{
  "best_metric": 17.63062286376953,
  "best_model_checkpoint": "./qwen2.5-0.5b/qwen2.5-0.5b-expo-L2EXPO-ES-0.01/checkpoint-50",
  "epoch": 0.8502598016060463,
  "eval_steps": 50,
  "global_step": 300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "dpo_loss": 0.6931471824645996,
      "epoch": 0.002834199338686821,
      "grad_norm": 3.6884903788633325,
      "learning_rate": 2.840909090909091e-08,
      "logits": -1.359458565711975,
      "logps": -84.69721221923828,
      "loss": 0.3913,
      "objective": 0.3618059456348419,
      "ranking_idealized": 0.875,
      "ranking_idealized_expo": 0.5833333134651184,
      "ranking_simple": 0.5833333134651184,
      "regularize": 0.3618059456348419,
      "step": 1,
      "wo_beta": 14.830931663513184
    },
    {
      "dpo_loss": 0.6914557218551636,
      "epoch": 0.14170996693434104,
      "grad_norm": 4.3876993528860435,
      "learning_rate": 1.4204545454545458e-06,
      "logits": -1.524561882019043,
      "logps": -82.59858703613281,
      "loss": 0.4144,
      "objective": 0.41888776421546936,
      "ranking_idealized": 0.8945578336715698,
      "ranking_idealized_expo": 0.5348639488220215,
      "ranking_simple": 0.5284863710403442,
      "regularize": 0.4188877046108246,
      "step": 50,
      "wo_beta": 16.066608428955078
    },
    {
      "epoch": 0.14170996693434104,
      "eval_dpo_loss": 0.6905742287635803,
      "eval_logits": -1.6666735410690308,
      "eval_logps": -92.34442901611328,
      "eval_loss": 0.41158440709114075,
      "eval_objective": 0.4097130298614502,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5289855003356934,
      "eval_regularize": 0.4097130298614502,
      "eval_runtime": 308.7557,
      "eval_samples_per_second": 18.753,
      "eval_steps_per_second": 1.564,
      "eval_wo_beta": 17.63062286376953,
      "step": 50
    },
    {
      "dpo_loss": 0.6717649102210999,
      "epoch": 0.2834199338686821,
      "grad_norm": 9.179969250495992,
      "learning_rate": 2.8409090909090916e-06,
      "logits": -1.801299810409546,
      "logps": -98.4464111328125,
      "loss": 0.394,
      "objective": 0.39795053005218506,
      "ranking_idealized": 0.8958333134651184,
      "ranking_idealized_expo": 0.5270833373069763,
      "ranking_simple": 0.5887500047683716,
      "regularize": 0.39795053005218506,
      "step": 100,
      "wo_beta": 20.196491241455078
    },
    {
      "epoch": 0.2834199338686821,
      "eval_dpo_loss": 0.6862898468971252,
      "eval_logits": -2.0312013626098633,
      "eval_logps": -115.87567138671875,
      "eval_loss": 0.4077964425086975,
      "eval_objective": 0.40739303827285767,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5398550629615784,
      "eval_regularize": 0.40739303827285767,
      "eval_runtime": 307.8646,
      "eval_samples_per_second": 18.807,
      "eval_steps_per_second": 1.569,
      "eval_wo_beta": 22.105756759643555,
      "step": 100
    },
    {
      "dpo_loss": 0.6482623219490051,
      "epoch": 0.42512990080302315,
      "grad_norm": 9.270844197401257,
      "learning_rate": 4.2613636363636365e-06,
      "logits": -1.8636983633041382,
      "logps": -110.8066177368164,
      "loss": 0.3504,
      "objective": 0.35595157742500305,
      "ranking_idealized": 0.8862500190734863,
      "ranking_idealized_expo": 0.5358333587646484,
      "ranking_simple": 0.6287500262260437,
      "regularize": 0.35595157742500305,
      "step": 150,
      "wo_beta": 25.672834396362305
    },
    {
      "epoch": 0.42512990080302315,
      "eval_dpo_loss": 0.6803079843521118,
      "eval_logits": -2.0504531860351562,
      "eval_logps": -123.1670150756836,
      "eval_loss": 0.4044354259967804,
      "eval_objective": 0.4018007516860962,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5719461441040039,
      "eval_regularize": 0.4018007516860962,
