50/mistral-dpo-lr-5.0e-7-beta-0.01/train_results.json

{
    "epoch": 1.0,
    "total_flos": 0.0,
    "train_loss": 0.6616816390939311,
    "train_runtime": 3221.9386,
    "train_samples": 9383,
    "train_samples_per_second": 2.912,
    "train_steps_per_second": 0.046
}