[co63oc]

PR Category

Execute Infrastructure

PR Types

Improvements

Description

[0-size Tensor No.248] Add 0-size Tensor support for reshape

paddle/phi/kernels/reshape_grad_kernel.cc 中对所有backend注册

PaddleAPITest 测试没有cuda error, torch 和 paddle都有shape检查

如果x numel为0，shape numel不为0，torch没有错误，这是因为PaddleAPITest有修改torch规则配置，返回为全0
https://github.com/PFCCLab/PaddleAPITest/blob/480b501e1fd3c480e173284c8a94d15a5c42546e/tester/paddle_to_torch/rules.py#L4641

[paddle-bot]

你的PR提交成功，感谢你对开源项目的贡献!
请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

[co63oc]

/re-run all-failed

[co63oc]

/re-run all-failed

[co63oc]

@DanielSun11 CI已完成请review

[DanielSun11]

x numel为0，shape numel 不为0，torch会报错。

import torch
# import paddle
shape_a = [0,1,7,1]
shape_b = (52,7,1,)
a = torch.zeros(shape_a)
b = torch.reshape(a,shape_b)

Traceback (most recent call last):
  File "/home/code/test/test_reshape.py", line 6, in <module>
    b = torch.reshape(a,shape_b)
RuntimeError: shape '[52, 7, 1]' is invalid for input of size 0

请 @wanghuancoder 关注https://github.com/PFCCLab/PaddleAPITest/blob/480b501e1fd3c480e173284c8a94d15a5c42546e/tester/paddle_to_torch/rules.py#L4641 此处

[DanielSun11]

LGTM

[wanghuancoder]

LGTM