(François) Hoa T. Le

icon indicating a website link https://lethienhoablog.wordpress.com/

icon location Grenoble, France icon location Data Scientist, Ph.D. in Deep Learning & NLP

Results 2 issues of (François) Hoa T. Le

NLL & Perplexity Loss

2 comment

Hi, It seems that Perplexity is normalized twice & norm_term of NLLLoss should be masked out as well.

bug
medium priority

Masked attention

4 comment

Hi, I see that this implementation is lacking masked attention on encoder. Input_lengths should be passed to decoder (not just encoder) in order to compute this. OpenNMT already provided this...

medium priority