{
  "best_metric": 0.46216800808906555,
  "best_model_checkpoint": "./mistral/19-04-24-Weni-WeniGPT-Agents-Mistral-1.0.6-SFT-1.0.4-DPO_Experiment on DPO with other hyperparameters and best SFT model of WeniGPT-2_max_steps-732_batch_4_2024-04-19_ppid_9/checkpoint-90",
  "epoch": 1.4634146341463414,
  "eval_steps": 30,
  "global_step": 90,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.16,
      "grad_norm": 38.021175384521484,
      "learning_rate": 1.590909090909091e-06,
      "logits/chosen": -1.7808929681777954,
      "logits/rejected": -1.8136215209960938,
      "logps/chosen": -52.2663688659668,
      "logps/rejected": -88.76155853271484,
      "loss": 0.6927,
      "rewards/accuracies": 0.20000000298023224,
      "rewards/chosen": 0.0008833981119096279,
      "rewards/margins": 0.0009520913590677083,
      "rewards/rejected": -6.869318895041943e-05,
      "step": 10
    },
    {
      "epoch": 0.33,
      "grad_norm": 19.505126953125,
      "learning_rate": 3.863636363636364e-06,
      "logits/chosen": -1.865220308303833,
      "logits/rejected": -1.880090355873108,
      "logps/chosen": -34.3599967956543,
      "logps/rejected": -53.656646728515625,
      "loss": 0.6846,
      "rewards/accuracies": 0.22499999403953552,
      "rewards/chosen": 0.019346294924616814,
      "rewards/margins": 0.017682768404483795,
      "rewards/rejected": 0.0016635276842862368,
      "step": 20
    },
    {
      "epoch": 0.49,
      "grad_norm": 16.67194938659668,
      "learning_rate": 4.964788732394366e-06,
      "logits/chosen": -1.7692855596542358,
      "logits/rejected": -1.7932851314544678,
      "logps/chosen": -38.679237365722656,
      "logps/rejected": -79.23713684082031,
      "loss": 0.665,
      "rewards/accuracies": 0.375,
      "rewards/chosen": 0.07158979028463364,
      "rewards/margins": 0.061105482280254364,
      "rewards/rejected": 0.010484304279088974,
      "step": 30
    },
    {
      "epoch": 0.49,
      "eval_logits/chosen": -1.7801028490066528,
      "eval_logits/rejected": -1.7981160879135132,
      "eval_logps/chosen": -54.71597671508789,
      "eval_logps/rejected": -95.75775909423828,
      "eval_loss": 0.6322354674339294,
      "eval_rewards/accuracies": 0.4285714328289032,
      "eval_rewards/chosen": 0.14569346606731415,
      "eval_rewards/margins": 0.13385750353336334,
      "eval_rewards/rejected": 0.011835969984531403,
      "eval_runtime": 8.5124,
      "eval_samples_per_second": 3.289,
      "eval_steps_per_second": 1.645,
      "step": 30
    },
    {
      "epoch": 0.65,
      "grad_norm": 22.69721031188965,
      "learning_rate": 4.894366197183099e-06,
      "logits/chosen": -1.8458446264266968,
      "logits/rejected": -1.8594070672988892,
      "logps/chosen": -39.180992126464844,
      "logps/rejected": -57.04741287231445,
      "loss": 0.6164,
      "rewards/accuracies": 0.3499999940395355,
      "rewards/chosen": 0.19042351841926575,
      "rewards/margins": 0.18086455762386322,
      "rewards/rejected": 0.009558964520692825,
      "step": 40
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.0,
      "learning_rate": 4.823943661971832e-06,
      "logits/chosen": -1.8411815166473389,
      "logits/rejected": -1.8659855127334595,
      "logps/chosen": -31.194744110107422,
      "logps/rejected": -58.17326736450195,
      "loss": 0.6138,
      "rewards/accuracies": 0.30000001192092896,
      "rewards/chosen": 0.19823171198368073,
      "rewards/margins": 0.19669394195079803,
      "rewards/rejected": 0.0015377718955278397,
      "step": 50
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.77029800415039,
      "learning_rate": 4.753521126760564e-06,
      "logits/chosen": -1.816936731338501,
      "logits/rejected": -1.8421777486801147,
      "logps/chosen": -50.517478942871094,
      "logps/rejected": -80.6562728881836,
      "loss": 0.5445,
      "rewards/accuracies": 0.42500001192092896,
      "rewards/chosen": 0.4567543864250183,
      "rewards/margins": 0.42180705070495605,
      "rewards/rejected": 0.03494739532470703,
      "step": 60
    },
    {
      "epoch": 0.98,
      "eval_logits/chosen": -1.784444808959961,
      "eval_logits/rejected": -1.8030898571014404,
      "eval_logps/chosen": -53.55432891845703,
      "eval_logps/rejected": -95.73511505126953,
      "eval_loss": 0.527186930179596,
      "eval_rewards/accuracies": 0.4285714328289032,
      "eval_rewards/chosen": 0.4941868484020233,
      "eval_rewards/margins": 0.47555941343307495,
      "eval_rewards/rejected": 0.018627464771270752,
      "eval_runtime": 8.5149,
      "eval_samples_per_second": 3.288,
      "eval_steps_per_second": 1.644,
      "step": 60
    },
    {
      "epoch": 1.14,
      "grad_norm": 8.583382606506348,
      "learning_rate": 4.683098591549296e-06,
      "logits/chosen": -1.7579460144042969,
      "logits/rejected": -1.776581048965454,
      "logps/chosen": -54.420066833496094,
      "logps/rejected": -79.0608901977539,
      "loss": 0.451,
      "rewards/accuracies": 0.4749999940395355,
      "rewards/chosen": 0.8353117108345032,
      "rewards/margins": 0.8680375814437866,
      "rewards/rejected": -0.032725896686315536,
      "step": 70
    },
    {
      "epoch": 1.3,
      "grad_norm": 2.3610572814941406,
      "learning_rate": 4.612676056338028e-06,
      "logits/chosen": -1.8833833932876587,
      "logits/rejected": -1.9092212915420532,
      "logps/chosen": -25.5800724029541,
      "logps/rejected": -64.96099853515625,
      "loss": 0.5302,
      "rewards/accuracies": 0.30000001192092896,
      "rewards/chosen": 0.5431563258171082,
      "rewards/margins": 0.6763826012611389,
      "rewards/rejected": -0.13322624564170837,
      "step": 80
    },
    {
      "epoch": 1.46,
      "grad_norm": 3.739367723464966,
      "learning_rate": 4.542253521126761e-06,
      "logits/chosen": -1.8194096088409424,
      "logits/rejected": -1.8343127965927124,
      "logps/chosen": -41.80801773071289,
      "logps/rejected": -63.005592346191406,
      "loss": 0.4616,
      "rewards/accuracies": 0.4749999940395355,
      "rewards/chosen": 0.830731213092804,
      "rewards/margins": 0.9403120875358582,
      "rewards/rejected": -0.109580859541893,
      "step": 90
    },
    {
      "epoch": 1.46,
      "eval_logits/chosen": -1.7902206182479858,
      "eval_logits/rejected": -1.8099161386489868,
      "eval_logps/chosen": -52.01105880737305,
      "eval_logps/rejected": -95.83198547363281,
      "eval_loss": 0.46216800808906555,
      "eval_rewards/accuracies": 0.4285714328289032,
      "eval_rewards/chosen": 0.957168459892273,
      "eval_rewards/margins": 0.9676021337509155,
      "eval_rewards/rejected": -0.010433738119900227,
      "eval_runtime": 8.5144,
      "eval_samples_per_second": 3.289,
      "eval_steps_per_second": 1.644,
      "step": 90
    }
  ],
  "logging_steps": 10,
  "max_steps": 732,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 12,
  "save_steps": 90,
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}