Hongmei Zhu issues

Repositories
Issues
Comments

Results 2 issues of


                                            Hongmei Zhu

CSPN_ours的问题

我在调用x = conv2d(x, kernel, kernel_size=K, stride=1, padding=K // 2, dilation=1, native_impl=True)时，native_impl 取True时能backward，取False时不能backward。两者运算有本质的什么不同吗？

Models

Please release the final checkpoint for each model.