{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/on-using-very-large-target-vocabulary-for","title":"On Using Very Large Target Vocabulary for Neural Machine Translation","arxiv_id":"1412.2007","date":"2014-12-05","proceeding":"IJCNLP 2015 7","authors":["Sébastien Jean","Kyunghyun Cho","Roland Memisevic","Yoshua Bengio"],"abstract":"Neural machine translation, a recently proposed approach to machine\ntranslation based purely on neural networks, has shown promising results\ncompared to the existing approaches such as phrase-based statistical machine\ntranslation. Despite its recent success, neural machine translation has its\nlimitation in handling a larger vocabulary, as training complexity as well as\ndecoding complexity increase proportionally to the number of target words. In\nthis paper, we propose a method that allows us to use a very large target\nvocabulary without increasing training complexity, based on importance\nsampling. We show that decoding can be efficiently done even with the model\nhaving a very large target vocabulary by selecting only a small subset of the\nwhole target vocabulary. The models trained by the proposed approach are\nempirically found to outperform the baseline models with a small vocabulary as\nwell as the LSTM-based neural machine translation models. Furthermore, when we\nuse the ensemble of a few models with very large target vocabularies, we\nachieve the state-of-the-art translation performance (measured by BLEU) on the\nEnglish->German translation and almost as high performance as state-of-the-art\nEnglish->French translation system.","url_abs":"http://arxiv.org/abs/1412.2007v2","url_pdf":"http://arxiv.org/pdf/1412.2007v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"on-using-very-large-target-vocabulary-for","repo_url":"https://github.com/HIT-SCIR/ELMoForManyLangs","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"machine-translation","task_name":"Machine Translation"},{"task_slug":"translation","task_name":"Translation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1412.2007","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}