{
  "best_metric": 0.43594828248023987,
  "best_model_checkpoint": "./Mistral/13-03-24-Weni-ZeroShot-3.4.6-Mistral-7b-DPO-1.0.0_ZeroShot DPO Training a improved dataset and best hyperparameters found so far-2_max_steps-288_batch_32_2024-03-13_ppid_9/checkpoint-200",
  "epoch": 4.081632653061225,
  "eval_steps": 100,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.41,
      "grad_norm": 4.933152675628662,
      "learning_rate": 1.3793103448275862e-06,
      "logits/chosen": -1.3310586214065552,
      "logits/rejected": -1.2998794317245483,
      "logps/chosen": -17.44550895690918,
      "logps/rejected": -15.287511825561523,
      "loss": 0.6887,
      "rewards/accuracies": 0.6078125238418579,
      "rewards/chosen": 0.012908421456813812,
      "rewards/margins": 0.00912781897932291,
      "rewards/rejected": 0.0037806027103215456,
      "step": 20
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.840120553970337,
      "learning_rate": 1.915057915057915e-06,
      "logits/chosen": -1.342775583267212,
      "logits/rejected": -1.3136231899261475,
      "logps/chosen": -16.57306480407715,
      "logps/rejected": -15.076986312866211,
      "loss": 0.6579,
      "rewards/accuracies": 0.7734375,
      "rewards/chosen": 0.10680261999368668,
      "rewards/margins": 0.08031970262527466,
      "rewards/rejected": 0.02648291550576687,
      "step": 40
    },
    {
      "epoch": 1.22,
      "grad_norm": 2.2041518688201904,
      "learning_rate": 1.7606177606177606e-06,
      "logits/chosen": -1.3282166719436646,
      "logits/rejected": -1.300879716873169,
      "logps/chosen": -15.91553783416748,
      "logps/rejected": -15.11401653289795,
      "loss": 0.6314,
      "rewards/accuracies": 0.785937488079071,
      "rewards/chosen": 0.16994965076446533,
      "rewards/margins": 0.14337585866451263,
      "rewards/rejected": 0.026573771610856056,
      "step": 60
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4858603477478027,
      "learning_rate": 1.606177606177606e-06,
      "logits/chosen": -1.3553317785263062,
      "logits/rejected": -1.3338539600372314,
      "logps/chosen": -15.328280448913574,
      "logps/rejected": -15.158895492553711,
      "loss": 0.6066,
      "rewards/accuracies": 0.768750011920929,
      "rewards/chosen": 0.2215496301651001,
      "rewards/margins": 0.20472590625286102,
      "rewards/rejected": 0.016823694109916687,
      "step": 80
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4387667179107666,
      "learning_rate": 1.4517374517374517e-06,
      "logits/chosen": -1.3066260814666748,
      "logits/rejected": -1.282898187637329,
      "logps/chosen": -14.973272323608398,
      "logps/rejected": -15.511428833007812,
      "loss": 0.5697,
      "rewards/accuracies": 0.793749988079071,
      "rewards/chosen": 0.2805514931678772,
      "rewards/margins": 0.2973305881023407,
      "rewards/rejected": -0.016779109835624695,
      "step": 100
    },
    {
      "epoch": 2.04,
      "eval_logits/chosen": -1.3297994136810303,
      "eval_logits/rejected": -1.3079336881637573,
      "eval_logps/chosen": -14.785962104797363,
      "eval_logps/rejected": -15.702004432678223,
      "eval_loss": 0.5667449831962585,
      "eval_rewards/accuracies": 0.7651515603065491,
      "eval_rewards/chosen": 0.30174848437309265,
      "eval_rewards/margins": 0.31671077013015747,
      "eval_rewards/rejected": -0.014962326735258102,
      "eval_runtime": 81.5286,
      "eval_samples_per_second": 2.134,
      "eval_steps_per_second": 0.27,
      "step": 100
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5866434574127197,
      "learning_rate": 1.2972972972972972e-06,
      "logits/chosen": -1.3086926937103271,
      "logits/rejected": -1.2873843908309937,
      "logps/chosen": -14.411561965942383,
      "logps/rejected": -15.736058235168457,
      "loss": 0.5444,
      "rewards/accuracies": 0.8140624761581421,
      "rewards/chosen": 0.32512110471725464,
      "rewards/margins": 0.36463481187820435,
      "rewards/rejected": -0.0395137183368206,
      "step": 120
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9121856689453125,
      "learning_rate": 1.1428571428571428e-06,
      "logits/chosen": -1.277313470840454,
      "logits/rejected": -1.2543872594833374,
      "logps/chosen": -13.582036972045898,
      "logps/rejected": -15.926782608032227,
      "loss": 0.5144,
      "rewards/accuracies": 0.831250011920929,
      "rewards/chosen": 0.3723224699497223,
      "rewards/margins": 0.4507189691066742,
      "rewards/rejected": -0.07839655876159668,
      "step": 140
    },
    {
      "epoch": 3.27,
      "grad_norm": 2.457465171813965,
      "learning_rate": 9.884169884169883e-07,
      "logits/chosen": -1.266373872756958,
      "logits/rejected": -1.2449392080307007,
      "logps/chosen": -13.217763900756836,
      "logps/rejected": -16.542827606201172,
      "loss": 0.4729,
      "rewards/accuracies": 0.8453124761581421,
      "rewards/chosen": 0.4651583135128021,
      "rewards/margins": 0.5779576301574707,
      "rewards/rejected": -0.11279928684234619,
      "step": 160
    },
    {
      "epoch": 3.67,
      "grad_norm": 4.268860340118408,
      "learning_rate": 8.33976833976834e-07,
      "logits/chosen": -1.242867112159729,
      "logits/rejected": -1.220132827758789,
      "logps/chosen": -12.452630996704102,
      "logps/rejected": -16.665748596191406,
      "loss": 0.4601,
      "rewards/accuracies": 0.8578125238418579,
      "rewards/chosen": 0.503332257270813,
      "rewards/margins": 0.6378341913223267,
      "rewards/rejected": -0.1345018595457077,
      "step": 180
    },
    {
      "epoch": 4.08,
      "grad_norm": 2.2548317909240723,
      "learning_rate": 6.795366795366795e-07,
      "logits/chosen": -1.232925295829773,
      "logits/rejected": -1.210792899131775,
      "logps/chosen": -12.247258186340332,
      "logps/rejected": -17.310483932495117,
      "loss": 0.4274,
      "rewards/accuracies": 0.8828125,
      "rewards/chosen": 0.5342787504196167,
      "rewards/margins": 0.7390083074569702,
      "rewards/rejected": -0.2047295868396759,
      "step": 200
    },
    {
      "epoch": 4.08,
      "eval_logits/chosen": -1.2476983070373535,
      "eval_logits/rejected": -1.2272884845733643,
      "eval_logps/chosen": -12.130993843078613,
      "eval_logps/rejected": -17.42223358154297,
      "eval_loss": 0.43594828248023987,
      "eval_rewards/accuracies": 0.8446969985961914,
      "eval_rewards/chosen": 0.5672450661659241,
      "eval_rewards/margins": 0.7542301416397095,
      "eval_rewards/rejected": -0.1869850754737854,
      "eval_runtime": 81.5978,
      "eval_samples_per_second": 2.132,
      "eval_steps_per_second": 0.27,
      "step": 200
    }
  ],
  "logging_steps": 20,
  "max_steps": 288,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}