Skip to content
This repository has been archived by the owner on Mar 15, 2024. It is now read-only.

how to implement cosub training use deit-III #217

Open
xiaoguang-1 opened this issue Apr 28, 2023 · 2 comments
Open

how to implement cosub training use deit-III #217

xiaoguang-1 opened this issue Apr 28, 2023 · 2 comments

Comments

@xiaoguang-1
Copy link

Hi,
i want to use cosub trainging method for custom images classification work, but i don't find implemention in the code. Does this work contains the details of cosub training method. If yes, my computer can't support RASamper ,does it have an influence?
Best,
xiaoguang

@xiaoguang-1
Copy link
Author

Trouble you again,
I would like to know how the residual function at efficient stochastic depth is implemented? Thank you so much!
yours best,
xiaoguang

@TouvronHugo
Copy link
Contributor

Hi @xiaoguang-1 ,
Thanks for your message.
I have added cosub in the code you need to use stochastic depth with cosub but RASampler is not mandatory.
I still need to add the efficient version of Stochastic depth but the code works without it.
Best,
Hugo

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants