【cs224n-16】Low Resource Machine Translation
神经机器翻译（NMT）的成功往往依赖于大量高质量的双语语料作为训练数据。如果是蒙古语、尼泊尔语这些小语种，无法提供足够多的双语数据，更极端的现实情况是，有些语言几乎没有任何双语预料，这种情况下NMT就无能为力了。 松散定义:当并行句子数量在10,000或更少时，可以认为语言对资源不足。注:现代NMT系统现在有数亿个参数! 挑战: 数据:
相关文章
- 1. statistical machine translation system
- 2. Statistical Machine Translation Tutorial Reading
- 3. dual learning for machine translation
- 4. lecture8 - Machine Translation Seq2Seq and Attention
- 5. CS224n-Lecture8-Machine Translation, Seq2Seq and Attention
- 6. 【CS224n】Neural Machine Translation with Seq2Seq
- 7. Linguistic Input Features Improve Neural Machine Translation
- 8. An Effective Approach to Unsupervised Machine Translation
- 9. 【cs224n-7】Machine Translation, Seq2Seq and Attention
- 10. 论文解读：Multimodal Machine Translation with Embedding Prediction
- 更多相关文章...
- • Docker Machine - Docker教程
- • PHP range() 函数 - PHP参考手册
- • Java Agent入门实战（一）-Instrumentation介绍与使用
- • ☆基于Java Instrument的Agent实现
相关标签/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。
欢迎关注本站公众号,获取更多信息
相关文章
- 1. statistical machine translation system
- 2. Statistical Machine Translation Tutorial Reading
- 3. dual learning for machine translation
- 4. lecture8 - Machine Translation Seq2Seq and Attention
- 5. CS224n-Lecture8-Machine Translation, Seq2Seq and Attention
- 6. 【CS224n】Neural Machine Translation with Seq2Seq
- 7. Linguistic Input Features Improve Neural Machine Translation
- 8. An Effective Approach to Unsupervised Machine Translation
- 9. 【cs224n-7】Machine Translation, Seq2Seq and Attention
- 10. 论文解读：Multimodal Machine Translation with Embedding Prediction