{
  "best_metric": 9.357504844665527,
  "best_model_checkpoint": "./qwen2.5-0.5b/qwen2.5-0.5b-expo-DPO-ES-0.01/checkpoint-50",
  "epoch": 1.5588096362777515,
  "eval_steps": 50,
  "global_step": 550,
  "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": 1.8442555137313956,
      "learning_rate": 2.840909090909091e-08,
      "logits": -1.359458565711975,
      "logps": -84.69721221923828,
      "loss": 0.6931,
      "objective": 0.6931471824645996,
      "ranking_idealized": 0.875,
      "ranking_idealized_expo": 0.5833333134651184,
      "ranking_simple": 0.5833333134651184,
      "regularize": 0.6931471824645996,
      "step": 1,
      "wo_beta": 5.271125316619873
    },
    {
      "dpo_loss": 0.6914740800857544,
      "epoch": 0.14170996693434104,
      "grad_norm": 2.883110438441685,
      "learning_rate": 1.4204545454545458e-06,
      "logits": -1.5272094011306763,
      "logps": -83.2472152709961,
      "loss": 0.6907,
      "objective": 0.6914740800857544,
      "ranking_idealized": 0.8945578336715698,
      "ranking_idealized_expo": 0.5348639488220215,
      "ranking_simple": 0.5293367505073547,
      "regularize": 0.6914740800857544,
      "step": 50,
      "wo_beta": 7.299497604370117
    },
    {
      "epoch": 0.14170996693434104,
      "eval_dpo_loss": 0.6902037858963013,
      "eval_logits": -1.7611440420150757,
      "eval_logps": -99.38214111328125,
      "eval_loss": 0.6893583536148071,
      "eval_objective": 0.6902037858963013,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5305383205413818,
      "eval_regularize": 0.6902037858963013,
      "eval_runtime": 314.0138,
      "eval_samples_per_second": 18.439,
      "eval_steps_per_second": 1.538,
      "eval_wo_beta": 9.357504844665527,
      "step": 50
    },
    {
      "dpo_loss": 0.6664106249809265,
      "epoch": 0.2834199338686821,
      "grad_norm": 7.452572178923395,
      "learning_rate": 2.8409090909090916e-06,
      "logits": -1.6917132139205933,
      "logps": -118.66804504394531,
      "loss": 0.6701,
      "objective": 0.6664106249809265,
      "ranking_idealized": 0.8958333134651184,
      "ranking_idealized_expo": 0.5270833373069763,
      "ranking_simple": 0.5866666436195374,
      "regularize": 0.6664106249809265,
      "step": 100,
      "wo_beta": 9.330028533935547
    },
    {
      "epoch": 0.2834199338686821,
      "eval_dpo_loss": 0.6895959377288818,
      "eval_logits": -1.8466589450836182,
      "eval_logps": -153.5715789794922,
      "eval_loss": 0.6837136149406433,
      "eval_objective": 0.6895959377288818,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5517598390579224,
      "eval_regularize": 0.6895959377288818,
      "eval_runtime": 308.2112,
      "eval_samples_per_second": 18.786,
      "eval_steps_per_second": 1.567,
      "eval_wo_beta": 14.174129486083984,
      "step": 100
    },
    {
      "dpo_loss": 0.6343104243278503,
      "epoch": 0.42512990080302315,
      "grad_norm": 13.296765037353357,
      "learning_rate": 4.2613636363636365e-06,
      "logits": -1.825290322303772,
      "logps": -173.7165985107422,
      "loss": 0.637,
      "objective": 0.6343104243278503,
      "ranking_idealized": 0.8862500190734863,
      "ranking_idealized_expo": 0.5358333587646484,
      "ranking_simple": 0.6316666603088379,
      "regularize": 0.6343104243278503,
      "step": 150,
      "wo_beta": 12.891901016235352
    },
    {
      "epoch": 0.42512990080302315,
      "eval_dpo_loss": 0.6764841079711914,
      "eval_logits": -2.461402416229248,
      "eval_logps": -198.92694091796875,
      "eval_loss": 0.6722760796546936,
      "eval_objective": 0.6764841079711914,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5822981595993042,
      "eval_regularize": 0.6764841079711914,
      "eval_runtime": 308.7795,
      "eval_samples_per_second": 18.751,
      "eval_steps_per_second": 1.564,
      "eval_wo_beta": 17.20980453491211,
      "step": 150
    },
    {
      "dpo_loss": 0.5824262499809265,
      "epoch": 0.5668398677373642,
      "grad_norm": 14.182378256685057,
      "learning_rate": 4.997168347957521e-06,
      "logits": -2.928657054901123,
      "logps": -216.66006469726562,
      "loss": 0.5833,
      "objective": 0.5824262499809265,
      "ranking_idealized": 0.8883333206176758,
      "ranking_idealized_expo": 0.527916669845581,
