qwen72B_models/qwen_72B_d0_iter1_model/train_results.json

{
    "epoch": 2.0,
    "total_flos": 294651875557376.0,
    "train_loss": 0.6822677301323932,
    "train_runtime": 1850.6815,
    "train_samples_per_second": 1.982,
    "train_steps_per_second": 0.124
}