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 improving a model's foundational language capabilities using a large amount of newly collected unlabeled text?
①Manually label all of that data.
②Add evaluation benchmarks and monitor them.
③Handle it with instruction tuning.
④Update the model's representations through continual pretraining.
⑤Train with differential privacy.