PyTorch 1.7 Stable support #3821

ydcjeff · 2020-10-03T16:54:57Z

What does this PR do?

Conda 1.7 test is failing again and found out PyTorch 1.7 is in RC stage and 1.7 RC binary cannot be found on conda.

This PR

prepares for PyTorch 1.7 stable once it's released
doesn't add PyTorch 1.7 new features
will fix the failed tests for 1.7 Stable where applicable

Before submitting

Was this discussed/approved via a Github issue? (no need for typos and docs improvements)
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? Otherwise, we ask you to create a separate PR for every change.
Did you make sure to update the documentation with your changes?
Did you write any new necessary tests?
Did you verify new and existing tests pass locally with your changes?
If you made a notable change (that affects users), did you update the CHANGELOG?

PR review

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

codecov · 2020-10-03T17:11:21Z

Codecov Report

Merging #3821 into master will not change coverage.
The diff coverage is 100%.

@@          Coverage Diff           @@
##           master   #3821   +/-   ##
======================================
  Coverage      93%     93%           
======================================
  Files         113     113           
  Lines        8179    8179           
======================================
  Hits         7600    7600           
  Misses        579     579

Borda · 2020-10-03T23:59:43Z

in meantime, we can drop it from CI #3833 or add a special install task as it is available on pip --pre as pre-release

ydcjeff · 2020-10-04T04:53:36Z

Is RC version available on pypi? I didn't find it.
PyTorch released RC version via only github release

Borda · 2020-10-26T17:31:21Z

@ydcjeff can we now install PT 1.7 from regular channels - PyPI and Conda?

ydcjeff · 2020-10-26T17:35:40Z

@ydcjeff can we now install PT 1.7 from regular channels - PyPI and Conda?

I think not yet

pep8speaks · 2020-10-27T10:15:13Z

Hello @ydcjeff! Thanks for updating this PR.

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2020-10-30 14:46:36 UTC

ydcjeff · 2020-10-27T15:42:44Z

@Borda There is no 1.7 docker on docker hub.
Should I push it from here?

ydcjeff · 2020-10-29T16:28:58Z

@awaelchli @tchaton @SeanNaren
ready for final review.

tests/checkpointing/test_model_checkpoint.py

ydcjeff · 2020-10-30T14:34:38Z

Once the CI passed, if ok with the changes, ready to merge.

SeanNaren · 2020-10-30T14:35:52Z

There is a failing test which seems to not find a pytorch 1.7 base docker image. Do we need to manually create this?

ydcjeff · 2020-10-30T14:39:00Z

There is a failing test which seems to not find a pytorch 1.7 base docker image. Do we need to manually create this?

@SeanNaren we create docker image nightly and after this is merged, that will be created as well.
But yes, if we wish to create and push to docker hub, will wait for that.

tests/checkpointing/test_model_checkpoint.py

Co-authored-by: Rohit Gupta <rohitgr1998@gmail.com>

ydcjeff · 2020-10-30T15:14:48Z

pinning @SeanNaren for final review.

SeanNaren · 2020-10-30T15:42:45Z

thanks @ydcjeff, will monitor to see if everything goes smoothly with the ci docker workflows!

SeanNaren · 2020-10-30T15:58:09Z

hey @ydcjeff do we need to add 1.7 to the docker build push here? https://github.com/PyTorchLightning/pytorch-lightning/blob/master/.github/workflows/docker-builds.yml#L17

ydcjeff · 2020-10-30T16:01:13Z

hey @ydcjeff do we need to add 1.7 to the docker build push here? https://github.com/PyTorchLightning/pytorch-lightning/blob/master/.github/workflows/docker-builds.yml#L17

no need currently, this is pytorch lightning docker image to be used by our users
this is not for CI testing.

SeanNaren · 2020-10-30T16:38:59Z

Are we missing the public pytorch 1.7 docker image for this test? https://github.com/PyTorchLightning/pytorch-lightning/runs/1332900421

Any idea what's happening here? https://github.com/PyTorchLightning/pytorch-lightning/runs/1332770138

ydcjeff · 2020-10-30T16:41:53Z

Are we missing the public pytorch 1.7 docker image for this test? https://github.com/PyTorchLightning/pytorch-lightning/runs/1332900421

Yes, the image will be built at UTC 0 nightly, so the test will run tmr.
can't trigger the docker build manually.

Any idea what's happening here? https://github.com/PyTorchLightning/pytorch-lightning/runs/1332770138

No idea, that probably comes from master before this PR.

* prepare for 1.7 support [ci skip] * tpu [ci skip] * test run 1.7 * all 1.7, needs to fix tests * couple with torchvision * windows try * remove windows * 1.7 is here * on purpose fail [ci skip] * return [ci skip] * 1.7 docker * back to normal [ci skip] * change to some_val [ci skip] * add seed [ci skip] * 4 places [ci skip] * fail on purpose [ci skip] * verbose=True [ci skip] * use filename to track * use filename to track * monitor epoch + changelog * Update tests/checkpointing/test_model_checkpoint.py Co-authored-by: Rohit Gupta <rohitgr1998@gmail.com> Co-authored-by: Sean Naren <sean.narenthiran@gmail.com> Co-authored-by: Rohit Gupta <rohitgr1998@gmail.com> (cherry picked from commit 0f584fa)

ydcjeff marked this pull request as draft October 3, 2020 16:55

ydcjeff changed the title ~~PyTorch 1.7 Stable support [CI SKIP]~~ [WIP] PyTorch 1.7 Stable support [CI SKIP] Oct 3, 2020

Borda added the 3rd party Related to a 3rd-party label Oct 3, 2020

Borda mentioned this pull request Oct 4, 2020

Add step index in checkpoint name #3807

Merged

3 tasks

Borda changed the title ~~[WIP] PyTorch 1.7 Stable support [CI SKIP]~~ PyTorch 1.7 Stable support Oct 4, 2020

Borda added this to the 1.0.x milestone Oct 20, 2020

Borda added bug Something isn't working feature Is an improvement or enhancement priority: 0 High priority task labels Oct 26, 2020

ydcjeff added 3 commits October 27, 2020 00:09

prepare for 1.7 support [ci skip]

0e9c831

tpu [ci skip]

cfe1a33

test run 1.7

708b5db

ydcjeff changed the title ~~PyTorch 1.7 Stable support~~ PyTorch 1.7 Stable support [ci skip] Oct 26, 2020

ydcjeff added 4 commits October 27, 2020 00:58

all 1.7, needs to fix tests

65212e3

Merge branch 'master' into torch/prepare-1.7

f01bb42

couple with torchvision

d716c2c

windows try

705fc47

ydcjeff added 2 commits October 27, 2020 17:29

remove windows

d1037fa

1.7 is here

0042b2c

ydcjeff added 4 commits October 27, 2020 23:08

on purpose fail [ci skip]

e2c8ca6

return [ci skip]

b3c714f

Merge branch 'master' into torch/prepare-1.7

cc45669

1.7 docker

fde61df

fail on purpose [ci skip]

9c649c7

SeanNaren self-requested a review October 29, 2020 12:12

ydcjeff added 3 commits October 29, 2020 20:54

verbose=True [ci skip]

133901d

use filename to track

bc873f2

use filename to track

775089c

ydcjeff requested review from awaelchli and tchaton October 29, 2020 16:29

SeanNaren and others added 2 commits October 29, 2020 16:52

Merge branch 'master' into torch/prepare-1.7

b8ada75

Merge branch 'master' into torch/prepare-1.7

c5f7d06

awaelchli approved these changes Oct 29, 2020

View reviewed changes

tests/checkpointing/test_model_checkpoint.py Show resolved Hide resolved

Merge branch 'master' into torch/prepare-1.7

bbd1cb8

ydcjeff mentioned this pull request Oct 30, 2020

[docs] Fix ModelCheckpoint examples #4426

Merged

1 task

Jeff Yang added 2 commits October 30, 2020 20:11

Merge branch 'master' into torch/prepare-1.7

b6d29d6

monitor epoch + changelog

a4d4302

rohitgr7 reviewed Oct 30, 2020

View reviewed changes

tests/checkpointing/test_model_checkpoint.py Outdated Show resolved Hide resolved

rohitgr7 approved these changes Oct 30, 2020

View reviewed changes

Update tests/checkpointing/test_model_checkpoint.py

1f771e9

Co-authored-by: Rohit Gupta <rohitgr1998@gmail.com>

SeanNaren merged commit 0f584fa into Lightning-AI:master Oct 30, 2020

ydcjeff deleted the torch/prepare-1.7 branch October 30, 2020 15:43

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

PyTorch 1.7 Stable support #3821

PyTorch 1.7 Stable support #3821

ydcjeff commented Oct 3, 2020

codecov bot commented Oct 3, 2020 •

edited

Loading

Borda commented Oct 3, 2020

ydcjeff commented Oct 4, 2020

Borda commented Oct 26, 2020

ydcjeff commented Oct 26, 2020

pep8speaks commented Oct 27, 2020 •

edited

Loading

ydcjeff commented Oct 27, 2020 •

edited

Loading

ydcjeff commented Oct 29, 2020

ydcjeff commented Oct 30, 2020

SeanNaren commented Oct 30, 2020

ydcjeff commented Oct 30, 2020

ydcjeff commented Oct 30, 2020

SeanNaren commented Oct 30, 2020

SeanNaren commented Oct 30, 2020

ydcjeff commented Oct 30, 2020

SeanNaren commented Oct 30, 2020

ydcjeff commented Oct 30, 2020 •

edited

Loading

PyTorch 1.7 Stable support #3821

PyTorch 1.7 Stable support #3821

Conversation

ydcjeff commented Oct 3, 2020

What does this PR do?

Before submitting

PR review

Did you have fun?

codecov bot commented Oct 3, 2020 • edited Loading

Codecov Report

Borda commented Oct 3, 2020

ydcjeff commented Oct 4, 2020

Borda commented Oct 26, 2020

ydcjeff commented Oct 26, 2020

pep8speaks commented Oct 27, 2020 • edited Loading

Comment last updated at 2020-10-30 14:46:36 UTC

ydcjeff commented Oct 27, 2020 • edited Loading

ydcjeff commented Oct 29, 2020

ydcjeff commented Oct 30, 2020

SeanNaren commented Oct 30, 2020

ydcjeff commented Oct 30, 2020

ydcjeff commented Oct 30, 2020

SeanNaren commented Oct 30, 2020

SeanNaren commented Oct 30, 2020

ydcjeff commented Oct 30, 2020

SeanNaren commented Oct 30, 2020

ydcjeff commented Oct 30, 2020 • edited Loading

codecov bot commented Oct 3, 2020 •

edited

Loading

pep8speaks commented Oct 27, 2020 •

edited

Loading

ydcjeff commented Oct 27, 2020 •

edited

Loading

ydcjeff commented Oct 30, 2020 •

edited

Loading