試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 11 / 65
What is the most appropriate metric for evaluating a multi-step automation agent?
①The small number of lines in the agent's source code.
②Only the amount of training data for each model.
③Only the local machine's CPU usage.
④The large number of parameters in the model used by the agent.
⑤A comprehensive evaluation including end-to-end task success rate, safety, and explainability.