Search Results for author: Taku Kudo

Found 5 papers, 3 papers with code

SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processing

3 code implementations EMNLP 2018 Taku Kudo, John Richardson

We perform a validation experiment of NMT on English-Japanese machine translation, and find that it is possible to achieve comparable accuracy to direct subword training from raw sentences.

Machine Translation NMT +2

Phrase-based Machine Translation using Multiple Preordering Candidates

no code implementations COLING 2016 Yusuke Oda, Taku Kudo, Tetsuji Nakagawa, Taro Watanabe

In this paper, we propose a new decoding method for phrase-based statistical machine translation which directly uses multiple preordering candidates as a graph structure.

Machine Translation Translation

Cannot find the paper you are looking for? You can Submit a new open access paper.