Adding Weights classes for Resnet classification models #4655

jdsgomes · 2021-10-19T14:09:10Z

Helps #4672

cc @pmeier @bjuncek

datumbox

@jdsgomes Look great. Just a minor typo:

torchvision/prototype/models/resnet.py

jdsgomes · 2021-10-19T20:44:24Z

Local prototype test results

~/v/test ❯❯❯ pytest test_prototype_models.py -vvv                                                                                                                                            ✘ 1 
====================================================================================== test session starts =======================================================================================
platform darwin -- Python 3.8.8, pytest-6.2.3, py-1.10.0, pluggy-0.13.1 -- /Users/jdsgomes/homebrew/anaconda3/bin/python
cachedir: .pytest_cache
rootdir: /Users/jdsgomes/vision, configfile: pytest.ini
plugins: anyio-2.2.0
collected 36 items                                                                                                                                                                               

test_prototype_models.py::test_classification_model[cpu-resnet18] PASSED                                                                                                                   [  2%]
test_prototype_models.py::test_classification_model[cpu-resnet34] PASSED                                                                                                                   [  5%]
test_prototype_models.py::test_classification_model[cpu-resnet50] PASSED                                                                                                                   [  8%]
test_prototype_models.py::test_classification_model[cpu-resnet101] PASSED                                                                                                                  [ 11%]
test_prototype_models.py::test_classification_model[cpu-resnet152] PASSED                                                                                                                  [ 13%]
test_prototype_models.py::test_classification_model[cpu-resnext50_32x4d] PASSED                                                                                                            [ 16%]
test_prototype_models.py::test_classification_model[cpu-resnext101_32x8d] PASSED                                                                                                           [ 19%]
test_prototype_models.py::test_classification_model[cpu-wide_resnet50_2] PASSED                                                                                                            [ 22%]
test_prototype_models.py::test_classification_model[cpu-wide_resnet101_2] PASSED                                                                                                           [ 25%]
test_prototype_models.py::test_classification_model[cuda-resnet18] SKIPPED (CUDA device not available)                                                                                     [ 27%]
test_prototype_models.py::test_classification_model[cuda-resnet34] SKIPPED (CUDA device not available)                                                                                     [ 30%]
test_prototype_models.py::test_classification_model[cuda-resnet50] SKIPPED (CUDA device not available)                                                                                     [ 33%]
test_prototype_models.py::test_classification_model[cuda-resnet101] SKIPPED (CUDA device not available)                                                                                    [ 36%]
test_prototype_models.py::test_classification_model[cuda-resnet152] SKIPPED (CUDA device not available)                                                                                    [ 38%]
test_prototype_models.py::test_classification_model[cuda-resnext50_32x4d] SKIPPED (CUDA device not available)                                                                              [ 41%]
test_prototype_models.py::test_classification_model[cuda-resnext101_32x8d] SKIPPED (CUDA device not available)                                                                             [ 44%]
test_prototype_models.py::test_classification_model[cuda-wide_resnet50_2] SKIPPED (CUDA device not available)                                                                              [ 47%]
test_prototype_models.py::test_classification_model[cuda-wide_resnet101_2] SKIPPED (CUDA device not available)                                                                             [ 50%]
test_prototype_models.py::test_old_vs_new_classification_factory[cpu-resnet18] PASSED                                                                                                      [ 52%]
test_prototype_models.py::test_old_vs_new_classification_factory[cpu-resnet34] PASSED                                                                                                      [ 55%]
test_prototype_models.py::test_old_vs_new_classification_factory[cpu-resnet50] PASSED                                                                                                      [ 58%]
test_prototype_models.py::test_old_vs_new_classification_factory[cpu-resnet101] PASSED                                                                                                     [ 61%]
test_prototype_models.py::test_old_vs_new_classification_factory[cpu-resnet152] PASSED                                                                                                     [ 63%]
test_prototype_models.py::test_old_vs_new_classification_factory[cpu-resnext50_32x4d] PASSED                                                                                               [ 66%]
test_prototype_models.py::test_old_vs_new_classification_factory[cpu-resnext101_32x8d] PASSED                                                                                              [ 69%]
test_prototype_models.py::test_old_vs_new_classification_factory[cpu-wide_resnet50_2] PASSED                                                                                               [ 72%]
test_prototype_models.py::test_old_vs_new_classification_factory[cpu-wide_resnet101_2] PASSED                                                                                              [ 75%]
test_prototype_models.py::test_old_vs_new_classification_factory[cuda-resnet18] SKIPPED (CUDA device not available)                                                                        [ 77%]
test_prototype_models.py::test_old_vs_new_classification_factory[cuda-resnet34] SKIPPED (CUDA device not available)                                                                        [ 80%]
test_prototype_models.py::test_old_vs_new_classification_factory[cuda-resnet50] SKIPPED (CUDA device not available)                                                                        [ 83%]
test_prototype_models.py::test_old_vs_new_classification_factory[cuda-resnet101] SKIPPED (CUDA device not available)                                                                       [ 86%]
test_prototype_models.py::test_old_vs_new_classification_factory[cuda-resnet152] SKIPPED (CUDA device not available)                                                                       [ 88%]
test_prototype_models.py::test_old_vs_new_classification_factory[cuda-resnext50_32x4d] SKIPPED (CUDA device not available)                                                                 [ 91%]
test_prototype_models.py::test_old_vs_new_classification_factory[cuda-resnext101_32x8d] SKIPPED (CUDA device not available)                                                                [ 94%]
test_prototype_models.py::test_old_vs_new_classification_factory[cuda-wide_resnet50_2] SKIPPED (CUDA device not available)                                                                 [ 97%]
test_prototype_models.py::test_old_vs_new_classification_factory[cuda-wide_resnet101_2] SKIPPED (CUDA device not available)                                                                [100%]

