Skip to content

【New Feature】W4afp8 supports per group quantization#4987

Merged
EmmonsCurse merged 7 commits intoPaddlePaddle:developfrom
yangjianfengo1:new_w4afp8
Nov 13, 2025
Merged

【New Feature】W4afp8 supports per group quantization#4987
EmmonsCurse merged 7 commits intoPaddlePaddle:developfrom
yangjianfengo1:new_w4afp8

Conversation

@yangjianfengo1
Copy link
Contributor

@yangjianfengo1 yangjianfengo1 commented Nov 12, 2025

Motivation

此pr #4272 被revert掉,现重新提交
TP并行下w4afp8可直接通过本PR适配动态量化,EP并行下的动态量化需要结合PaddlePaddle/Paddle#76262 该PR中对deep ep的修改使用

Modifications

Usage or Command

Accuracy Tests

Checklist

@paddle-bot
Copy link

paddle-bot bot commented Nov 12, 2025

Thanks for your contribution!

@EmmonsCurse EmmonsCurse merged commit ae7bee8 into PaddlePaddle:develop Nov 13, 2025
13 of 16 checks passed
EmmonsCurse added a commit that referenced this pull request Nov 16, 2025
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.

5 participants