Qwen2.5-7B-MPO/all_results.json

{
    "epoch": 2.0,
    "total_flos": 0.0,
    "train_loss": 6.450036816596985,
    "train_runtime": 2055.1895,
    "train_samples_per_second": 0.584,
    "train_steps_per_second": 0.073
}