已合并
qbmm int4 非对称量化 #2281
bynshard创建于 3月3日
qbmm int4 非对称量化 #2281
已合并
Pull Request已成功合入, 合并人@CANN-robot
(感谢 bynshard 的贡献)CANN-robot
3月3日 评论:
3月3日 评论:
Thanks for your pull-request.
The full list of commands accepted by me can be found at here。
You can get sig-info at here
PR Approval Progress
✅ Congratulations! All modules have met the lgtm and approve requirements.
Module Approval Details
| module | lgtm status | approve status |
|---|---|---|
| */*/docs/acl*.md | ✅ chenqi317, liubo75, gitcode-chenjiao (3/2) | ✅ gitcode-chenjiao (1/1) |
| */*/op_api/*.h | ✅ tang-lei01, chenqi317, liubo75 (3/2) | ✅ tang-lei01 (1/1) |
| */*/op_host/*_def.cpp | ✅ tang-lei01, liubo75, chenqi317 (3/2) | ✅ tang-lei01 (1/1) |
| */*/op_host/op_api/*.h | ✅ chenqi317, tang-lei01, liubo75 (3/2) | ✅ tang-lei01 (1/1) |
| matmul | ✅ liubo75, chenqi317 (2/2) | ✅ chenqi317, liubo75 (2/1) |
💡 Tip:
- Committer can comment
/approveor/lgtm- Commenting
/approveimplies both code review (lgtm) and intent to merge (approve)
CLA Signature Pass
bynshard, thanks for your pull request. All authors of the commits have signed the CLA. 👍


3月3日 添加了label:cann-cla/yes
3月3日 将sxb154714,chaotang233,crystalhu,fanqirui,zhou-qilong,chenqi317,liubo75,tangweiwei2,wangzitao_leo,zhang-wu设为评审人
3月3日 将sxb154714,chaotang233,crystalhu,fanqirui,zhou-qilong,chenqi317,liubo75,tangweiwei2,wangzitao_leo,zhang-wu设为评审人
3月3日 将sxb154714,chaotang233,crystalhu,fanqirui,chenqi317,liubo75,tangweiwei2,wangzitao_leo,zhang-wu设为审查人
3月3日 将sxb154714,chaotang233,crystalhu,fanqirui,chenqi317,liubo75,tangweiwei2,wangzitao_leo,zhang-wu设为审查人
此处折叠了474条消息 查看更多
3月28日 添加了label:lgtm
CANN-robot
3月28日 评论:
3月28日 评论:
Review Guide
This pull-request passes review.
Committers who wrote a comment of /approve are: chenqi317, liubo75, tang-lei01, gitcode-chenjiao.
Reviewers who wrote a comment of /lgtm are: tang-lei01, gitcode-chenjiao, chenqi317, liubo75.


3月28日 关闭了关联的issue
3月28日 合入了pull request
描述
补齐A5上QBMM支持int4输入的K-G 非对称量化,A3上有该功能。
A5 上芯片不支持的 int4,所以方案为:
关联的Issue
https://gitcode.com/cann/ops-nn/issues/1745
测试
进行了回归测试
对该功能在的A5 进行了测试,通过门槛用例
文档更新
aclnnQuantMatmulV5.md
类型标签