remove import torchvision when testing torch.hub#21132
remove import torchvision when testing torch.hub#21132ailzhang wants to merge 3 commits intopytorch:masterfrom
Conversation
|
@pytorchbot rebase this please |
facebook-github-bot
left a comment
There was a problem hiding this comment.
@ailzhang has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
25996de to
74babd3
Compare
facebook-github-bot
left a comment
There was a problem hiding this comment.
@ailzhang has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
|
Lint test failure unrelated. Landing since it only failed windows test in my previous commit and now it passed |
|
Any reason why this reverts #19606? The hub test now again prints like 40% lines of CI log. |
|
@ssnl I removed it since the line it tried to suppress was removed. Maybe it's a different place not flushes huge logs? (I know there're some logs in python27 case but that's expected). Would you mind sharing an example? |
Summary: #21132 reverted #19606. Now these tests again print like 40% lines of CI outputs (e.g., https://circleci.com/gh/pytorch/pytorch/2041825?utm_campaign=vcs-integration-link&utm_medium=referral&utm_source=github-build-link) This PR now uses the functionality introduced in pytorch/vision#862. Pull Request resolved: #22043 Differential Revision: D15947268 Pulled By: ailzhang fbshipit-source-id: f84f4d6b86203dbe8687e04ae3ed8c99df0bdff8
This should pass once pytorch/vision#971 is merged.
To remove torchvision as baseline, we just compare to sum of all param.sum() in pretrained resnet18 model, which means we need to manually update the number only when that pretrained weights are changed, which is generally rare.