Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Local Attention for GPT2 #10834

Closed
leogao2 opened this issue Mar 21, 2021 · 1 comment
Closed

Local Attention for GPT2 #10834

leogao2 opened this issue Mar 21, 2021 · 1 comment

Comments

@leogao2
Copy link
Contributor

leogao2 commented Mar 21, 2021 •
edited
Loading

🚀 Feature request

Our model uses local attention in some layers (i.e each position can only see the last k=256 tokens in every other layer). We would like to be able to specify this in the config on the model hub.

Motivation

Right now we can't integrate the 1.3B and 2.7B EleutherAI GPT models because local attention is not supported in transformers.
@LysandreJik
Copy link
Member

@patil-suraj is working on implementing GPT Neo over at #10848! The 1.3B and 2.7B should be loadable in that architecture once finalized.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
3 participants
@patil-suraj @LysandreJik @leogao2