all_results.json

{
    "epoch": 1.998667732480682,
    "eval_logits/chosen": -13.45787239074707,
    "eval_logits/rejected": -12.986124992370605,
    "eval_logps/chosen": -209.70858764648438,
    "eval_logps/rejected": -266.514404296875,
    "eval_loss": 0.535685658454895,
    "eval_rewards/accuracies": 0.7565789818763733,
    "eval_rewards/chosen": -4.100185394287109,
    "eval_rewards/margins": 1.7237099409103394,
    "eval_rewards/rejected": -5.823895454406738,
    "eval_runtime": 15.7482,
    "eval_samples_per_second": 19.304,
    "eval_steps_per_second": 2.413,
    "total_flos": 174218888085504.0,
    "train_loss": 0.33488652056087054,
    "train_runtime": 7850.5465,
    "train_samples_per_second": 7.649,
    "train_steps_per_second": 0.239
}