{
  "best_metric": 0.04710844159126282,
  "best_model_checkpoint": "./mistral/22-04-24-Weni-WeniGPT-Agents-Mistral-1.0.0-SFT-1.0.24-DPO_Experiment on DPO with other hyperparameters and best SFT model of WeniGPT-2_max_steps-180_batch_8_2024-04-22_ppid_9/checkpoint-90",
  "epoch": 2.903225806451613,
  "eval_steps": 30,
  "global_step": 90,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 18.102449417114258,
      "learning_rate": 4.971264367816092e-06,
      "logits/chosen": -1.8189502954483032,
      "logits/rejected": -1.8384132385253906,
      "logps/chosen": -183.32998657226562,
      "logps/rejected": -217.82275390625,
      "loss": 0.6772,
      "rewards/accuracies": 0.3499999940395355,
      "rewards/chosen": 0.028210829943418503,
      "rewards/margins": 0.022311249747872353,
      "rewards/rejected": 0.0058995820581912994,
      "step": 10
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 20.25644302368164,
      "learning_rate": 4.71264367816092e-06,
      "logits/chosen": -1.8640209436416626,
      "logits/rejected": -1.8766177892684937,
      "logps/chosen": -152.9895477294922,
      "logps/rejected": -246.1815948486328,
      "loss": 0.4471,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.5268799662590027,
      "rewards/margins": 0.6505464315414429,
      "rewards/rejected": -0.12366630882024765,
      "step": 20
    },
    {
      "epoch": 0.967741935483871,
      "grad_norm": 23.647602081298828,
      "learning_rate": 4.42528735632184e-06,
      "logits/chosen": -1.8179765939712524,
      "logits/rejected": -1.8386437892913818,
      "logps/chosen": -154.68057250976562,
      "logps/rejected": -222.60986328125,
      "loss": 0.2964,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 1.1595659255981445,
      "rewards/margins": 1.5337896347045898,
      "rewards/rejected": -0.3742235600948334,
      "step": 30
    },
    {
      "epoch": 0.967741935483871,
      "eval_logits/chosen": -1.922228217124939,
      "eval_logits/rejected": -1.9377918243408203,
      "eval_logps/chosen": -190.84751892089844,
      "eval_logps/rejected": -194.19583129882812,
      "eval_loss": 0.25011441111564636,
      "eval_rewards/accuracies": 0.8571428656578064,
      "eval_rewards/chosen": 0.8715044856071472,
      "eval_rewards/margins": 1.0091807842254639,
      "eval_rewards/rejected": -0.13767634332180023,
      "eval_runtime": 15.9917,
      "eval_samples_per_second": 1.751,
      "eval_steps_per_second": 0.438,
      "step": 30
    },
    {
      "epoch": 1.2903225806451613,
      "grad_norm": 14.360973358154297,
      "learning_rate": 4.137931034482759e-06,
      "logits/chosen": -1.7312577962875366,
      "logits/rejected": -1.720571517944336,
      "logps/chosen": -133.35958862304688,
      "logps/rejected": -174.8995361328125,
      "loss": 0.1702,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 2.0476880073547363,
      "rewards/margins": 2.6601643562316895,
      "rewards/rejected": -0.6124762296676636,
      "step": 40
    },
    {
      "epoch": 1.6129032258064515,
      "grad_norm": 7.5603814125061035,
      "learning_rate": 3.850574712643678e-06,
      "logits/chosen": -1.814079999923706,
      "logits/rejected": -1.8412790298461914,
      "logps/chosen": -135.43051147460938,
      "logps/rejected": -218.2666473388672,
      "loss": 0.1671,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 2.6564764976501465,
      "rewards/margins": 3.553129196166992,
      "rewards/rejected": -0.8966524004936218,
      "step": 50
    },
    {
      "epoch": 1.935483870967742,
      "grad_norm": 1.6607310771942139,
      "learning_rate": 3.563218390804598e-06,
      "logits/chosen": -1.8398892879486084,
      "logits/rejected": -1.8503402471542358,
      "logps/chosen": -147.40121459960938,
      "logps/rejected": -175.20462036132812,
      "loss": 0.1148,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 2.9025120735168457,
      "rewards/margins": 3.795964002609253,
      "rewards/rejected": -0.8934518098831177,
      "step": 60
    },
    {
      "epoch": 1.935483870967742,
      "eval_logits/chosen": -1.933964729309082,
      "eval_logits/rejected": -1.9501203298568726,
      "eval_logps/chosen": -186.97340393066406,
      "eval_logps/rejected": -197.0532989501953,
      "eval_loss": 0.11131088435649872,
      "eval_rewards/accuracies": 0.8571428656578064,
      "eval_rewards/chosen": 1.6463254690170288,
      "eval_rewards/margins": 2.355492115020752,
      "eval_rewards/rejected": -0.7091667056083679,
      "eval_runtime": 15.9564,
      "eval_samples_per_second": 1.755,
      "eval_steps_per_second": 0.439,
      "step": 60
    },
    {
      "epoch": 2.258064516129032,
      "grad_norm": 1.0933343172073364,
      "learning_rate": 3.2758620689655175e-06,
      "logits/chosen": -1.8214714527130127,
      "logits/rejected": -1.8398475646972656,
      "logps/chosen": -123.23341369628906,
      "logps/rejected": -169.8323974609375,
      "loss": 0.0958,
      "rewards/accuracies": 0.8999999761581421,
      "rewards/chosen": 3.013068914413452,
      "rewards/margins": 3.9151065349578857,
      "rewards/rejected": -0.9020377993583679,
      "step": 70
    },
    {
      "epoch": 2.5806451612903225,
      "grad_norm": 1.6529628038406372,
      "learning_rate": 2.988505747126437e-06,
      "logits/chosen": -1.8991115093231201,
      "logits/rejected": -1.9303562641143799,
      "logps/chosen": -126.10279846191406,
      "logps/rejected": -249.8067169189453,
      "loss": 0.0542,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 2.2060089111328125,
      "rewards/margins": 4.4772820472717285,
      "rewards/rejected": -2.271273136138916,
      "step": 80
    },
    {
      "epoch": 2.903225806451613,
      "grad_norm": 1.3197133541107178,
      "learning_rate": 2.7011494252873567e-06,
      "logits/chosen": -1.9221817255020142,
      "logits/rejected": -1.926163911819458,
      "logps/chosen": -130.5477752685547,
      "logps/rejected": -212.64401245117188,
      "loss": 0.0655,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 3.013507127761841,
      "rewards/margins": 4.679448127746582,
      "rewards/rejected": -1.665941596031189,
      "step": 90
    },
    {
      "epoch": 2.903225806451613,
      "eval_logits/chosen": -1.9456537961959839,
      "eval_logits/rejected": -1.9615535736083984,
      "eval_logps/chosen": -184.64419555664062,
      "eval_logps/rejected": -200.91307067871094,
      "eval_loss": 0.04710844159126282,
      "eval_rewards/accuracies": 1.0,
      "eval_rewards/chosen": 2.112164258956909,
      "eval_rewards/margins": 3.5932810306549072,
      "eval_rewards/rejected": -1.4811171293258667,
      "eval_runtime": 15.9489,
      "eval_samples_per_second": 1.756,
      "eval_steps_per_second": 0.439,
      "step": 90
    }
  ],
  "logging_steps": 10,
  "max_steps": 180,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 90,
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}