datumbox

Thanks for adding tests, this is very much needed. I left a couple of comments, let me know what you think.

.circleci/unittest/linux/scripts/run_test.sh

test/test_prototype_models.py

Co-authored-by: Vasilis Vryniotis <[email protected]>

datumbox

Last set of optional nitpicks, feel free to accept/decline at will.

Also we will need to adapt this slightly once #4664 is merged but other than this, we should be good to go.

test/test_prototype_models.py

Co-authored-by: Vasilis Vryniotis <[email protected]>

test/test_prototype_models.py

Co-authored-by: Vasilis Vryniotis <[email protected]>

datumbox

Thanks a lot for your PR and patience @jdsgomes.

I think it looks great, I'll merge once the CI turns green.

github-actions · 2021-10-20T13:21:56Z

Hey @datumbox!

You merged this PR, but no labels were added. The list of valid labels is available at https://github.com/pytorch/vision/blob/main/.github/process_commit.py

Summary: * adding Weights classes for Resnet classification models * Replacing BasicBlock by Bottleneck in all but 3 model contructors * adding tests for prototype models * fixing typo in environment variable * Update test/test_prototype_models.py * changing default value for PYTORCH_TEST_WITH_PROTOTYPE * adding checks to compare outputs of the prototype vs old models * refactoring prototype tests * removing unused imports * applying ufmt * Update test/test_prototype_models.py * Update test/test_prototype_models.py * Update test/test_prototype_models.py * Update test/test_prototype_models.py * Update test/test_prototype_models.py * Update test/test_prototype_models.py * Update test/test_prototype_models.py Reviewed By: NicolasHug Differential Revision: D31916320 fbshipit-source-id: 6a379c2318de8448bf2b89174e789b866071eba5 Co-authored-by: Vasilis Vryniotis <[email protected]> Co-authored-by: Vasilis Vryniotis <[email protected]> Co-authored-by: Vasilis Vryniotis <[email protected]> Co-authored-by: Vasilis Vryniotis <[email protected]> Co-authored-by: Vasilis Vryniotis <[email protected]> Co-authored-by: Vasilis Vryniotis <[email protected]> Co-authored-by: Vasilis Vryniotis <[email protected]> Co-authored-by: Vasilis Vryniotis <[email protected]> Co-authored-by: Vasilis Vryniotis <[email protected]>

* adding Weights classes for Resnet classification models * Replacing BasicBlock by Bottleneck in all but 3 model contructors * adding tests for prototype models * fixing typo in environment variable * Update test/test_prototype_models.py Co-authored-by: Vasilis Vryniotis <[email protected]> * changing default value for PYTORCH_TEST_WITH_PROTOTYPE * adding checks to compare outputs of the prototype vs old models * refactoring prototype tests * removing unused imports * applying ufmt * Update test/test_prototype_models.py Co-authored-by: Vasilis Vryniotis <[email protected]> * Update test/test_prototype_models.py Co-authored-by: Vasilis Vryniotis <[email protected]> * Update test/test_prototype_models.py Co-authored-by: Vasilis Vryniotis <[email protected]> * Update test/test_prototype_models.py Co-authored-by: Vasilis Vryniotis <[email protected]> * Update test/test_prototype_models.py Co-authored-by: Vasilis Vryniotis <[email protected]> * Update test/test_prototype_models.py Co-authored-by: Vasilis Vryniotis <[email protected]> * Update test/test_prototype_models.py Co-authored-by: Vasilis Vryniotis <[email protected]> Co-authored-by: Vasilis Vryniotis <[email protected]>

