CMU Neural Nets for NLP 2019 (9): Debugging Neural Nets (for NLP)

preview_player
Показать описание
This lecture (by Graham Neubig) for CMU CS 11-747, Neural Networks for NLP (Spring 2019) covers:

* Identifying problems
* Debugging training time problems
* Debugging test time problems

Рекомендации по теме
Комментарии
Автор

Hi, Thanks a lot for such great lectures, these are turning out to be life-saving. However, I have a small query: In sequence to sequence model setting each direction of BiLSTM should have at least 512 nodes or combined 512 nodes(i.e., each direction 256 nodes)? Thank you!

kaushalkumarmaurya