About the word_emb for cross attention

ZcyMonkey / AttT2M

Code of ICCV 2023 paper: "AttT2M: Text-Driven Human Motion Generation with Multi-Perspective Attention Mechanism"

https://arxiv.org/abs/2309.00796

Apache License 2.0

37 stars 3 forks source link

Open buptxyb666 opened 2 months ago

buptxyb666 commented 2 months ago

Thanks for your great work!

I wonder that the length of text usually less than 77, so why not mask the padding tokens in word_emb when performing cross attention?