Skip to content

[OPs] MoE support wfp8afp8(channelwise) and improve per_token_quant_fp8 #4238

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
Merged
ckl117 merged 5 commits into from
Sep 24, 2025
Merged

[OPs] MoE support wfp8afp8(channelwise) and improve per_token_quant_fp8 #4238

ckl117 merged 5 commits into from
Sep 24, 2025

Conversation

@ckl117
Copy link
Collaborator

@ckl117 ckl117 commented Sep 23, 2025 •
edited
Loading

CP from 2.2 PR#4115

  • Optimize per_token_quant_fp8 kernel performance improved by 50%
  • Support Wfp8Afp8MoEMethod(weight quant in channel-wise)
python -m fastdeploy.entrypoints.openai.api_server \
    --model ${model_path} \
    --max-model-len 32768 \
    --max-num-seqs 128 \
    --tensor-parallel-size 1 \
    --load_choices "default_v1" \
    --quantization wfp8afp8 \

@paddle-bot
Copy link

paddle-bot bot commented Sep 23, 2025

Thanks for your contribution!

@ckl117 ckl117 changed the title Dev moe wfp8afp8 [OPs] MoE support wfp8afp8(channelwise) and improve per_token_quant_fp8 Sep 24, 2025
@ckl117 ckl117 merged commit 7c1fd19 into PaddlePaddle:develop Sep 24, 2025
26 of 28 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@zhoutianzi666 zhoutianzi666 zhoutianzi666 approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@ckl117 @zhoutianzi666