Use numbers for input check in `roi_{average|max}_{pooling|align}_2d.py` #5636

knorth55 · 2018-11-07T10:25:35Z

~~Merge after #5634 and #5635.~~

use numpy.issubdtype instead of isinstance to check function args.
numpy.issubdtype supports int, float and numpy.integer, numpy.floating.

stale · 2019-02-19T02:15:14Z

This issue has been automatically marked as stale because it has not had recent activity. It will be closed after 30 days if no further activity occurs. Thank you for your contributions.

knorth55 · 2019-02-19T06:16:57Z

waiting for #5635, not stale

mitmul · 2019-02-21T07:50:20Z

@knorth55 Could you fix the conflicts?

knorth55 · 2019-02-21T15:27:16Z

I merged master branch and travis passed.

mitmul · 2019-02-28T09:15:13Z

Jenkins, test this please

chainer-ci · 2019-02-28T09:55:34Z

Jenkins CI test (for commit 3cd5e48, target branch master) failed with status FAILURE.
(For contributors, please wait until the reviewer confirms the details of the error.)

…d.py

chainer-ci · 2019-04-09T16:31:40Z

Jenkins CI test (for commit ae7d9ca, target branch master) succeeded!

knorth55 · 2019-04-09T18:44:31Z

@mitmul can you review this again ?

knorth55 · 2019-04-24T09:34:04Z

@Hakuyume kindly ping.

Hakuyume · 2019-05-07T03:43:30Z

numbers.Integral can do similar thing and I think it is more simple.

import numbers
import numpy as np

isinstance(int(0), numbers.Integral)  # True
isinstance(np.int32(0), numbers.Integral)  # True
isinstance(float(0), numbers.Integral)  # False

knorth55 · 2019-05-07T10:52:19Z

@Hakuyume i update to use numbers.

chainer-ci · 2019-06-11T06:39:50Z

Jenkins CI test (for commit 095a94d, target branch master) failed with status FAILURE.

mitmul · 2019-07-16T17:46:32Z

Jenkins, test this please

chainer-ci · 2019-07-16T18:34:07Z

Jenkins CI test (for commit 095a94d, target branch master) succeeded!

mitmul · 2019-07-17T04:32:52Z

@knorth55 I'm very sorry for late reaction to your PRs... Could you fix the conflicts? Then I'll re-run the tests and merge this.

knorth55 · 2019-07-17T04:41:56Z

@mitmul No problem. I resolved the conflict.

chainer-ci · 2019-07-20T09:54:06Z

@mitmul This pull-request is marked as st:test-and-merge, but there were no activities for the last 3 days. Could you check?

mitmul · 2019-07-21T10:01:39Z

Jenkins, test this please

chainer-ci · 2019-07-21T10:42:57Z

Jenkins CI test (for commit ccc6293, target branch master) failed with status FAILURE.

knorth55 · 2019-07-22T09:16:45Z

The test exceeds timeout limit in Python2.7.
Should I make the test ligheter?

chainer-ci · 2019-07-26T09:54:11Z

@mitmul This pull-request is marked as st:test-and-merge, but there were no activities for the last 3 days. Could you check?

chainer-ci · 2019-07-29T09:54:08Z

@mitmul This pull-request is marked as st:test-and-merge, but there were no activities for the last 3 days. Could you check?

chainer-ci · 2019-08-01T09:54:08Z

@mitmul This pull-request is marked as st:test-and-merge, but there were no activities for the last 3 days. Could you check?

chainer-ci · 2019-08-04T09:54:06Z

@mitmul This pull-request is marked as st:test-and-merge, but there were no activities for the last 3 days. Could you check?

toslunar · 2019-08-05T07:51:55Z

I'll rerun tests because the last two commits LGTM.

Jenkins, test this please.

chainer-ci · 2019-08-05T08:42:20Z

Jenkins CI test (for commit 89e90b4, target branch master) failed with status FAILURE.

knorth55 · 2019-08-06T09:13:52Z

Jenkins test failed, but it is not related to this PR.

