all_results.json

{
    "epoch": 0.9991024490319271,
    "eval_count/fg_chosen": 16.01369857788086,
    "eval_count/fg_rejected": 4.0824174880981445,
    "eval_fg_kl": NaN,
    "eval_fg_logps/policy_KL": -21.24127197265625,
    "eval_fg_logps/policy_chosen": -15.721322059631348,
    "eval_fg_logps/policy_rejected": -20.08335304260254,
    "eval_fg_logps/reference_KL": -20.207021713256836,
    "eval_fg_logps/reference_chosen": -14.929487228393555,
    "eval_fg_logps/reference_rejected": -18.78682518005371,
    "eval_fg_loss": 0.738543689250946,
    "eval_fg_rewards/chosen_sum": -0.5872737169265747,
    "eval_fg_rewards/rejected_sum": -0.5185446739196777,
    "eval_kl": 0.005938735790550709,
    "eval_logps/chosen": -300.693661971831,
    "eval_logps/rejected": -365.812429138322,
    "eval_loss": 0.481334924697876,
    "eval_rewards/chosen": -0.7582848777233715,
    "eval_rewards/margins": 2.0290243493528757,
    "eval_rewards/rejected": -2.787309227076247,
    "eval_runtime": 811.484,
    "eval_samples": 1733,
    "eval_samples_per_second": 2.136,
    "eval_steps_per_second": 1.068,
    "total_flos": 0.0,
    "train_loss": 0.5029905087159644,
    "train_runtime": 13638.2416,
    "train_samples": 15597,
    "train_samples_per_second": 1.144,
    "train_steps_per_second": 0.071
}