Remove deprecated trainer flag `Trainer.distributed_backend` in favor of `Trainer.accelerator` #9246

Tshimanga · 2021-09-01T05:50:47Z

What does this PR do?

Removes deprecated Trainer.distributed_backend flag as scheduled for 1.5

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)
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

Anyone in the community is welcome to review the PR.
Before you start reviewing make sure you have read Review guidelines. In short, see the following bullet-list:

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

Did you have fun?

Make sure you had fun coding 🙃

tchaton

LGTM ! @Tshimanga You are missing some and you have failing tests due to this :)

kaushikb11

Thanks @Tshimanga! But Failing tests. Also, at the moment accelerator flag supports Accelerators as well as Training Type plugins. We wish to decouple it, we would be taking a step back with the changes in this PR.

To read more about it, you could refer to #9053

Tshimanga · 2021-09-01T14:25:48Z

@tchaton @kaushikb11 Reading over the conversation in #9053 it sounds like there's a lack of consensus for a pretty impactful redesign of accelerators and training type plugins. I can resolve the failures in this PR, but it feels like the refactor should come first to do this right. I think the deprecation of distribution_backend was perhaps premature and it's removal should be kicked back to 1.6.

Thoughts?

tchaton · 2021-09-02T08:06:04Z

distribution_backend

Hey @Tshimanga,

Yes, we are still debating and the proposal from #9053 is definitely a big BC. If you have any suggestion, feel free to add to the conversation.

But let's pause this depreciation until resolution. Thanks for your hard work !

Best,
T.C

Tshimanga · 2021-09-02T23:23:13Z

@tchaton should I move the deprecation test to 1.6 in the meantime?

tchaton · 2021-09-03T07:28:17Z

Hey @Tshimanga,

We still have a bit of time, let's see if we consensus is reached :)

Still 4-5 weeks left before 1.5.

Best,
T.C

stale · 2021-09-19T00:46:05Z

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed in 7 days if no further activity occurs. If you need further help see our docs: https://pytorch-lightning.readthedocs.io/en/latest/generated/CONTRIBUTING.html#pull-request or ask the assistance of a core contributor here or on Slack. Thank you for your contributions.

stale · 2021-09-24T10:34:46Z

This pull request is going to be closed. Please feel free to reopen it create a new from the actual master.

for more information, see https://pre-commit.ci

…StatsMonitor` (Lightning-AI#9924) Co-authored-by: Carlos Mocholí <[email protected]> Co-authored-by: Nicki Skafte Detlefsen <[email protected]> Co-authored-by: Rohit Gupta <[email protected]>

Co-authored-by: tchaton <[email protected]>

Co-authored-by: Adrian Wälchli <[email protected]> Co-authored-by: thomas chaton <[email protected]>

… `DeviceStatsMonitor` callback (Lightning-AI#9921) Co-authored-by: Kaushik B <[email protected]> Co-authored-by: Rohit Gupta <[email protected]>

Co-authored-by: tchaton <[email protected]>

…rategies (Lightning-AI#9931) Co-authored-by: Carlos Mocholí <[email protected]>

…ightning-AI#9934)

…AI#9963) Co-authored-by: ronif <[email protected]> Co-authored-by: Adrian Wälchli <[email protected]>

…ightning-AI#9937)

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by: Carlos Mocholí <[email protected]> Co-authored-by: thomas chaton <[email protected]> Co-authored-by: edenlightning <[email protected]>

Co-authored-by: thomas chaton <[email protected]>

* fix * update tests * chlog * skip windows

Co-authored-by: Carlos Mocholí <[email protected]>

rohitgr7 · 2021-10-18T16:38:21Z

I somewhere messed up the with git here.. closing this one .. will open another

Tshimanga requested review from awaelchli, Borda, carmocca, justusschock, kaushikb11, SeanNaren, tchaton and williamFalcon as code owners September 1, 2021 05:50

tchaton approved these changes Sep 1, 2021

View reviewed changes

mergify bot added the has conflicts label Sep 1, 2021

kaushikb11 suggested changes Sep 1, 2021

View reviewed changes

mergify bot added has conflicts and removed has conflicts labels Sep 1, 2021

mergify bot added the has conflicts label Sep 2, 2021

carmocca marked this pull request as draft September 4, 2021 02:28

stale bot added the won't fix This will not be worked on label Sep 19, 2021

stale bot closed this Sep 24, 2021

awaelchli mentioned this pull request Sep 28, 2021

Remove deprecated property is_using_torchelastic from AcceleratorConnector #9729

Merged

12 tasks

awaelchli reopened this Sep 28, 2021

stale bot removed the won't fix This will not be worked on label Sep 28, 2021

Tshimanga and others added 3 commits September 28, 2021 18:36

scheduled removal of Trainer.distributed_backend

cc2b274

update CHANGELOG.md

21becf0

[pre-commit.ci] auto fixes from pre-commit.com hooks

d7a9c45

for more information, see https://pre-commit.ci

daniellepintz and others added 27 commits October 18, 2021 21:48

Deprecate GPUStatsMonitor and XLAStatsMonitor in favor of `Device…

8eb832b

…StatsMonitor` (Lightning-AI#9924) Co-authored-by: Carlos Mocholí <[email protected]> Co-authored-by: Nicki Skafte Detlefsen <[email protected]> Co-authored-by: Rohit Gupta <[email protected]>

