{
  "best_metric": 0.529485821723938,
  "best_model_checkpoint": "./mistral/20-04-24-Weni-WeniGPT-Agents-Mistral-1.0.6-SFT-1.0.5-DPO_Experiment on DPO with other hyperparameters and best SFT model of WeniGPT-2_max_steps-366_batch_4_2024-04-20_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": 8.378021240234375,
      "learning_rate": 4.0909090909090915e-06,
      "logits/chosen": -1.830958604812622,
      "logits/rejected": -1.8507845401763916,
      "logps/chosen": -28.701984405517578,
      "logps/rejected": -54.28569793701172,
      "loss": 0.6924,
      "rewards/accuracies": 0.20000000298023224,
      "rewards/chosen": 0.0008967495523393154,
      "rewards/margins": 0.0014666033675894141,
      "rewards/rejected": -0.0005698538152500987,
      "step": 10
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.193418502807617,
      "learning_rate": 4.887323943661972e-06,
      "logits/chosen": -1.7550897598266602,
      "logits/rejected": -1.770708680152893,
      "logps/chosen": -47.344207763671875,
      "logps/rejected": -64.0368423461914,
      "loss": 0.6852,
      "rewards/accuracies": 0.4000000059604645,
      "rewards/chosen": 0.017231885343790054,
      "rewards/margins": 0.01606021076440811,
      "rewards/rejected": 0.0011716745793819427,
      "step": 20
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.308932304382324,
      "learning_rate": 4.746478873239437e-06,
      "logits/chosen": -1.781267762184143,
      "logits/rejected": -1.8114898204803467,
      "logps/chosen": -54.274559020996094,
      "logps/rejected": -95.20500183105469,
      "loss": 0.6635,
      "rewards/accuracies": 0.5,
      "rewards/chosen": 0.0641159638762474,
      "rewards/margins": 0.061691801995038986,
      "rewards/rejected": 0.0024241588544100523,
      "step": 30
    },
    {
      "epoch": 0.49,
      "eval_logits/chosen": -1.7831767797470093,
      "eval_logits/rejected": -1.8043663501739502,
      "eval_logps/chosen": -55.16960906982422,
      "eval_logps/rejected": -97.32585144042969,
      "eval_loss": 0.6523757576942444,
      "eval_rewards/accuracies": 0.4642857015132904,
      "eval_rewards/chosen": 0.09036973863840103,
      "eval_rewards/margins": 0.08673857897520065,
      "eval_rewards/rejected": 0.0036311547737568617,
      "eval_runtime": 8.141,
      "eval_samples_per_second": 3.439,
      "eval_steps_per_second": 1.72,
      "step": 30
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.0,
      "learning_rate": 4.6056338028169015e-06,
      "logits/chosen": -1.889905333518982,
      "logits/rejected": -1.9024461507797241,
      "logps/chosen": -27.918941497802734,
      "logps/rejected": -42.093284606933594,
      "loss": 0.668,
      "rewards/accuracies": 0.25,
      "rewards/chosen": 0.054457180202007294,
      "rewards/margins": 0.0539846234023571,
      "rewards/rejected": 0.0004725646285805851,
      "step": 40
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.53225326538086,
      "learning_rate": 4.464788732394367e-06,
      "logits/chosen": -1.8278567790985107,
      "logits/rejected": -1.849957823753357,
      "logps/chosen": -43.8238639831543,
      "logps/rejected": -68.02179718017578,
      "loss": 0.6358,
      "rewards/accuracies": 0.3499999940395355,
      "rewards/chosen": 0.13941256701946259,
      "rewards/margins": 0.13133978843688965,
      "rewards/rejected": 0.008072790689766407,
      "step": 50
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.436968803405762,
      "learning_rate": 4.3239436619718315e-06,
      "logits/chosen": -1.805991768836975,
      "logits/rejected": -1.8437427282333374,
      "logps/chosen": -43.8873291015625,
      "logps/rejected": -95.2943115234375,
      "loss": 0.6026,
      "rewards/accuracies": 0.44999998807907104,
      "rewards/chosen": 0.18793432414531708,
      "rewards/margins": 0.21308371424674988,
      "rewards/rejected": -0.025149401277303696,
      "step": 60
    },
    {
      "epoch": 0.98,
      "eval_logits/chosen": -1.7877694368362427,
      "eval_logits/rejected": -1.8098936080932617,
      "eval_logps/chosen": -53.567203521728516,
      "eval_logps/rejected": -97.33795928955078,
      "eval_loss": 0.5890871286392212,
      "eval_rewards/accuracies": 0.4642857015132904,
      "eval_rewards/chosen": 0.25061002373695374,
      "eval_rewards/margins": 0.2481890469789505,
      "eval_rewards/rejected": 0.002420984674245119,
      "eval_runtime": 8.1404,
      "eval_samples_per_second": 3.44,
      "eval_steps_per_second": 1.72,
      "step": 60
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.0,
      "learning_rate": 4.183098591549296e-06,
      "logits/chosen": -1.8344879150390625,
      "logits/rejected": -1.8489716053009033,
      "logps/chosen": -40.38930892944336,
      "logps/rejected": -60.9084358215332,
      "loss": 0.6031,
      "rewards/accuracies": 0.375,
      "rewards/chosen": 0.19739331305027008,
      "rewards/margins": 0.22638121247291565,
      "rewards/rejected": -0.028987903147935867,
      "step": 70
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.49536657333374,
      "learning_rate": 4.042253521126761e-06,
      "logits/chosen": -1.7903095483779907,
      "logits/rejected": -1.8362411260604858,
      "logps/chosen": -44.288116455078125,
      "logps/rejected": -90.21073913574219,
      "loss": 0.5357,
      "rewards/accuracies": 0.4749999940395355,
      "rewards/chosen": 0.34061312675476074,
      "rewards/margins": 0.40679749846458435,
      "rewards/rejected": -0.06618441641330719,
      "step": 80
    },
    {
      "epoch": 1.46,
      "grad_norm": 13.401692390441895,
      "learning_rate": 3.901408450704225e-06,
      "logits/chosen": -1.8004281520843506,
      "logits/rejected": -1.8247934579849243,
      "logps/chosen": -42.32465362548828,
      "logps/rejected": -70.9749984741211,
      "loss": 0.5387,
      "rewards/accuracies": 0.4749999940395355,
      "rewards/chosen": 0.3678433299064636,
      "rewards/margins": 0.4186524450778961,
      "rewards/rejected": -0.05080908536911011,
      "step": 90
    },
    {
      "epoch": 1.46,
      "eval_logits/chosen": -1.7943389415740967,
      "eval_logits/rejected": -1.8181126117706299,
      "eval_logps/chosen": -51.677486419677734,
      "eval_logps/rejected": -97.63689422607422,
      "eval_loss": 0.529485821723938,
      "eval_rewards/accuracies": 0.4642857015132904,
      "eval_rewards/chosen": 0.4395819306373596,
      "eval_rewards/margins": 0.4670555889606476,
      "eval_rewards/rejected": -0.027473628520965576,
      "eval_runtime": 8.1412,
      "eval_samples_per_second": 3.439,
      "eval_steps_per_second": 1.72,
      "step": 90
    }
  ],
  "logging_steps": 10,
  "max_steps": 366,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 90,
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}