trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9905956112852664,
  "eval_steps": 500,
  "global_step": 79,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.012539184952978056,
      "grad_norm": 7.019775825313518,
      "learning_rate": 6.25e-08,
      "logits/chosen": -2.8930726051330566,
      "logits/rejected": -2.8629798889160156,
      "logps/chosen": -357.41943359375,
      "logps/rejected": -303.773193359375,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.12539184952978055,
      "grad_norm": 7.210056679422731,
      "learning_rate": 4.990217055187362e-07,
      "logits/chosen": -2.7940280437469482,
      "logits/rejected": -2.7685537338256836,
      "logps/chosen": -260.2227783203125,
      "logps/rejected": -250.45333862304688,
      "loss": 0.692,
      "rewards/accuracies": 0.5034722089767456,
      "rewards/chosen": 0.0014079404063522816,
      "rewards/margins": 0.0026502651162445545,
      "rewards/rejected": -0.0012423248263075948,
      "step": 10
    },
    {
      "epoch": 0.2507836990595611,
      "grad_norm": 6.938600485483928,
      "learning_rate": 4.655786431300069e-07,
      "logits/chosen": -2.7941102981567383,
      "logits/rejected": -2.7533040046691895,
      "logps/chosen": -273.05743408203125,
      "logps/rejected": -248.7327880859375,
      "loss": 0.6766,
      "rewards/accuracies": 0.6499999761581421,
      "rewards/chosen": 0.03514797240495682,
      "rewards/margins": 0.03713010624051094,
      "rewards/rejected": -0.0019821308087557554,
      "step": 20
    },
    {
      "epoch": 0.3761755485893417,
      "grad_norm": 7.437427526131843,
      "learning_rate": 3.9061232191019517e-07,
      "logits/chosen": -2.7480266094207764,
      "logits/rejected": -2.7372658252716064,
      "logps/chosen": -258.54705810546875,
      "logps/rejected": -261.3839416503906,
      "loss": 0.6524,
      "rewards/accuracies": 0.6812499761581421,
      "rewards/chosen": -0.0330645926296711,
      "rewards/margins": 0.083428755402565,
      "rewards/rejected": -0.1164933443069458,
      "step": 30
    },
    {
      "epoch": 0.5015673981191222,
      "grad_norm": 9.260816868546575,
      "learning_rate": 2.8856223324132555e-07,
      "logits/chosen": -2.7586328983306885,
      "logits/rejected": -2.7407941818237305,
      "logps/chosen": -277.0296325683594,
      "logps/rejected": -269.89105224609375,
      "loss": 0.6274,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.15935972332954407,
      "rewards/margins": 0.19388534128665924,
      "rewards/rejected": -0.3532450497150421,
      "step": 40
    },
    {
      "epoch": 0.6269592476489029,
      "grad_norm": 13.440996020894056,
      "learning_rate": 1.7908455541642582e-07,
      "logits/chosen": -2.758403778076172,
      "logits/rejected": -2.735839366912842,
      "logps/chosen": -311.75067138671875,
      "logps/rejected": -301.5154724121094,
      "loss": 0.6091,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": -0.2620142996311188,
      "rewards/margins": 0.27029839158058167,
      "rewards/rejected": -0.5323126316070557,
      "step": 50
    },
    {
      "epoch": 0.7523510971786834,
      "grad_norm": 10.982206859063186,
      "learning_rate": 8.32661172908373e-08,
      "logits/chosen": -2.7525084018707275,
      "logits/rejected": -2.7330567836761475,
      "logps/chosen": -266.4403381347656,
      "logps/rejected": -292.4945373535156,
      "loss": 0.5938,
      "rewards/accuracies": 0.7406250238418579,
      "rewards/chosen": -0.28137117624282837,
      "rewards/margins": 0.3296356201171875,
      "rewards/rejected": -0.6110068559646606,
      "step": 60
    },
    {
      "epoch": 0.877742946708464,
      "grad_norm": 10.390233056901057,
      "learning_rate": 1.956279997278043e-08,
      "logits/chosen": -2.7587547302246094,
      "logits/rejected": -2.733307123184204,
      "logps/chosen": -305.1893615722656,
      "logps/rejected": -302.3923034667969,
      "loss": 0.5905,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.31602221727371216,
      "rewards/margins": 0.29854491353034973,
      "rewards/rejected": -0.6145671606063843,
      "step": 70
    },
    {
      "epoch": 0.9905956112852664,
      "step": 79,
      "total_flos": 0.0,
      "train_loss": 0.6319818134549298,
      "train_runtime": 2210.8428,
      "train_samples_per_second": 9.217,
      "train_steps_per_second": 0.036
    }
  ],
  "logging_steps": 10,
  "max_steps": 79,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}