Add `W` parameter as written in paper.

graykode / nlp-tutorial

Natural Language Processing Tutorial for Deep Learning Researchers

https://www.reddit.com/r/MachineLearning/comments/amfinl/project_nlptutoral_repository_who_is_studying/

MIT License

14.07k stars 3.91k forks source link

Add `W` parameter as written in paper. #4

Closed likejazz closed 5 years ago

likejazz commented 5 years ago

http://www.jmlr.org/papers/volume3/bengio03a/bengio03a.pdf

According to paper, W is used for directed connection, and is optionally zero. But W is written in the formula of the paper, and therefore needs to be implemented in code.

graykode commented 5 years ago

Thanks for detailed implementation @likejazz I will apply your Pr in *.ipynb and tensor file.