adding Weights classes for Resnet classification models

e48df51

jdsgomes requested a review from datumbox October 19, 2021 14:09

pytorch-probot bot added the ciflow/default label Oct 19, 2021

datumbox reviewed Oct 19, 2021

View reviewed changes

Replacing BasicBlock by Bottleneck in all but 3 model contructors

7354381

facebook-github-bot added the cla signed label Oct 19, 2021

datumbox mentioned this pull request Oct 19, 2021

Extend TorchVision's existing model builders to support multiple weights #4611

Closed

adding tests for prototype models

9264a52

fixing typo in environment variable

17fa569

datumbox reviewed Oct 20, 2021

View reviewed changes

.circleci/unittest/linux/scripts/run_test.sh Outdated Show resolved Hide resolved

test/test_prototype_models.py Show resolved Hide resolved

test/test_prototype_models.py Outdated Show resolved Hide resolved

jdsgomes and others added 6 commits October 20, 2021 09:24

Update test/test_prototype_models.py

72ba4e3

Co-authored-by: Vasilis Vryniotis <[email protected]>

changing default value for PYTORCH_TEST_WITH_PROTOTYPE

ae412e1

adding checks to compare outputs of the prototype vs old models

df6cd1f

refactoring prototype tests

9e4200d

removing unused imports

0b49b70

applying ufmt

b877d6b

datumbox reviewed Oct 20, 2021

View reviewed changes

jdsgomes and others added 6 commits October 20, 2021 10:51

Update test/test_prototype_models.py

069c832

Co-authored-by: Vasilis Vryniotis <[email protected]>

Update test/test_prototype_models.py

20fa4da

Co-authored-by: Vasilis Vryniotis <[email protected]>

Update test/test_prototype_models.py

d594973

Co-authored-by: Vasilis Vryniotis <[email protected]>

Update test/test_prototype_models.py

da969ce

Co-authored-by: Vasilis Vryniotis <[email protected]>

Update test/test_prototype_models.py

b737bf6

Co-authored-by: Vasilis Vryniotis <[email protected]>

Update test/test_prototype_models.py

814aeb8

Co-authored-by: Vasilis Vryniotis <[email protected]>

datumbox reviewed Oct 20, 2021

View reviewed changes

test/test_prototype_models.py Outdated Show resolved Hide resolved

jdsgomes and others added 2 commits October 20, 2021 13:18

Merge branch 'main' into multiweight/resnet

c4048b0

Update test/test_prototype_models.py

344b84c

Co-authored-by: Vasilis Vryniotis <[email protected]>

datumbox approved these changes Oct 20, 2021

View reviewed changes

Merge branch 'main' into multiweight/resnet

c93a653

datumbox merged commit 80e6aff into pytorch:main Oct 20, 2021

datumbox added module: tests prototype labels Oct 20, 2021

datumbox linked an issue Oct 20, 2021 that may be closed by this pull request

Multi-pretrained weights: Add model builders for Classification #4672

Closed

13 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adding Weights classes for Resnet classification models #4655

Adding Weights classes for Resnet classification models #4655

jdsgomes commented Oct 19, 2021 •

edited by datumbox

Loading

datumbox left a comment

jdsgomes commented Oct 19, 2021

datumbox left a comment

datumbox left a comment

datumbox left a comment

github-actions bot commented Oct 20, 2021

Adding Weights classes for Resnet classification models #4655

Adding Weights classes for Resnet classification models #4655

Conversation

jdsgomes commented Oct 19, 2021 • edited by datumbox Loading

datumbox left a comment

Choose a reason for hiding this comment

jdsgomes commented Oct 19, 2021

datumbox left a comment

Choose a reason for hiding this comment

datumbox left a comment

Choose a reason for hiding this comment

datumbox left a comment

Choose a reason for hiding this comment

github-actions bot commented Oct 20, 2021

jdsgomes commented Oct 19, 2021 •

edited by datumbox

Loading