all_results.json

{
    "epoch": 1.0,
    "eval_logits/chosen": -0.24525223672389984,
    "eval_logits/rejected": -0.1380803883075714,
    "eval_logps/chosen": -103.02954864501953,
    "eval_logps/rejected": -131.91891479492188,
    "eval_loss": 0.6906174421310425,
    "eval_pred_label": 4192.14306640625,
    "eval_rewards/accuracies": 0.363095223903656,
    "eval_rewards/chosen": -0.3412899374961853,
    "eval_rewards/margins": 0.22391849756240845,
    "eval_rewards/rejected": -0.5652084350585938,
    "eval_runtime": 247.5585,
    "eval_samples": 2000,
    "eval_samples_per_second": 8.079,
    "eval_steps_per_second": 0.254,
    "eval_use_label": 15879.857421875,
    "train_loss": 0.6880922077838039,
    "train_runtime": 20023.3666,
    "train_samples": 61135,
    "train_samples_per_second": 3.053,
    "train_steps_per_second": 0.048
}