all_results.json

{
    "epoch": 2.998430141287284,
    "eval_logits/chosen": -10.0625,
    "eval_logits/rejected": -8.75,
    "eval_logps/chosen": -752.0,
    "eval_logps/rejected": -824.0,
    "eval_loss": 0.6881640553474426,
    "eval_rewards/accuracies": 0.73828125,
    "eval_rewards/chosen": -4.34375,
    "eval_rewards/margins": 0.9921875,
    "eval_rewards/rejected": -5.34375,
    "eval_runtime": 46.6028,
    "eval_samples": 2000,
    "eval_samples_per_second": 42.916,
    "eval_steps_per_second": 0.687,
    "total_flos": 0.0,
    "train_loss": 0.0,
    "train_runtime": 0.0555,
    "train_samples": 61134,
    "train_samples_per_second": 3306285.721,
    "train_steps_per_second": 51648.884
}