all_results.json

{
    "epoch": 2.0,
    "eval_logits/chosen": 0.7808946371078491,
    "eval_logits/rejected": 0.8833128213882446,
    "eval_logps/chosen": -261.0694274902344,
    "eval_logps/rejected": -233.27114868164062,
    "eval_loss": 0.0021080097649246454,
    "eval_rewards/accuracies": 0.6919999718666077,
    "eval_rewards/chosen": -0.008252721279859543,
    "eval_rewards/margins": 0.01009758934378624,
    "eval_rewards/rejected": -0.018350308761000633,
    "eval_runtime": 325.1898,
    "eval_samples": 2000,
    "eval_samples_per_second": 6.15,
    "eval_steps_per_second": 0.384,
    "train_loss": 0.0021909422920118682,
    "train_runtime": 18127.9992,
    "train_samples": 30567,
    "train_samples_per_second": 3.372,
    "train_steps_per_second": 0.053
}