all_results.json

{
    "epoch": 2.0,
    "eval_logits/chosen": -2.962392568588257,
    "eval_logits/rejected": -2.956583023071289,
    "eval_logps/chosen": -70.84989929199219,
    "eval_logps/rejected": -78.1531753540039,
    "eval_loss": 0.6807694435119629,
    "eval_rewards/accuracies": 0.6089683771133423,
    "eval_rewards/chosen": -0.12137996405363083,
    "eval_rewards/margins": 0.028350669890642166,
    "eval_rewards/rejected": -0.1497306227684021,
    "eval_runtime": 382.3469,
    "eval_samples": 4304,
    "eval_samples_per_second": 11.257,
    "eval_steps_per_second": 1.407,
    "total_flos": 0.0,
    "train_loss": 0.6777890619945723,
    "train_runtime": 94641.0715,
    "train_samples": 92858,
    "train_samples_per_second": 1.962,
    "train_steps_per_second": 0.123
}