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

{
    "epoch": 1.9900497512437811,
    "eval_loss": 0.20492638647556305,
    "eval_runtime": 1373.2772,
    "eval_samples_per_second": 1.82,
    "eval_steps_per_second": 1.82
}