      "eval_runtime": 308.5087,
      "eval_samples_per_second": 18.768,
      "eval_steps_per_second": 1.566,
      "eval_wo_beta": 23.56603240966797,
      "step": 150
    },
    {
      "dpo_loss": 0.6220158934593201,
      "epoch": 0.5668398677373642,
      "grad_norm": 8.001451658093092,
      "learning_rate": 4.997168347957521e-06,
      "logits": -1.9987245798110962,
      "logps": -114.0056381225586,
      "loss": 0.3135,
      "objective": 0.3158811330795288,
      "ranking_idealized": 0.8883333206176758,
      "ranking_idealized_expo": 0.527916669845581,
      "ranking_simple": 0.67208331823349,
      "regularize": 0.3158811330795288,
      "step": 200,
      "wo_beta": 29.25323486328125
    },
    {
      "epoch": 0.5668398677373642,
      "eval_dpo_loss": 0.678061842918396,
      "eval_logits": -2.1408538818359375,
      "eval_logps": -121.60306549072266,
      "eval_loss": 0.4005555212497711,
      "eval_objective": 0.397381991147995,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5621117949485779,
      "eval_regularize": 0.397381991147995,
      "eval_runtime": 307.9864,
      "eval_samples_per_second": 18.8,
      "eval_steps_per_second": 1.568,
      "eval_wo_beta": 23.097728729248047,
      "step": 200
    },
    {
      "dpo_loss": 0.5992652773857117,
      "epoch": 0.7085498346717053,
      "grad_norm": 8.50379539811797,
      "learning_rate": 4.973122855144066e-06,
      "logits": -2.132956027984619,
      "logps": -119.59600067138672,
      "loss": 0.2807,
      "objective": 0.28073957562446594,
      "ranking_idealized": 0.8787500262260437,
      "ranking_idealized_expo": 0.5266666412353516,
      "ranking_simple": 0.71875,
      "regularize": 0.28073957562446594,
      "step": 250,
      "wo_beta": 33.336734771728516
    },
    {
      "epoch": 0.7085498346717053,
      "eval_dpo_loss": 0.6789642572402954,
      "eval_logits": -2.3711183071136475,
      "eval_logps": -122.16390228271484,
      "eval_loss": 0.4042646586894989,
      "eval_objective": 0.4010438024997711,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5600414276123047,
      "eval_regularize": 0.4010438024997711,
      "eval_runtime": 308.0112,
      "eval_samples_per_second": 18.798,
      "eval_steps_per_second": 1.568,
      "eval_wo_beta": 23.668811798095703,
      "step": 250
    },
    {
      "dpo_loss": 0.5932748317718506,
      "epoch": 0.8502598016060463,
      "grad_norm": 7.481377422405124,
      "learning_rate": 4.924776641419513e-06,
      "logits": -2.1575162410736084,
      "logps": -116.334716796875,
      "loss": 0.2532,
      "objective": 0.2530641555786133,
      "ranking_idealized": 0.8733333349227905,
      "ranking_idealized_expo": 0.5104166865348816,
      "ranking_simple": 0.7279166579246521,
      "regularize": 0.2530641555786133,
      "step": 300,
      "wo_beta": 33.5853271484375
    },
    {
      "epoch": 0.8502598016060463,
      "eval_dpo_loss": 0.6769444346427917,
      "eval_logits": -2.2620654106140137,
      "eval_logps": -123.38658905029297,
      "eval_loss": 0.3974739611148834,
      "eval_objective": 0.3952558934688568,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5740165710449219,
      "eval_regularize": 0.3952558934688568,
      "eval_runtime": 308.2767,
      "eval_samples_per_second": 18.782,
      "eval_steps_per_second": 1.567,
      "eval_wo_beta": 23.67264175415039,
      "step": 300
    },
    {
      "epoch": 0.8502598016060463,
      "step": 300,
      "total_flos": 0.0,
      "train_loss": 0.3342974317073822,
      "train_runtime": 8266.6282,
      "train_samples_per_second": 30.727,
      "train_steps_per_second": 0.213
    }
  ],
  "logging_steps": 50,
  "max_steps": 1760,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "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": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}