llama-factory/saves/Qwen2-72B-Instruct/eval_results.json

{
    "epoch": 5.997322623828648,
    "eval_loss": 2.630276679992676,
    "eval_runtime": 17.6344,
    "eval_samples_per_second": 2.609,
    "eval_steps_per_second": 2.609
}