      "ranking_simple": 0.6833333373069763,
      "regularize": 0.5824262499809265,
      "step": 200,
      "wo_beta": 13.758678436279297
    },
    {
      "epoch": 0.5668398677373642,
      "eval_dpo_loss": 0.6780264377593994,
      "eval_logits": -3.3478028774261475,
      "eval_logps": -256.0312194824219,
      "eval_loss": 0.6728978157043457,
      "eval_objective": 0.6780264377593994,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5797101259231567,
      "eval_regularize": 0.6780264377593994,
      "eval_runtime": 312.7328,
      "eval_samples_per_second": 18.514,
      "eval_steps_per_second": 1.544,
      "eval_wo_beta": 21.710927963256836,
      "step": 200
    },
    {
      "dpo_loss": 0.541719377040863,
      "epoch": 0.7085498346717053,
      "grad_norm": 16.105604013075503,
      "learning_rate": 4.973122855144066e-06,
      "logits": -3.4248011112213135,
      "logps": -254.68023681640625,
      "loss": 0.5439,
      "objective": 0.541719377040863,
      "ranking_idealized": 0.8787500262260437,
      "ranking_idealized_expo": 0.5266666412353516,
      "ranking_simple": 0.7054166793823242,
      "regularize": 0.541719377040863,
      "step": 250,
      "wo_beta": 15.057023048400879
    },
    {
      "epoch": 0.7085498346717053,
      "eval_dpo_loss": 0.6857688426971436,
      "eval_logits": -3.626901865005493,
      "eval_logps": -257.3546447753906,
      "eval_loss": 0.6781259775161743,
      "eval_objective": 0.6857688426971436,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5683229565620422,
      "eval_regularize": 0.6857688426971436,
      "eval_runtime": 308.1728,
      "eval_samples_per_second": 18.788,
      "eval_steps_per_second": 1.567,
      "eval_wo_beta": 22.9139461517334,
      "step": 250
    },
    {
      "dpo_loss": 0.49844247102737427,
      "epoch": 0.8502598016060463,
      "grad_norm": 14.80120970807248,
      "learning_rate": 4.924776641419513e-06,
      "logits": -4.0557355880737305,
      "logps": -292.7536315917969,
      "loss": 0.5077,
      "objective": 0.49844247102737427,
      "ranking_idealized": 0.8733333349227905,
      "ranking_idealized_expo": 0.5104166865348816,
      "ranking_simple": 0.747083306312561,
      "regularize": 0.49844247102737427,
      "step": 300,
      "wo_beta": 13.577784538269043
    },
    {
      "epoch": 0.8502598016060463,
      "eval_dpo_loss": 0.6684697866439819,
      "eval_logits": -4.609988689422607,
      "eval_logps": -319.3935241699219,
      "eval_loss": 0.663960874080658,
      "eval_objective": 0.6684697866439819,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5828157067298889,
      "eval_regularize": 0.6684697866439819,
      "eval_runtime": 309.749,
      "eval_samples_per_second": 18.693,
      "eval_steps_per_second": 1.559,
      "eval_wo_beta": 23.650588989257812,
      "step": 300
    },
    {
      "dpo_loss": 0.4693358838558197,
      "epoch": 0.9919697685403873,
      "grad_norm": 17.283303885786346,
      "learning_rate": 4.8526047530778175e-06,
      "logits": -4.669376373291016,
      "logps": -331.3940124511719,
      "loss": 0.4786,
      "objective": 0.4693358838558197,
      "ranking_idealized": 0.8804166913032532,
      "ranking_idealized_expo": 0.5387499928474426,
      "ranking_simple": 0.7683333158493042,
      "regularize": 0.4693358838558197,
      "step": 350,
      "wo_beta": 12.894118309020996
    },
    {
      "epoch": 0.9919697685403873,
      "eval_dpo_loss": 0.6896602511405945,
      "eval_logits": -4.919209957122803,
      "eval_logps": -368.701416015625,
      "eval_loss": 0.6866692304611206,
      "eval_objective": 0.6896602511405945,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5750517845153809,
      "eval_regularize": 0.6896602511405945,
      "eval_runtime": 308.8788,
      "eval_samples_per_second": 18.745,
      "eval_steps_per_second": 1.564,
      "eval_wo_beta": 27.71128273010254,
      "step": 350
    },
    {
      "dpo_loss": 0.3561079800128937,
      "epoch": 1.1336797354747283,
      "grad_norm": 15.512579363366164,
      "learning_rate": 4.757316345716554e-06,
      "logits": -4.89206075668335,
      "logps": -377.2569580078125,
      "loss": 0.3619,
      "objective": 0.3561079800128937,
      "ranking_idealized": 0.8895833492279053,
      "ranking_idealized_expo": 0.5450000166893005,
