Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Drop torch 1.6 support #10367

Merged
merged 16 commits into from Nov 8, 2021
Merged

Drop torch 1.6 support #10367

merged 16 commits into from Nov 8, 2021

Conversation

carmocca
Copy link
Member

@carmocca carmocca commented Nov 5, 2021

What does this PR do?

We have a 4 PyTorch version policy. With the release of 1.10, that means we can drop 1.6

[1.6, 1.7, 1.8, 1.9, 1.10] -> [1.7, 1.8, 1.9, 1.10]

Part of #10075

Does your PR introduce any breaking changes? If yes, please list them.

None

Before submitting

  • Was this discussed/approved via a GitHub issue? (not for typos and docs)
  • Did you read the contributor guideline, Pull Request section?
  • Did you make sure your PR does only one thing, instead of bundling different changes together?
  • Did you make sure to update the documentation with your changes? (if necessary)
  • [n/a] Did you write any new necessary tests? (not for typos and docs)
  • Did you verify new and existing tests pass locally with your changes?
  • Did you list all the breaking changes introduced by this pull request?
  • Did you update the CHANGELOG? (not for typos, docs, test updates, or internal minor changes/refactorings)

PR review

  • Is this pull request ready for review? (if not, please submit in draft mode)
  • Check that all items from Before submitting are resolved
  • Make sure the title is self-explanatory and the description concisely explains the PR
  • Add labels and milestones (and optionally projects) to the PR so it can be classified

@carmocca carmocca added the ci Continuous Integration label Nov 5, 2021
@carmocca carmocca added this to the 1.5.x milestone Nov 5, 2021
@carmocca carmocca self-assigned this Nov 5, 2021
@github-actions
Copy link
Contributor

github-actions bot commented Nov 5, 2021

Build Error! No Linked Issue found. Please link an issue or mention it in the body using #<issue_id>

Copy link
Member

@justusschock justusschock left a comment

Choose a reason for hiding this comment

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

LGTM overall.

I think whereever possible (all runs/images with python 3.8) we should still fall back to 1.7 and only for the 3.9 versions lift it to 1.7.1

environment.yml Outdated Show resolved Hide resolved
.github/workflows/events-nightly.yml Outdated Show resolved Hide resolved
.github/workflows/ci_test-conda.yml Outdated Show resolved Hide resolved
.github/workflows/ci_dockers.yml Outdated Show resolved Hide resolved
@mergify mergify bot added the has conflicts label Nov 5, 2021
@mergify mergify bot added ready PRs ready to be merged and removed has conflicts labels Nov 5, 2021
@carmocca
Copy link
Member Author

carmocca commented Nov 5, 2021

@justusschock we could do that but would complicate CI a bit since we would need logic to handle 1.7 and 1.7.1 separately, additionally, we would need to generate both docker images.

I think 1.7.1 is fine looking at the release notes: https://github.com/pytorch/pytorch/releases/tag/v1.7.1
and download stats per version: https://pepy.tech/project/torch?versions=1.7.0&versions=1.7.1&versions=1.10.0&versions=1.6.0

And technically this should have been done before the 1.5 release, because it supports 1.6-1.10 (5 versions)

so we should actually drop 1.7 too before the 1.6 release, as it should support 1.8-1.11 (4 versions)

edit: found a solution using wildcard in c498a3b

@carmocca carmocca changed the title Drop torch 1.6 support and testing Drop torch 1.6 support Nov 6, 2021
Copy link
Contributor

@akihironitta akihironitta left a comment

Choose a reason for hiding this comment

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

@carmocca carmocca mentioned this pull request Nov 7, 2021
10 tasks
@carmocca
Copy link
Member Author

carmocca commented Nov 7, 2021

Does this mean we can also deprecate the following lines and remove it in pl v1.7, right?

yes, in #10390

@carmocca carmocca enabled auto-merge (squash) November 7, 2021 02:01
@Borda
Copy link
Member

Borda commented Nov 7, 2021

@carmocca I would rather keep it on master as we are going to merge bug fixes which will be conservatively cherry picked to release/1.5.x where we still support PT 1.6
So it may yield in extra work and debugging if the fix is touching some old PT functionality...

@carmocca
Copy link
Member Author

carmocca commented Nov 7, 2021

@Borda This PR should have landed with 1.5 already to comply with the 4 torch versions window.

Still, this can be merged to master and the milestone changed if you want to land a specific fix for 1.5.0 with 1.6

@mergify mergify bot added the has conflicts label Nov 8, 2021
Copy link
Contributor

@tchaton tchaton left a comment

Choose a reason for hiding this comment

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

LGTM !

requirements/adjust_versions.py Show resolved Hide resolved
@mergify mergify bot removed the has conflicts label Nov 8, 2021
@tchaton tchaton enabled auto-merge (squash) November 8, 2021 10:58
@tchaton tchaton merged commit 5b90253 into master Nov 8, 2021
@tchaton tchaton deleted the ci/drop-1.6 branch November 8, 2021 11:29
@carmocca carmocca modified the milestones: 1.5.x, 1.6 Nov 8, 2021
Raalsky pushed a commit to neptune-ai/pytorch-lightning that referenced this pull request Nov 23, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
ci Continuous Integration ready PRs ready to be merged
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

6 participants