-
Notifications
You must be signed in to change notification settings - Fork 5.7k
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
[NewIR]Fix new ir concat split bug #55419
Conversation
… fix_new_ir_concat_op
你的PR提交成功,感谢你对开源项目的贡献! |
❌ The PR is not created using PR's template. You can refer to this Demo. |
… fix_new_ir_concat_op
… fix_new_ir_concat_op
… fix_new_ir_concat_op
@@ -1087,6 +1088,26 @@ struct FetchOpTranscriber : public OpTranscriber { | |||
} | |||
}; | |||
|
|||
// NOTE, add_n op in legacy ops don't have a kernel, so we use a new op for now |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
我看原本不管外面是不是inplace都会调用同一个kernel实现,这样的话是不是不管 Op的输入输出有没有同一个变量,都把它翻译为非inplace op,然后去掉现在pd_op.yaml
里的add_n_
下划线版本定义,只对应到一个add_n
比较好。
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
像这种op有必要区分它们的inplace语义吗?
"Only support DenseTensorType in vector")); | ||
} | ||
} | ||
} else if (type.isa<paddle::dialect::AllocatedDenseTensorType>()) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
} else if (type.isa<paddle::dialect::AllocatedDenseTensorType>()) { | |
} else if (type.isa<paddle::dialect::AllocatedSelectedRowsType>()) { |
这里是写错了?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
done
… fix_new_ir_concat_op
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM for YAML
* fix new ir concat op bug * fix bug * using add_n_with_kernel instead of add_n impl * fix pd_op yaml bug * fix bug
* fix new ir concat op bug * fix bug * using add_n_with_kernel instead of add_n impl * fix pd_op yaml bug * fix bug
PR types
Bug fixes
PR changes
Others
Description
修复新IR下, concat split op的单测错误
Others
Pcard-67164