KEMBAR78
Makes a streaming backward test try gradient stealing more directly by mcarilli · Pull Request #60065 · pytorch/pytorch · GitHub
Skip to content

Conversation

@mcarilli
Copy link
Collaborator

@mcarilli mcarilli commented Jun 16, 2021

Closes #59846.

#59846 is likely paranoia, and some of the test_streaming_backward_* in test_cuda.py already use gradient stealing (ie, they start with .grads as None before backward). Regardless, this PR augments one of the tests to stress gradient stealing a bit more directly.

@facebook-github-bot
Copy link
Contributor

facebook-github-bot commented Jun 16, 2021

💊 CI failures summary and remediations

As of commit e2769ad (more details on the Dr. CI page and at hud.pytorch.org/pr/60065):


  • 1/1 failures introduced in this PR

1 failure not recognized by patterns:

Job Step Action
GitHub Actions Lint / shellcheck Assert that regenerating the workflows didn't change them 🔁 rerun

Preview docs built from this PR

This comment was automatically generated by Dr. CI (expand for details).Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions to the (internal) Dr. CI Users group.

Click here to manually regenerate this comment.

@mcarilli mcarilli requested review from mruberry and ngimel June 16, 2021 02:59
@mcarilli mcarilli added module: cuda Related to torch.cuda, and CUDA support in general module: autograd Related to torch.autograd, and the autograd engine in general labels Jun 16, 2021
@mcarilli mcarilli changed the title Small streaming backwards test augment Makes a streaming backward test try gradient stealing more directly Jun 16, 2021
@mruberry
Copy link
Collaborator

This seems fine but I'll let @ngimel review.

@anjali411 anjali411 added the triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module label Jun 16, 2021
@facebook-github-bot
Copy link
Contributor

@ngimel has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

@facebook-github-bot
Copy link
Contributor

@ngimel merged this pull request in 9fb6b40.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

cla signed Merged module: autograd Related to torch.autograd, and the autograd engine in general module: cuda Related to torch.cuda, and CUDA support in general open source triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Need a test to make sure the autograd engine inserts proper leaf stream syncs for stolen gradients

6 participants