768970787525177,
      "learning_rate": 1.7855227882037534e-05,
      "loss": 0.09952692985534668,
      "step": 2200
    },
    {
      "epoch": 2.0554066130473636,
      "grad_norm": 0.057887863367795944,
      "learning_rate": 1.7656637871115084e-05,
      "loss": 0.05328631401062012,
      "step": 2300
    },
    {
      "epoch": 2.1447721179624666,
      "grad_norm": 0.004557018168270588,
      "learning_rate": 1.7458047860192634e-05,
      "loss": 0.057413887977600095,
      "step": 2400
    },
    {
      "epoch": 2.234137622877569,
      "grad_norm": 0.092039555311203,
      "learning_rate": 1.7259457849270184e-05,
      "loss": 0.05033339023590088,
      "step": 2500
    },
    {
      "epoch": 2.234137622877569,
      "eval_accuracy": 0.980569514237856,
      "eval_f1_NEED_CLARIFICATION": 0.9959839357429718,
      "eval_f1_SAT": 1.0,
      "eval_f1_WANT_DIFFERENT": 0.9414182939362795,
      "eval_f1_WRONG_ANSWER": 0.9460255152109912,
      "eval_f1_macro": 0.9708569362225606,
      "eval_f1_weighted": 0.9805517644741512,
      "eval_loss": 0.08076681196689606,
      "eval_runtime": 4.4113,
      "eval_samples_per_second": 676.667,
      "eval_steps_per_second": 21.309,
      "step": 2500
    },
    {
      "epoch": 2.323503127792672,
      "grad_norm": 0.08486219495534897,
      "learning_rate": 1.7060867838347733e-05,
      "loss": 0.0629350757598877,
      "step": 2600
    },
    {
      "epoch": 2.4128686327077746,
      "grad_norm": 0.005150920711457729,
      "learning_rate": 1.6862277827425283e-05,
      "loss": 0.07242729187011719,
      "step": 2700
    },
    {
      "epoch": 2.5022341376228776,
      "grad_norm": 10.539729118347168,
      "learning_rate": 1.6663687816502833e-05,
      "loss": 0.07086173057556153,
      "step": 2800
    },
    {
      "epoch": 2.5915996425379806,
      "grad_norm": 0.5994746685028076,
      "learning_rate": 1.6465097805580382e-05,
      "loss": 0.03949449300765991,
      "step": 2900
    },
    {
      "epoch": 2.680965147453083,
      "grad_norm": 0.03495287150144577,
      "learning_rate": 1.6266507794657932e-05,
      "loss": 0.05908185482025147,
      "step": 3000
    },
    {
      "epoch": 2.680965147453083,
      "eval_accuracy": 0.9835845896147404,
      "eval_f1_NEED_CLARIFICATION": 0.995004995004995,
      "eval_f1_SAT": 1.0,
      "eval_f1_WANT_DIFFERENT": 0.9524752475247524,
      "eval_f1_WRONG_ANSWER": 0.9539406345957011,
      "eval_f1_macro": 0.9753552192813622,
      "eval_f1_weighted": 0.9835536337716829,
      "eval_loss": 0.06656040251255035,
      "eval_runtime": 4.4851,
      "eval_samples_per_second": 665.535,
      "eval_steps_per_second": 20.958,
      "step": 3000
    },
    {
      "epoch": 2.7703306523681857,
      "grad_norm": 0.0323098860681057,
      "learning_rate": 1.6067917783735482e-05,
      "loss": 0.05086075782775879,
      "step": 3100
    },
    {
      "epoch": 2.8596961572832886,
      "grad_norm": 3.2358248233795166,
      "learning_rate": 1.5869327772813028e-05,
      "loss": 0.053619518280029296,
      "step": 3200
    },
    {
      "epoch": 2.9490616621983916,
      "grad_norm": 0.0017350203124806285,
      "learning_rate": 1.5670737761890578e-05,
      "loss": 0.07301931381225586,
      "step": 3300
    },
    {
      "epoch": 3.038427167113494,
      "grad_norm": 0.062237728387117386,
      "learning_rate": 1.5472147750968128e-05,
      "loss": 0.04660813331604004,
      "step": 3400
    },
    {
      "epoch": 3.127792672028597,
      "grad_norm": 0.16048528254032135,
      "learning_rate": 1.5273557740045677e-05,
      "loss": 0.021572010517120363,
      "step": 3500
    },
    {
      "epoch": 3.127792672028597,