17:35:21 =================================== FAILURES ===================================
17:35:21 __________ test_TrilTriu_param_4_{k=1, shape=(4, 3)}[float16-cuda:0] ___________
17:35:21 
17:35:21 device = cuda:0, args = (), kwargs = {'float_dtype': 'float16'}
17:35:21 backend_config = <BackendConfig use_chainerx=True chainerx_device='cuda:0' use_cuda=False cuda_device=None use_cudnn='never' cudnn_deterministic=False autotune=False cudnn_fast_batch_normalization=False use_ideep='never'>
17:35:21 obj = <chainer.testing._bundle.TestTrilTriu_param_4_{k=1, shape=(4, 3)} object at 0x7f53f8b31908>
17:35:21 
17:35:21     @pytest.mark.parametrize_device(devices)
17:35:21     def entry_func(device, *args, **kwargs):
17:35:21         backend_config = _make_backend_config(device.name)
17:35:21     
17:35:21         # Forward test
17:35:21         obj = cls()
17:35:21         try:
17:35:21             obj.setup(*args, **kwargs)
17:35:21             obj.run_test_forward(backend_config)
17:35:21         finally:
17:35:21             obj.teardown()
17:35:21     
17:35:21         # If this is a NumpyOpTest instance, skip backward/double-backward
17:35:21         # tests if the forward test succeeds with acceptable errors.
17:35:21         if isinstance(obj, NumpyOpTest):
17:35:21             if obj.is_forward_successful_with_accept_errors:
17:35:21                 return  # success with expected errors
17:35:21     
17:35:21         # Backward test
17:35:21         obj = cls()
17:35:21         try:
17:35:21             obj.setup(*args, **kwargs)
17:35:21 >           obj.run_test_backward(backend_config)
17:35:21 
17:35:21 args       = ()
17:35:21 backend_config = <BackendConfig use_chainerx=True chainerx_device='cuda:0' use_cuda=False cuda_device=None use_cudnn='never' cudnn_deterministic=False autotune=False cudnn_fast_batch_normalization=False use_ideep='never'>
17:35:21 cls        = <class 'chainer.testing._bundle.TestTrilTriu_param_4_{k=1, shape=(4, 3)}'>
17:35:21 device     = cuda:0
17:35:21 kwargs     = {'float_dtype': 'float16'}
17:35:21 obj        = <chainer.testing._bundle.TestTrilTriu_param_4_{k=1, shape=(4, 3)} object at 0x7f53f8b31908>
17:35:21 
17:35:21 /repo/tests/chainerx_tests/op_utils.py:366: 
17:35:21 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
17:35:21 /repo/tests/chainerx_tests/op_utils.py:95: in run_test_backward
17:35:21     super(OpTest, self).run_test_backward(backend_config)
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/testing/function_link.py:255: in run_test_backward
17:35:21     do_check()
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/testing/function_link.py:244: in do_check
17:35:21     **self.check_backward_options)
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/contextlib.py:99: in __exit__
17:35:21     self.gen.throw(type, value, traceback)
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/testing/function_link.py:36: in raise_if_fail
17:35:21     cls.fail(message, e)
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/testing/function_link.py:27: in fail
17:35:21     utils._raise_from(cls, message, exc)
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/utils/__init__.py:106: in _raise_from
17:35:21     six.raise_from(new_exc.with_traceback(orig_exc.__traceback__), None)
17:35:21 <string>:3: in raise_from
17:35:21     ???
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/testing/function_link.py:34: in raise_if_fail
17:35:21     yield
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/testing/function_link.py:244: in do_check
17:35:21     **self.check_backward_options)
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/gradient_check.py:900: in check_backward
17:35:21     detect_nondifferentiable, is_immutable_params=False
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/gradient_check.py:464: in run
17:35:21     self._run()
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/gradient_check.py:507: in _run
17:35:21     self._compare_gradients(gx_numeric, gx_backward, directions)
17:35:21 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
17:35:21 
17:35:21 self = <chainer.gradient_check._CheckBackward object at 0x7f53f8b31748>
17:35:21 gx_numeric = array(-0.00447911, shape=(), dtype=float64, device='cuda:0')
17:35:21 gx_backward = array(-0.00457345, shape=(), dtype=float64, device='cuda:0')
17:35:21 directions = [array([[-0.19084183, -0.13525285, 0.24186938],
17:35:21        [0.38141085, -0.54487473, -0.3863876 ],
17:35:21        [-0.17815417, 0.04524124, -0.48183729],
17:35:21        [0.11480988, -0.03716954, 0.12087778]], shape=(4, 3), dtype=float64, device='cuda:0')]
17:35:21 
17:35:21     def _compare_gradients(self, gx_numeric, gx_backward, directions):
17:35:21         atol = self.atol
17:35:21         rtol = self.rtol
17:35:21         # Compare the gradients
17:35:21         try:
17:35:21             testing.assert_allclose(
17:35:21                 gx_numeric, gx_backward, atol=atol, rtol=rtol)
17:35:21         except AssertionError as e:
17:35:21             eps = self.eps
17:35:21             xs = self.xs
17:35:21             gys = self.gys
17:35:21             f = six.StringIO()
17:35:21             f.write('check_backward failed (eps={} atol={} rtol={})\n'.format(
17:35:21                 eps, atol, rtol))
17:35:21             for i, x in enumerate(xs):
17:35:21                 f.write('inputs[{}]:\n'.format(i))
17:35:21                 f.write('{}\n'.format(x))
17:35:21             for i, gy in enumerate(gys):
17:35:21                 f.write('grad_outputs[{}]:\n'.format(i))
17:35:21                 f.write('{}\n'.format(gy))
17:35:21             for i, d in enumerate(directions):
17:35:21                 f.write('directions[{}]:\n'.format(i))
17:35:21                 f.write('{}\n'.format(d))
17:35:21             f.write('gradients (numeric):  {}\n'.format(gx_numeric))
17:35:21             f.write('gradients (backward): {}\n'.format(gx_backward))
17:35:21             f.write('\n')
17:35:21             f.write('x: numeric gradient, y: backward gradient')
17:35:21             f.write(str(e))
17:35:21 >           raise AssertionError(f.getvalue())
17:35:21 E           chainer.testing.function_link.FunctionTestError: backward is not implemented correctly
17:35:21 E           
17:35:21 E           (caused by)
17:35:21 E           AssertionError: check_backward failed (eps=0.001 atol=1e-05 rtol=0.005)
17:35:21 E           inputs[0]:
17:35:21 E           array([[0.9765625 , 0.35913086, 0.16003418],
17:35:21 E                  [0.87304688, 0.41284180, 0.34838867],
17:35:21 E                  [0.61523438, 0.55126953, 0.71191406],
17:35:21 E                  [0.40600586, 0.0482788 , 0.73046875]], shape=(4, 3), dtype=float16, device='cuda:0')
17:35:21 E           grad_outputs[0]:
17:35:21 E           array([[-0.29565430, 0.82910156, -0.86621094],
17:35:21 E                  [0.45214844, -0.57568359, 0.51074219],
17:35:21 E                  [-0.91455078, -0.56347656, 0.04678345],
17:35:21 E                  [-0.18273926, 0.54541016, -0.71386719]], shape=(4, 3), dtype=float16, device='cuda:0')
17:35:21 E           grad_outputs[1]:
17:35:21 E           array([[-0.82617188, -0.45629883, -0.77978516],
17:35:21 E                  [0.63574219, -0.13049316, 0.2536621 ],
17:35:21 E                  [-0.75195312, -0.43554688, -0.35913086],
17:35:21 E                  [0.80859375, 0.69677734, 0.96728516]], shape=(4, 3), dtype=float16, device='cuda:0')
17:35:21 E           directions[0]:
17:35:21 E           array([[-0.19084183, -0.13525285, 0.24186938],
17:35:21 E                  [0.38141085, -0.54487473, -0.3863876 ],
17:35:21 E                  [-0.17815417, 0.04524124, -0.48183729],
17:35:21 E                  [0.11480988, -0.03716954, 0.12087778]], shape=(4, 3), dtype=float64, device='cuda:0')
17:35:21 E           gradients (numeric):  array(-0.00447911, shape=(), dtype=float64, device='cuda:0')
17:35:21 E           gradients (backward): array(-0.00457345, shape=(), dtype=float64, device='cuda:0')
17:35:21 E           
17:35:21 E           x: numeric gradient, y: backward gradient
17:35:21 E           Not equal to tolerance rtol=0.005, atol=1e-05
17:35:21 E           
17:35:21 E           Mismatch: 100%
17:35:21 E           Max absolute difference: 9.43329122e-05
17:35:21 E           Max relative difference: 0.02062621
17:35:21 E            x: array(-0.004479)
17:35:21 E            y: array(-0.004573)
17:35:21 E           
17:35:21 E           assert_allclose failed: 
17:35:21 E             shape: () ()
17:35:21 E             dtype: float64 float64
17:35:21 E             i: (0,)
17:35:21 E             x[i]: -0.004479114874565315
17:35:21 E             y[i]: -0.0045734477868034085
17:35:21 E             relative error[i]: 0.020626213884039273
17:35:21 E             absolute error[i]: 9.433291223809315e-05
17:35:21 E             relative tolerance * |y[i]|: 2.286723893401704e-05
17:35:21 E             absolute tolerance: 1e-05
17:35:21 E             total tolerance: 3.2867238934017044e-05
17:35:21 E           x: -0.00447911
17:35:21 E           y: -0.00457345
17:35:21 
17:35:21 atol       = 1e-05
17:35:21 d          = array([[-0.19084183, -0.13525285, 0.24186938],
17:35:21        [0.38141085, -0.54487473, -0.3863876 ],
17:35:21        [-0.17815417, 0.04524124, -0.48183729],
17:35:21        [0.11480988, -0.03716954, 0.12087778]], shape=(4, 3), dtype=float64, device='cuda:0')
17:35:21 directions = [array([[-0.19084183, -0.13525285, 0.24186938],
17:35:21        [0.38141085, -0.54487473, -0.3863876 ],
17:35:21        [-0.17815417, 0.04524124, -0.48183729],
17:35:21        [0.11480988, -0.03716954, 0.12087778]], shape=(4, 3), dtype=float64, device='cuda:0')]
17:35:21 eps        = 0.001
17:35:21 f          = <_io.StringIO object at 0x7f53f8b3c168>
17:35:21 gx_backward = array(-0.00457345, shape=(), dtype=float64, device='cuda:0')
17:35:21 gx_numeric = array(-0.00447911, shape=(), dtype=float64, device='cuda:0')
17:35:21 gy         = array([[-0.82617188, -0.45629883, -0.77978516],
17:35:21        [0.63574219, -0.13049316, 0.2536621 ],
17:35:21        [-0.75195312, -0.43554688, -0.35913086],
17:35:21        [0.80859375, 0.69677734, 0.96728516]], shape=(4, 3), dtype=float16, device='cuda:0')
17:35:21 gys        = (array([[-0.29565430, 0.82910156, -0.86621094],
17:35:21        [0.45214844, -0.57568359, 0.51074219],
17:35:21        [-0.91455078, -0....-0.43554688, -0.35913086],
17:35:21        [0.80859375, 0.69677734, 0.96728516]], shape=(4, 3), dtype=float16, device='cuda:0'))
17:35:21 i          = 0
17:35:21 rtol       = 0.005
17:35:21 self       = <chainer.gradient_check._CheckBackward object at 0x7f53f8b31748>
17:35:21 x          = array([[0.9765625 , 0.35913086, 0.16003418],
17:35:21        [0.87304688, 0.41284180, 0.34838867],
17:35:21        [0.61523438, 0.55126953, 0.71191406],
17:35:21        [0.40600586, 0.0482788 , 0.73046875]], shape=(4, 3), dtype=float16, device='cuda:0')
17:35:21 xs         = (array([[0.9765625 , 0.35913086, 0.16003418],
17:35:21        [0.87304688, 0.41284180, 0.34838867],
17:35:21        [0.61523438, 0.55126953, 0.71191406],
17:35:21        [0.40600586, 0.0482788 , 0.73046875]], shape=(4, 3), dtype=float16, device='cuda:0'),)
17:35:21 
17:35:21 /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/gradient_check.py:537: FunctionTestError
17:35:22 =============================== warnings summary ===============================
17:35:22 tests/chainerx_tests/unit_tests/routines_tests/test_explog.py::test_Erf_param_0_{shape=(2, 2), in_dtypes=('float16',), out_dtype='float16'}[float16-native:0]
17:35:22   /workspace/conda/envs/testenv/lib/python3.6/site-packages/chainer/functions/math/erf.py:36: PerformanceWarning: SciPy is not available. Forward computation of erf in CPU can be slow without SciPy.
17:35:22     chainer.warnings.PerformanceWarning)
17:35:22 
17:35:22 -- Docs: https://docs.pytest.org/en/latest/warnings.html
17:35:22 =========================== short test summary info ============================
17:35:22 FAIL ../../../repo/tests/chainerx_tests/unit_tests/routines_tests/test_creation.py::test_TrilTriu_param_4_{k=1, shape=(4, 3)}[float16-cuda:0]
17:35:22 = 1 failed, 165337 passed, 5907 skipped, 41 xfailed, 1 warnings in 2079.43 seconds =
17:35:38 Build step 'Execute shell' marked build as failure

