試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 44 / 65
Which design should be given top priority for a real-time chat inference platform to handle sudden traffic spikes such as during Black Friday?
①Precompute and cache all possible responses
②Prioritize frequently fine-tuning the model to keep it up to date.
③Introduce containerization and autoscaling to dynamically scale inference nodes up and down.
④Substantially increase the maximum number of tokens per response.
⑤Adopt a policy of shortening responses at the expense of model accuracy.