Refactor tests for TPU Accelerator (Lightning-AI#9718)

830839d

Co-authored-by: tchaton <[email protected]>

Single-process multi-node CPU training (Lightning-AI#9603)

07ba0b9

Co-authored-by: Adrian Wälchli <[email protected]> Co-authored-by: thomas chaton <[email protected]>

Deprecate log_gpu_memory, gpu_metrics, and util funcs in favor of…

ba2efa2

… `DeviceStatsMonitor` callback (Lightning-AI#9921) Co-authored-by: Kaushik B <[email protected]> Co-authored-by: Rohit Gupta <[email protected]>

Add support for len(datamodule) (Lightning-AI#9895)

12ac06b

Co-authored-by: tchaton <[email protected]>

Validate the precision input earlier (Lightning-AI#9763)

52221c0

Use non-deprecated options in tests (Lightning-AI#9949)

7938922

(1/n) tests: Use strategy flag instead of accelerator for training st…

99af0c5

…rategies (Lightning-AI#9931) Co-authored-by: Carlos Mocholí <[email protected]>

Fixed use of LightningCLI in computer_vision_fine_tuning.py example (L…

bca1b66

…ightning-AI#9934)

Avoid deprecation warning after Lightning-AI#9901 (Lightning-AI#9951)

b66ecfc

Fix issue with no-init dataclass fields in move_to_device (Lightning-…

2c5e330

…AI#9963) Co-authored-by: ronif <[email protected]> Co-authored-by: Adrian Wälchli <[email protected]>

Fix LightningOptimizer step and toggling logic (Lightning-AI#9958)

e6ec14e

Update accelerator connector messages after the addition of strategy (L…

0e36a1c

…ightning-AI#9937)

loop customization docs (Lightning-AI#9609)

84dd799

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by: Carlos Mocholí <[email protected]> Co-authored-by: thomas chaton <[email protected]> Co-authored-by: edenlightning <[email protected]>

reset val dataloader for binsearch (Lightning-AI#9975)

da126ba

Fix self.log(on_epoch=True) on_batch_start (Lightning-AI#9780)

db8470a

Remove deprecated DataModule.dims usage in tests (Lightning-AI#9948)

206d6a0

Update resume_from_checkpoint docs (Lightning-AI#9952)

4dcc078

Remove manual tracking of optimizer steps (Lightning-AI#9957)

73e0a57

Introduce PrecisionPlugin.forward_context() (Lightning-AI#9988)

6a385be

Co-authored-by: thomas chaton <[email protected]>

Fix logic to check for spawn in worker_check (Lightning-AI#9902)

13fdba7

* fix * update tests * chlog * skip windows

Add unit tests for pl.utilities.grads (Lightning-AI#9765)

220fea7

Co-authored-by: Carlos Mocholí <[email protected]>

Add KFold Loop example (Lightning-AI#9965)

431919f

Add typing for LightningOptimizer (Lightning-AI#9990)

3c303e9

scheduled removal of Trainer.distributed_backend

3100dab

update CHANGELOG.md

4a0c479

update

a2e2e52

rohitgr7 closed this Oct 18, 2021

rohitgr7 mentioned this pull request Oct 19, 2021

Remove deprecated distributed_backend from Trainer #10017

Merged

12 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Remove deprecated trainer flag `Trainer.distributed_backend` in favor of `Trainer.accelerator` #9246

Remove deprecated trainer flag `Trainer.distributed_backend` in favor of `Trainer.accelerator` #9246

Uh oh!

Tshimanga commented Sep 1, 2021

Uh oh!

tchaton left a comment •

edited

Loading

Uh oh!

kaushikb11 left a comment •

edited

Loading

Uh oh!

Tshimanga commented Sep 1, 2021

Uh oh!

tchaton commented Sep 2, 2021

Uh oh!

Tshimanga commented Sep 2, 2021

Uh oh!

tchaton commented Sep 3, 2021 •

edited by carmocca

Loading

Uh oh!

stale bot commented Sep 19, 2021

Uh oh!

stale bot commented Sep 24, 2021

Uh oh!

rohitgr7 commented Oct 18, 2021

Uh oh!

Uh oh!

Remove deprecated trainer flag Trainer.distributed_backend in favor of Trainer.accelerator #9246

Remove deprecated trainer flag Trainer.distributed_backend in favor of Trainer.accelerator #9246

Uh oh!

Conversation

Tshimanga commented Sep 1, 2021

What does this PR do?

Before submitting

PR review

Did you have fun?

Uh oh!

tchaton left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kaushikb11 left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Tshimanga commented Sep 1, 2021

Uh oh!

tchaton commented Sep 2, 2021

Uh oh!

Tshimanga commented Sep 2, 2021

Uh oh!

tchaton commented Sep 3, 2021 • edited by carmocca Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

stale bot commented Sep 19, 2021

Uh oh!

stale bot commented Sep 24, 2021

Uh oh!

rohitgr7 commented Oct 18, 2021

Uh oh!

Uh oh!

Remove deprecated trainer flag `Trainer.distributed_backend` in favor of `Trainer.accelerator` #9246

Remove deprecated trainer flag `Trainer.distributed_backend` in favor of `Trainer.accelerator` #9246

tchaton left a comment •

edited

Loading

kaushikb11 left a comment •

edited

Loading

tchaton commented Sep 3, 2021 •

edited by carmocca

Loading