all_results.json

{
    "epoch": 3.0,
    "eval_logits/chosen": -0.33491453528404236,
    "eval_logits/rejected": -0.3404884338378906,
    "eval_logps/chosen": -12.90090274810791,
    "eval_logps/rejected": -19.88900375366211,
    "eval_loss": 0.8390035629272461,
    "eval_rewards/accuracies": 0.6317567825317383,
    "eval_rewards/chosen": -0.5836451649665833,
    "eval_rewards/margins": 0.680999755859375,
    "eval_rewards/rejected": -1.2646448612213135,
    "eval_runtime": 29.3135,
    "eval_samples_per_second": 10.098,
    "eval_steps_per_second": 2.524,
    "total_flos": 11469721436160.0,
    "train_loss": 0.693384567896525,
    "train_runtime": 1185.9878,
    "train_samples_per_second": 6.716,
    "train_steps_per_second": 0.21
}