all_results.json

{
    "epoch": 9.876543209876543,
    "eval_logits/chosen": -0.34801948070526123,
    "eval_logits/rejected": -0.4030517041683197,
    "eval_logps/chosen": -0.10840340703725815,
    "eval_logps/rejected": -0.5819076299667358,
    "eval_loss": 0.15150243043899536,
    "eval_odds_ratio_loss": 1.3828411102294922,
    "eval_rewards/accuracies": 0.8222222328186035,
    "eval_rewards/chosen": -0.010840341448783875,
    "eval_rewards/margins": 0.04735042154788971,
    "eval_rewards/rejected": -0.058190762996673584,
    "eval_runtime": 3.7487,
    "eval_samples_per_second": 24.008,
    "eval_sft_loss": 0.013218320906162262,
    "eval_steps_per_second": 12.004,
    "total_flos": 8.823432944104243e+16,
    "train_loss": 0.3188942909240723,
    "train_runtime": 1116.7498,
    "train_samples_per_second": 7.253,
    "train_steps_per_second": 0.448
}