all_results.json

{
    "epoch": 3.0,
    "eval_logits/chosen": -2.4882912635803223,
    "eval_logits/rejected": -2.058779001235962,
    "eval_logps/chosen": -731.7378540039062,
    "eval_logps/rejected": -1996.604736328125,
    "eval_loss": 3.1195009796647355e-05,
    "eval_rewards/accuracies": 1.0,
    "eval_rewards/chosen": -6.646634101867676,
    "eval_rewards/margins": 12.863931655883789,
    "eval_rewards/rejected": -19.51056480407715,
    "eval_runtime": 4.5785,
    "eval_samples": 5,
    "eval_samples_per_second": 1.092,
    "eval_steps_per_second": 0.218,
    "train_loss": 0.0,
    "train_runtime": 0.0617,
    "train_samples": 134157,
    "train_samples_per_second": 6527181.256,
    "train_steps_per_second": 815915.902
}