試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 35 / 65
When training a chatbot base model using a multilingual corpus, which method is most effective for sharing vocabulary across languages while suppressing unknown words and vocabulary bloat?
①Subword tokenization using BPE and SentencePiece
②Use completely independent vocabularies for each language.
③Use only character-level tokenization.
④Use only a fixed dictionary at the word level.
⑤Manage the vocabulary using one-hot encoding.