qwen72B_models/qwen_72B_d5_iter6_model/train_results.json

{
    "epoch": 2.0,
    "total_flos": 268558516879360.0,
    "train_loss": 0.34019078340756115,
    "train_runtime": 2667.3925,
    "train_samples_per_second": 2.024,
    "train_steps_per_second": 0.127
}