SUN Dong comments

Repositories
Issues
Comments

Results 33 comments of


                                            SUN Dong

[0-size Tensor No.132、236、240、241、259] Add 0-size Tensor support for masked_select

CPU上测试都过了吗？

[0-size Tensor No.132、236、240、241、259] Add 0-size Tensor support for masked_select

https://github.com/PFCCLab/PaddleAPITest/blame/main/report/0size_tensor_gpu/error_config.txt 中有paddle.masked_select的0-size 配置https://github.com/PFCCLab/PaddleAPITest/blob/main/report/0size_tensor_gpu/error_config_paddleonly.txt#L6571有 paddle.polar的0-size 配置，请验证一下

[0-size Tensor No.132、236、240、241、259] Add 0-size Tensor support for masked_select

> 下面这个算 0 size 问题吗,cpu 和 GPU 的梯度不一样了，torch 是[0,0].paddle 的cpu是对的 > > ```python > import paddle > import numpy as np > # paddle.device.set_device('cpu') > x = paddle.to_tensor([1,2], dtype='float32') >...