all_results.json

{
    "epoch": 2.981366459627329,
    "eval_logits/chosen": -2.5412092208862305,
    "eval_logits/rejected": -2.5580978393554688,
    "eval_logps/chosen": -253.88226318359375,
    "eval_logps/rejected": -211.97067260742188,
    "eval_loss": 0.6183735132217407,
    "eval_rewards/accuracies": 0.7708333134651184,
    "eval_rewards/chosen": 0.7816598415374756,
    "eval_rewards/margins": 1.6348735094070435,
    "eval_rewards/rejected": -0.8532136082649231,
    "eval_runtime": 74.5085,
    "eval_samples_per_second": 15.3,
    "eval_steps_per_second": 0.242,
    "total_flos": 1414680891359232.0,
    "train_loss": 0.3313312023878098,
    "train_runtime": 4261.9082,
    "train_samples_per_second": 7.22,
    "train_steps_per_second": 0.028
}