all_results.json

{
    "epoch": 2.999297541394882,
    "eval_logits/chosen": 0.11797090619802475,
    "eval_logits/rejected": 0.22131487727165222,
    "eval_logps/chosen": -1.7544060945510864,
    "eval_logps/rejected": -2.372403860092163,
    "eval_loss": 0.4698396623134613,
    "eval_rewards/accuracies": 0.6839762330055237,
    "eval_rewards/chosen": -1.7544060945510864,
    "eval_rewards/margins": 0.6179976463317871,
    "eval_rewards/rejected": -2.372403860092163,
    "eval_runtime": 40.4595,
    "eval_samples": 1345,
    "eval_samples_per_second": 33.243,
    "eval_steps_per_second": 8.329,
    "total_flos": 0.0,
    "train_loss": 0.4793110236919071,
    "train_runtime": 30370.8284,
    "train_samples": 59790,
    "train_samples_per_second": 5.906,
    "train_steps_per_second": 0.185
}