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.

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

Implement attention sliding window for Mistral #341

Merged
merged 3 commits into from
Feb 23, 2024

Conversation

jonatanklosko
Copy link
Member

Attention sliding window means that each token "attends to" N prior and N later tokens.

Comment on lines 35 to 37
# TODO test once we know the expected behaviour
# spec = Bumblebee.configure(spec, attention_window_size: 2)
spec = Bumblebee.configure(spec, attention_window_size: 1)
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There are two conflicting interpretations of the size. I am waiting for a resolution upstream (huggingface/transformers#29176).

@jonatanklosko jonatanklosko merged commit a6019fd into main Feb 23, 2024
2 checks passed
@jonatanklosko jonatanklosko deleted the jk-attention-sliding-window branch February 23, 2024 10:20
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant