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

[SW-207965] Add batch splitting in attention layer to hide NIC latency(#14) #1640

Open
wants to merge 3 commits into
base: main
Choose a base branch
from

Conversation

kalyank007
Copy link

- Introduced the `--attn_batch_split` parameter to enable batch splitting in the attention and mlp layer.
- This approach aims to overlap communication and computation, effectively hiding NIC latency during distributed attention operations.

- Perform the add in the beginning of the next layer for better pipelining

- Updated Readme

- [SW-212702] Fix the  attn_batch_split argument specific to llama config (#74)

huggingface#14)

    - Introduced the `--attn_batch_split` parameter to enable batch
    splitting in the attention and mlp layer.
    - This approach aims to overlap communication and computation,
    effectively hiding NIC latency during distributed attention operations.

    - Perform the add in the beginning of the next layer for better pipelining

    - Updated Readme

    - [SW-212702] Fix the  attn_batch_split argument specific to llama config (huggingface#74)

Co-authored-by: Kalyan <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants