all_results.json

{
    "epoch": 0.9997172745264349,
    "eval_logits/chosen": -0.9504738450050354,
    "eval_logits/rejected": -0.9573480486869812,
    "eval_logps/chosen": -4.477466106414795,
    "eval_logps/rejected": -5.212347030639648,
    "eval_loss": 2.455941915512085,
    "eval_rewards/accuracies": 0.8001337051391602,
    "eval_rewards/chosen": -44.774662017822266,
    "eval_rewards/margins": 7.348804473876953,
    "eval_rewards/rejected": -52.12346649169922,
    "eval_runtime": 100.659,
    "eval_samples_per_second": 29.595,
    "eval_steps_per_second": 1.858,
    "total_flos": 134366991482880.0,
    "train_loss": 3.275575773208929,
    "train_runtime": 7589.4389,
    "train_samples_per_second": 7.456,
    "train_steps_per_second": 0.058
}