[Accuracy diff No.52] Fix accuracy diff for reciprocal API #73128
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
你的PR提交成功，感谢你对开源项目的贡献! |
lshpku
reviewed
Jun 10, 2025
Comment on lines
+3597
to
+3600
| MPType dout = static_cast<MPType>(arg_dout); | ||
| MPType out = static_cast<MPType>(arg_out); | ||
| return static_cast<T>(-dout * | ||
| static_cast<MPType>(static_cast<T>(out * out))); |
There was a problem hiding this comment.
lshpku
approved these changes
Jun 10, 2025
shanjiang7
pushed a commit
to shanjiang7/Paddle
that referenced
this pull request
Jun 12, 2025
…dle#73128) * fix reciprocal_grad accuracy * fix compile * refine
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PR Category
Execute Infrastructure
PR Types
Improvements
Description
日志显示反向精度问题，都出现在 torch 包含 inf 中，torch 的自动求导是
-grad * (result * result).conj()，先进行了 output 的相乘，如果出现 inf ，就会传递下去，paddle中按顺序 -grad*result，就导致不会出现 inf。对齐 torch 的行为，感觉在数学上也更加合理。（这里加了一个精度提升，主要是看见 torch 的mul算子有这个行为，但是没加 PaddleAPITest 也能通过