-
Notifications
You must be signed in to change notification settings - Fork 39
【Hackathon 9th No.16、19】Fix accuracy for index_put #533
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
【Hackathon 9th No.16、19】Fix accuracy for index_put #533
Conversation
|
Thanks for your contribution! |
b6e0b8a to
6aa1545
Compare
cangtianhuang
left a comment
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.
|
同学可以修改一下题目,暂时把 copysign 去掉😉 |
已修改~ |
|
emmm,我记得这是个遗留问题了,暂时没有解决,参考:#480 可以先把配置放入 tester/api_config/torch_error_skip.txt 中,视作两者行为有差异,不当成错误报出🫡 |
好的,提交了 #536 进行修改 |

重新整理了一下 _handle_indices_arg 的逻辑
由于涉及广播机制,生成不重复的索引比较复杂,当前尽量保证张量内不重复,所以还是选择迁移失败的测试到 random_calculation.txt
在 torch 2.8.0 版本中 paddle.index_put(Tensor([110, 42, 56, 56],"float64"), tuple(Tensor([16, 16],"int64"),Tensor([16, 16],"int64"),Tensor([1, 16],"int64"),), Tensor([56],"float64"), True, ) 是一个 torch 的 bug,所以先放进 torch_error
No.9 的 copysign 一直没有复现出来,是指要修复 output dtype 类型不同吗?但是别的 output dtype 类型不同的算子也不在列表里
No.19 的 put_along_axis 可能被 #365 修复了,但是我当时的分析 #218 是 Paddle assign 的时候应该不会随机,所以我还是选择提了一个 paddle pr PaddlePaddle/Paddle#74854 ,经过验证,遗留的测试通过了