      "ranking_simple": 0.8650000095367432,
      "regularize": 0.3561079800128937,
      "step": 400,
      "wo_beta": 8.194127082824707
    },
    {
      "epoch": 1.1336797354747283,
      "eval_dpo_loss": 0.6989732384681702,
      "eval_logits": -5.580094814300537,
      "eval_logps": -392.70184326171875,
      "eval_loss": 0.6960746049880981,
      "eval_objective": 0.6989732384681702,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5848861336708069,
      "eval_regularize": 0.6989732384681702,
      "eval_runtime": 308.969,
      "eval_samples_per_second": 18.74,
      "eval_steps_per_second": 1.563,
      "eval_wo_beta": 32.072998046875,
      "step": 400
    },
    {
      "dpo_loss": 0.3608836829662323,
      "epoch": 1.2753897024090695,
      "grad_norm": 16.08489095978575,
      "learning_rate": 4.639847716126855e-06,
      "logits": -5.059500217437744,
      "logps": -364.7832336425781,
      "loss": 0.3679,
      "objective": 0.3608836829662323,
      "ranking_idealized": 0.8845833539962769,
      "ranking_idealized_expo": 0.5266666412353516,
      "ranking_simple": 0.8412500023841858,
      "regularize": 0.3608836829662323,
      "step": 450,
      "wo_beta": 9.603814125061035
    },
    {
      "epoch": 1.2753897024090695,
      "eval_dpo_loss": 0.6952692866325378,
      "eval_logits": -5.245004177093506,
      "eval_logps": -349.20294189453125,
      "eval_loss": 0.6843227744102478,
      "eval_objective": 0.6952692866325378,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5885093212127686,
      "eval_regularize": 0.6952692866325378,
      "eval_runtime": 308.243,
      "eval_samples_per_second": 18.784,
      "eval_steps_per_second": 1.567,
      "eval_wo_beta": 31.319866180419922,
      "step": 450
    },
    {
      "dpo_loss": 0.3647218942642212,
      "epoch": 1.4170996693434104,
      "grad_norm": 16.426230236098732,
      "learning_rate": 4.501353102310901e-06,
      "logits": -5.431642055511475,
      "logps": -371.9469299316406,
      "loss": 0.3662,
      "objective": 0.3647218942642212,
      "ranking_idealized": 0.877916693687439,
      "ranking_idealized_expo": 0.5095833539962769,
      "ranking_simple": 0.8420833349227905,
      "regularize": 0.3647218942642212,
      "step": 500,
      "wo_beta": 10.912171363830566
    },
    {
      "epoch": 1.4170996693434104,
      "eval_dpo_loss": 0.6903234720230103,
      "eval_logits": -5.823331832885742,
      "eval_logps": -350.71368408203125,
      "eval_loss": 0.6858127117156982,
      "eval_objective": 0.6903234720230103,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.589026927947998,
      "eval_regularize": 0.6903234720230103,
      "eval_runtime": 308.1061,
      "eval_samples_per_second": 18.792,
      "eval_steps_per_second": 1.568,
      "eval_wo_beta": 30.27263641357422,
      "step": 500
    },
    {
      "dpo_loss": 0.3470539152622223,
      "epoch": 1.5588096362777515,
      "grad_norm": 13.971929990602174,
      "learning_rate": 4.34319334202531e-06,
      "logits": -5.66475248336792,
      "logps": -394.5787048339844,
      "loss": 0.3485,
      "objective": 0.3470539152622223,
      "ranking_idealized": 0.8870833516120911,
      "ranking_idealized_expo": 0.5220833420753479,
      "ranking_simple": 0.8383333086967468,
      "regularize": 0.3470539152622223,
      "step": 550,
      "wo_beta": 10.396123886108398
    },
    {
      "epoch": 1.5588096362777515,
      "eval_dpo_loss": 0.6972023844718933,
      "eval_logits": -5.249249458312988,
      "eval_logps": -349.5347900390625,
      "eval_loss": 0.6901025772094727,
      "eval_objective": 0.6972023844718933,
      "eval_ranking_idealized": 0.8731883764266968,
      "eval_ranking_idealized_expo": 0.5320910811424255,
      "eval_ranking_simple": 0.5719461441040039,
      "eval_regularize": 0.6972023844718933,
      "eval_runtime": 309.5507,
      "eval_samples_per_second": 18.705,
      "eval_steps_per_second": 1.56,
      "eval_wo_beta": 31.17648696899414,
      "step": 550
    },
    {
      "epoch": 1.5588096362777515,
      "step": 550,
      "total_flos": 0.0,
      "train_loss": 0.17483963706276634,
      "train_runtime": 6877.8936,
      "train_samples_per_second": 36.931,
      "train_steps_per_second": 0.256
    }
  ],
  "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
}