試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 52 / 65
Which method is most suitable for deploying a large-scale model to edge devices when you want to reduce model size while retaining as much performance as possible?
①Strong quantization (aggressive quantization)
②Parameter reduction through aggressive pruning
③Reacquiring knowledge through continual pretraining
④Transfer performance to a smaller model using knowledge distillation.
⑤Retrain the full model.