Skip to content

feat: [Deepseek] Redesign multi-stream API #3459

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

Merged
merged 1 commit into from
Apr 12, 2025
Merged

Conversation

hlu1
Copy link
Collaborator

@hlu1 hlu1 commented Apr 10, 2025

This PR applied the new API to Deepseek only. Llama will be updated later.

Applied multi-stream to overlap bmm and concat in MLA.forward_generation.

@hlu1 hlu1 changed the title [Deepseek] Redesign multi-stream API feat: [Deepseek] Redesign multi-stream API Apr 10, 2025
@hlu1
Copy link
Collaborator Author

hlu1 commented Apr 11, 2025

/bot run --disable-fail-fast

@tensorrt-cicd
Copy link
Collaborator

PR_Github #1825 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #1825 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #1350 completed with status: 'FAILURE'

Copy link
Collaborator

@QiJune QiJune left a comment

Choose a reason for hiding this comment

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

LGTM

@hlu1 hlu1 requested a review from litaotju April 11, 2025 20:51
@hlu1
Copy link
Collaborator Author

hlu1 commented Apr 11, 2025

/bot run --disable-fail-fast

@hlu1 hlu1 enabled auto-merge (squash) April 11, 2025 21:03
@tensorrt-cicd
Copy link
Collaborator

PR_Github #1967 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #1967 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #1450 completed with status: 'SUCCESS'

@hlu1 hlu1 disabled auto-merge April 12, 2025 06:00
@hlu1 hlu1 merged commit 4855431 into NVIDIA:main Apr 12, 2025
2 checks passed
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.

5 participants