Port test/test_models.py to pytest

[NicolasHug]

Currently, most tests in test/test_models.py rely on unittest.TestCase. Now that we support pytest, we want to remove the use of the unittest module.

For a similar issue: see #3945 and #3956

Instructions

There are many tests in this file, but it should be possible to port all of them in one PR as they're all pretty straightforward. If you're interested in this issue, please comment below to indicate that you started working on a PR. Look below for some porting tips, and please don't hesitate to ask for help. Thanks!

In this file there are already some test functions like:

def test_classification_model(model_name, dev):
    ModelTester()._test_classification_model(model_name, dev)

For those, we should just copy the body of _test_classification_model into test_classification_model so that we can get rid of the ModelTester class altogether.

@pytest.mark.parametrize('dev', _devs) should be changed into @pytest.mark.parametrize('dev', cpu_and_gpu()) where cpu_and_gpu() is in common_utils.

The tests that need cuda (e.g. test_fasterrcnn_switch_devices) should use the @needs_cuda decorator, also from common_utils. The test that don't need cuda should use the @cpu_only decorator.

How to port a test to pytest

Porting a test from unittest to pytest is usually fairly straightforward. For a typical example, see https://github.com/pytorch/vision/pull/3907/files:

• take the test method out of the Tester(unittest.TestCase) class and just declare it as a function
• Replace @unittest.skipIf with pytest.mark.skipif(cond, reason=...)
• remove any use of self.assertXYZ.
  • Typically assertEqual(a, b) can be replaced by assert a == b when a and b are pure python objects (scalars, tuples, lists), and otherwise we can rely on assert_equal which is already used in the file.
  • self.assertRaises should be replaced with the pytest.raises(Exp, match=...): context manager, as done in https://github.com/pytorch/vision/pull/3907/files. Same for warnings with pytest.warns
  • self.assertTrue should be replaced with a plain assert
• When a function uses for loops to tests multiple parameter values, one should usepytest.mark.parametrize instead, as done e.g. in https://github.com/pytorch/vision/pull/3907/files.

cc @pmeier

[AnirudhDagar]

Hi @NicolasHug,
I'll start working on a PR if no one has claimed it yet.

[saswatpp]

Hi @AnirudhDagar , i'd like to work on this issue. Let me know if you want me to help you.

[NicolasHug]

I'll soon open another similar issue @saswatpp , I'll ping you there

[saswatpp]

@NicolasHug Okay sure! Thank you 👍

[NicolasHug]

done in #3978, thanks again @AnirudhDagar