試験

問題 6 / 65

Which of the following is the most appropriate technical limitation of in-context learning?

It is essential to tune the learning rate and optimization hyperparameters.
There is a limit on input prompt length, so many examples cannot be provided at the same time.
Not being able to permanently update a model's weights isn't always a disadvantage.
The fact that labeled data is completely unusable.
Only works on-premises.

当サイトでは、ユーザー体験の向上を目的としてCookieを使用しています。サイトの利用を継続することで、Cookieの使用に同意したものとみなされます。