brotherb

icon company Baidu Inc. icon location Beijing icon location NLP

Results 2 issues of brotherb

why merge partition with one element to other partition?

Hi, I noitced that after k-means clustering, cluster with only one element is merged to the most similar cluster. My guess to this decision is that if this cluster has...

Does attention mask reduce computation cost？

2 comment

Hey, there. After I read the code, I am confused that the computation cost can be reduced by mask more tokens. Did I miss anything? PS. I see the FLOPS...