      "eval_accuracy": 0.9835845896147404,
      "eval_f1_NEED_CLARIFICATION": 0.996,
      "eval_f1_SAT": 1.0,
      "eval_f1_WANT_DIFFERENT": 0.9492227979274611,
      "eval_f1_WRONG_ANSWER": 0.9560117302052786,
      "eval_f1_macro": 0.975308632033185,
      "eval_f1_weighted": 0.9835225443919949,
      "eval_loss": 0.08207044005393982,
      "eval_runtime": 4.5888,
      "eval_samples_per_second": 650.501,
      "eval_steps_per_second": 20.485,
      "step": 3500
    },
    {
      "epoch": 3.2171581769436997,
      "grad_norm": 0.026359504088759422,
      "learning_rate": 1.5074967729123227e-05,
      "loss": 0.025213963985443115,
      "step": 3600
    },
    {
      "epoch": 3.3065236818588026,
      "grad_norm": 1.627691626548767,
      "learning_rate": 1.4876377718200777e-05,
      "loss": 0.019634859561920168,
      "step": 3700
    },
    {
      "epoch": 3.395889186773905,
      "grad_norm": 2.8381104469299316,
      "learning_rate": 1.4677787707278326e-05,
      "loss": 0.03549041748046875,
      "step": 3800
    },
    {
      "epoch": 3.485254691689008,
      "grad_norm": 0.0038880000356584787,
      "learning_rate": 1.4479197696355876e-05,
      "loss": 0.02453033924102783,
      "step": 3900
    },
    {
      "epoch": 3.5746201966041107,
      "grad_norm": 0.33596596121788025,
      "learning_rate": 1.4280607685433422e-05,
      "loss": 0.028336784839630126,
      "step": 4000
    },
    {
      "epoch": 3.5746201966041107,
      "eval_accuracy": 0.9862646566164154,
      "eval_f1_NEED_CLARIFICATION": 0.9989969909729187,
      "eval_f1_SAT": 1.0,
      "eval_f1_WANT_DIFFERENT": 0.9594460929772503,
      "eval_f1_WRONG_ANSWER": 0.9591836734693877,
      "eval_f1_macro": 0.9794066893548892,
      "eval_f1_weighted": 0.986257328373514,
      "eval_loss": 0.07474986463785172,
      "eval_runtime": 4.5088,
      "eval_samples_per_second": 662.045,
      "eval_steps_per_second": 20.848,
      "step": 4000
    },
    {
      "epoch": 3.6639857015192137,
      "grad_norm": 0.03944293409585953,
      "learning_rate": 1.4082017674510972e-05,
      "loss": 0.020915756225585936,
      "step": 4100
    },
    {
      "epoch": 3.753351206434316,
      "grad_norm": 0.007337045390158892,
      "learning_rate": 1.3883427663588522e-05,
      "loss": 0.02648637056350708,
      "step": 4200
    },
    {
      "epoch": 3.842716711349419,
      "grad_norm": 0.6361276507377625,
      "learning_rate": 1.3684837652666072e-05,
      "loss": 0.030732865333557128,
      "step": 4300
    },
    {
      "epoch": 3.932082216264522,
      "grad_norm": 6.5225138664245605,
      "learning_rate": 1.3486247641743621e-05,
      "loss": 0.033404347896575926,
      "step": 4400
    },
    {
      "epoch": 4.021447721179625,
      "grad_norm": 0.06466234475374222,
      "learning_rate": 1.3287657630821171e-05,
      "loss": 0.015966687202453613,
      "step": 4500
    },
    {
      "epoch": 4.021447721179625,
      "eval_accuracy": 0.9882747068676717,
      "eval_f1_NEED_CLARIFICATION": 0.9979919678714859,
      "eval_f1_SAT": 1.0,
      "eval_f1_WANT_DIFFERENT": 0.9646107178968655,
      "eval_f1_WRONG_ANSWER": 0.967098703888335,
      "eval_f1_macro": 0.9824253474141716,
      "eval_f1_weighted": 0.9882717896311659,
      "eval_loss": 0.07450290769338608,
      "eval_runtime": 4.4144,
      "eval_samples_per_second": 676.201,
      "eval_steps_per_second": 21.294,
      "step": 4500
    }
  ],
  "logging_steps": 100,
  "max_steps": 11190,
  "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": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.751305188717363e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}