Pull Request已成功合入, 合并人@ascend-robot
（感谢 Ambi 的贡献）【openlibing.ci】识别到代码检查告警抑制注释，匹配工具：flake8，请Committer检视其合理性。
【openlibing.ci】识别到代码检查告警抑制注释，匹配工具：flake8，请Committer检视其合理性。
合理添加
compile
当前仓库存在以下 保护分支 ：
| Protected Branch | Version | Release |
|---|---|---|
| master | ||
| v2.12.0 | ||
| v2.11.0 | ||
| v2.10.0 | ||
| v2.9.0 | ||
| v2.7.1 | ||
| v2.12.0-26.1.0 | ||
| v2.10.0-26.1.0 | ||
| v2.11.0-26.1.0 | ||
| v2.9.0-26.1.0 | ||
| v2.7.1-26.1.0 | ||
| ci-test | ||
| sync-pr28113--to-v2.9.0 |
评论 /sync <branch1> <branch2> ... 可将当前 PR 修改同步到其它分支（创建同步 PR）：
a) 如果当前 PR 是 Open 状态，同步操作将延迟到 PR 被合并时执行
b) 如果当前 PR 已经 Merged，将立即执行同步操作
注意：
- /sync 命令可以指定同步到多个分支，仅最后一个 /sync 命令生效
- 如果创建的同步 PR 不正确，可通过向同步 PR 的源分支提交轻量级 PR 完善，或使用 /close 命令关闭
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 |
|---|---|---|
| test | ✅ liangsongwei, zichun_ye (2/2) | ✅ liangsongwei, zichun_ye (2/1) |
| torch_npu/_inductor | ✅ zichun_ye, liangsongwei (2/2) | ✅ liangsongwei, zichun_ye (2/1) |
💡 Tip:
- Committer can comment
/approveor/lgtm- Commenting
/approveimplies both code review (lgtm) and intent to merge (approve)
CLA Signature Pass
sunhaolun, thanks for your pull request. All authors of the commits have signed the CLA. 👍
compile
Ascend docs pipeline is running...
✅ 跳过 docs ci 检查，没有需要检查的文档文件
| 阶段 | 任务名 | 状态 | 详情 |
|---|---|---|---|
| 编译构建 | Build_X86 | ✅ | >>> |
| Build_ARM | ✅ | >>> | |
| Build_LibTorch_x86 | ✅ | >>> | |
| Build_LibTorch_ARM | ✅ | >>> | |
| Build_X86_torchair | 🛑 | >>> | |
| Build_ARM_torchair | 🛑 | >>> | |
| patch_test | 🛑 | >>> | |
| 恶意代码检查 | Antipoison | ✅ | >>> |
| 编码安全与规范检查 | CodeCheck | ✅ | >>> |
| check_error | ✅ | >>> | |
| CodeCheck_lintrunner | ✅ | >>> | |
| 开源片段检查 | SCA | ✅ | >>> |
| 开发者测试 | UT_X86_Part_01 | 🛑 | >>> |
| UT_X86_Part_02 | 🛑 | >>> | |
| UT_ARM_A3_Part_01 | 🛑 | >>> | |
| UT_ARM_A3_Part_02 | 🛑 | >>> | |
| UT_ARM_A2_Part_01 | ✅ | >>> | |
| UT_ARM_A2_Part_02 | ✅ | >>> | |
| UT_ARM_A2_Part_03 | ✅ | >>> | |
| UT_inductor_Part_01 | ✅ | >>> | |
| UT_inductor_Part_02 | ✅ | >>> | |
| UT_inductor_Part_03 | ✅ | >>> | |
| UT_inductor_Part_04 | ✅ | >>> | |
| UT_DIST_ARM_Part_01 | 🛑 | >>> | |
| UT_DIST_ARM_Part_02 | 🛑 | >>> | |
| UT_DIST_ARM_Part_03 | 🛑 | >>> | |
| UT_DIST_ARM_Part_04 | 🛑 | >>> | |
| UT_ARM_A2_Select_Part_01 | ✅ | >>> | |
| UT_ARM_A2_Select_Part_02 | ✅ | >>> | |
| 流水线 | PR-pipeline_pytorch | ✅ | >>> |
Linking Issue Notice
@sunhaolun , the pull request must be linked to at least one issue.
If an issue has already been linked, but the needs-issue label remains, you can remove the label by commenting /check-issue .
/approve
/approve
The MR can not be merged, because of CodeReview discussion not resolved
If you want to solve this problem, you can click here to do it in the FAQs.
Pull Request 已合并或已关闭。
If you want to solve this problem, you can click here to do it in the FAQs.
https://gitcode.com/Ascend/pytorch/issues/2433
【合入来源】
【修改方案】
删除了对 torch._dynamo.utils.run_node 的 monkey patch：去掉 _patch_run_node() 以及 torch._dynamo.utils.run_node = _patch_run_node。
原先在 Dynamo run_node 阶段特殊处理 npu.npu_fusion_attention，把 actual_seq_qlen / actual_seq_kvlen 转成 list；现在这个逻辑下沉到了 npu_stream.py 的算子封装里。
npu_fusion_attention_grad 同步支持 softmax_layout、sink，返回值从 4 个扩展为 5 个，新增 dsink 梯度。
将 prefix、actual_seq_qlen、actual_seq_kvlen 的 tensor-to-list 转换放到 wrapper 内部处理，替代原来在 run_node patch 里的处理。
新增 InductorNpuAttentionFunction 的 autograd 封装，forward 保存 sink，backward 返回 grad_sink，并通过 apply_inductor_npu_attention_patch() 替换 torch.ops.npu.npu_fusion_attention。
【资料变更】
不涉及
【接口变更】
不涉及
【功能验证】
Ci UT全量自检
【CheckList】