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

fix decoding kernel for deepseekv2 #2688

Merged
merged 1 commit into from
Nov 6, 2024

Conversation

grimoire
Copy link
Collaborator

Fix decoding kernel for GQA with large kv_group.

@lvhan028 lvhan028 requested a review from RunningLeon October 31, 2024 10:16
Copy link
Collaborator

@RunningLeon RunningLeon left a comment

Choose a reason for hiding this comment

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

LGTM

@lvhan028
Copy link
Collaborator

lvhan028 commented Nov 1, 2024

@RunningLeon could you help start a oc evaluation with gsm8k and math datasets?

@lvhan028 lvhan028 added the Bug:P1 label Nov 1, 2024
@RunningLeon
Copy link
Collaborator

RunningLeon commented Nov 4, 2024

@RunningLeon could you help start a oc evaluation with gsm8k and math datasets?

Looks normal.

dataset version metric mode DeepSeek-V2-Chat-pytorch
gsm8k a58960 accuracy gen 91.36
race-high 9387ad accuracy ppl 90.94

@lvhan028 lvhan028 merged commit 354028b into InternLM:main Nov 6, 2024
4 of 5 checks passed
AllentDan pushed a commit to AllentDan/lmdeploy that referenced this pull request Nov 13, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants