論文閱讀：Neural Machine Translation with Byte-Level Subwords
文章簡介： 基於字符，子詞，詞的機器翻譯幾乎都是以詞頻top-k數量建立的詞典；但是針對字符相對雜亂的日文和字符較豐富的中文，往往他們的罕見詞難以表示； 本文提出採用字節級別的字詞BBPE(byte-level BPE)，不會出現oov的詞；比純用字節表示更方便，比只用字符表示又效果更好；當BBPE和BPE性能接近時，詞典size只是BPE的1/8； 主要方法： 整體思想是，把文本表示生字節級別的
相關文章
- 1. 論文閱讀——Unsupervised Neural Machine Translation with Indirect Supervision
- 2. 論文閱讀——Robust Unsupervised Neural Machine Translation with Adversarial Training
- 3. 論文閱讀：Fast Lexically Constrained Decoding with Dynamic Beam Allocation for Neural Machine Translation
- 4. 論文閱讀：ByteNet, Neural Machine Translation in Linear Time
- 5. Neural machine translation by jointly learning to align and translate論文閱讀筆記
- 6. 【論文閱讀】Neural Machine Translation By Jointly Learning To Align and Translate
- 7. [論文閱讀筆記] Massive Exploration of Neural Machine Translation Architectures
- 8. 【論文閱讀】Prior Knowledge Integration for Neural Machine Translation using Posterior Regularization
- 9. 論文解讀：Multimodal Machine Translation with Embedding Prediction
- 10. 論文閱讀——Toward Making the Most of Context in Neural Machine Translation
- 更多相關文章...
- • RSS 閱讀器 - RSS 教程
- • Docker Machine - Docker教程
- • JDK13 GA發佈：5大特性解讀
- • Scala 中文亂碼解決
相關標籤/搜索
每日一句
-
每一个你不满意的现在，都有一个你没有努力的曾经。
歡迎關注本站公眾號,獲取更多信息
相關文章
- 1. 論文閱讀——Unsupervised Neural Machine Translation with Indirect Supervision
- 2. 論文閱讀——Robust Unsupervised Neural Machine Translation with Adversarial Training
- 3. 論文閱讀：Fast Lexically Constrained Decoding with Dynamic Beam Allocation for Neural Machine Translation
- 4. 論文閱讀：ByteNet, Neural Machine Translation in Linear Time
- 5. Neural machine translation by jointly learning to align and translate論文閱讀筆記
- 6. 【論文閱讀】Neural Machine Translation By Jointly Learning To Align and Translate
- 7. [論文閱讀筆記] Massive Exploration of Neural Machine Translation Architectures
- 8. 【論文閱讀】Prior Knowledge Integration for Neural Machine Translation using Posterior Regularization
- 9. 論文解讀：Multimodal Machine Translation with Embedding Prediction
- 10. 論文閱讀——Toward Making the Most of Context in Neural Machine Translation