all_results.json

{
    "epoch": 1.0,
    "eval_count/fg_chosen": 30.183246612548828,
    "eval_count/fg_rejected": 6.92391300201416,
    "eval_fg_kl": NaN,
    "eval_fg_logps/policy_KL": -14.863760948181152,
    "eval_fg_logps/policy_chosen": -6.796307563781738,
    "eval_fg_logps/policy_rejected": -8.7196044921875,
    "eval_fg_logps/reference_KL": -11.47359848022461,
    "eval_fg_logps/reference_chosen": -6.041894912719727,
    "eval_fg_logps/reference_rejected": -7.58065938949585,
    "eval_fg_loss": 0.7616299986839294,
    "eval_fg_rewards/chosen_sum": -1.7056890726089478,
    "eval_fg_rewards/rejected_sum": -0.9767768979072571,
    "eval_kl": 0.01342250220477581,
    "eval_logps/chosen": -337.8388459335624,
    "eval_logps/rejected": -409.12525347624563,
    "eval_loss": 0.4091770052909851,
    "eval_rewards/chosen": 1.55619121608319,
    "eval_rewards/margins": 4.119737088494111,
    "eval_rewards/rejected": -2.5635458724109212,
    "eval_runtime": 463.0413,
    "eval_samples": 1733,
    "eval_samples_per_second": 3.743,
    "eval_steps_per_second": 0.937,
    "total_flos": 0.0,
    "train_loss": 0.45996271347388246,
    "train_runtime": 8430.3956,
    "train_samples": 15597,
    "train_samples_per_second": 1.85,
    "train_steps_per_second": 0.116
}