rayleizhu / BiFormer

[CVPR 2023] Official code release of our paper "BiFormer: Vision Transformer with Bi-Level Routing Attention"
https://arxiv.org/abs/2303.08810
MIT License
452 stars 36 forks source link

What are the functions of the parameters? #37

Open RayTan183 opened 8 months ago

RayTan183 commented 8 months ago

I have noticed some parameters in the code, such as kv_downsample diff_routing and so on. What are the functions of the parameters? Howwill they influence the model?

rayleizhu commented 8 months ago

No effect. They are legacy arguments I used during the exploration stage. See the illustration here

https://github.com/rayleizhu/BiFormer#news

image

RayTan183 commented 8 months ago

Thanks for your reply! I also noticed that there is a Token Labeling in the paper. Which part is the corresponding code?

rayleizhu commented 8 months ago

It has yet to be released.

I'm chasing the coming CVPR deadline. There will be some updates for this repository after the deadline.