KEMBAR78
Fix torch::nn::init::orthogonal_ with CNNs by Omegastick · Pull Request #18915 · pytorch/pytorch · GitHub
Skip to content

Conversation

@Omegastick
Copy link

Fixes #18518

I changed the C++ API torch::nn::init::orthogonal_ implementation to match the Python implementation.

@Omegastick
Copy link
Author

I haven't touched any Python files, so I assume the Python lint checks were already failing before this commit.

@Omegastick Omegastick changed the base branch from master to AA April 5, 2019 06:34
@Omegastick Omegastick changed the base branch from AA to master April 5, 2019 06:34
@ezyang
Copy link
Contributor

ezyang commented Apr 5, 2019

@Omegastick Thank you. Can we get a little test for this case?

@pytorchbot rebase this please

@Omegastick
Copy link
Author

@ezyang Right, how's that? Let me know if I did something wrong with the test, I'm not too familiar with PyTorch's tests.

Copy link
Contributor

@facebook-github-bot facebook-github-bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@ezyang is landing this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

@facebook-github-bot
Copy link
Contributor

@ezyang merged this pull request in 31ff0ec.

zhangguanheng66 pushed a commit to zhangguanheng66/pytorch that referenced this pull request May 6, 2019
Summary:
Fixes pytorch#18518

I changed the C++ API torch::nn::init::orthogonal_ implementation to match the Python implementation.
Pull Request resolved: pytorch#18915

Differential Revision: D14851833

Pulled By: ezyang

fbshipit-source-id: 45b5e9741582777c203e9ebed564ab3ac1f94baf
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

"shape '[32, 3]' is invalid for input of size 6144" when initializing Conv2d layer in C++ frontend

4 participants