chainer-ci · 2019-08-09T09:54:08Z

@mitmul This pull-request is marked as st:test-and-merge, but there were no activities for the last 3 days. Could you check?

toslunar · 2019-08-11T22:17:38Z

Jenkins, test this please.

chainer-ci · 2019-08-11T23:08:25Z

Jenkins CI test (for commit 89e90b4, target branch master) succeeded!

chainer-ci · 2019-08-22T03:16:20Z

Jenkins CI test (for commit 89e90b4, target branch master) failed with status FAILURE.

beam2d added the st:blocked-by-another-pr State indicating that another ticket is preventing this ticket from being closed/merged. label Nov 21, 2018

beam2d assigned mitmul Nov 21, 2018

stale bot added the stale Not updated for a longer period of time. label Feb 19, 2019

stale bot removed the stale Not updated for a longer period of time. label Feb 19, 2019

knorth55 added 10 commits March 18, 2019 13:10

use numpy.issubdtype to support numpy.scalar in roi_average_align_2d.py

8a884f3

use numpy.issubdtype to support numpy.scalar in roi_average_pooling_2…

3eabb9d

…d.py

use numpy.issubdtype to support numpy.scalar in roi_max_pooling_2d.py

597df00

test numpy.scalar in test_roi_average_pooling_2d.py

320532b

test numpy.scalar in test_roi_max_pooling_2d.py

016cf7a

use numpy.issubdtype to support numpy.scalar in roi_max_align_2d.py

8fd38a9

test numpy.scalar in test_roi_max_align_2d.py

1a20421

refactor variable order in roi_max_align_2d.py

06a5048

test numpy.scalar in test_roi_average_align_2d.py

8c33f29

reduce test number for test_roi_{max|average}_{pooling|align}_2d.py

be3f948

knorth55 force-pushed the useissubdtype branch from 3cd5e48 to be3f948 Compare March 18, 2019 04:15

Merge remote-tracking branch 'origin/master' into useissubdtype

ae7d9ca

knorth55 added 2 commits May 7, 2019 19:25

Merge remote-tracking branch 'origin/master' into useissubdtype

ee12dfb

use numbers instead of numpy.issubdtype

8c90214

knorth55 changed the title ~~use numpy.issubdtype in roi_{average|max}_{pooling|align}_2d.py~~ use numbers for input check in roi_{average|max}_{pooling|align}_2d.py May 7, 2019

mitmul added st:test-and-merge State indicating that pull request is approved by a reviewer and can be merged after CI passes. and removed st:blocked-by-another-pr State indicating that another ticket is preventing this ticket from being closed/merged. labels Jul 16, 2019

Merge branch 'master' into useissubdtype

ccc6293

knorth55 added 2 commits July 23, 2019 17:51

Merge remote-tracking branch 'origin/master' into useissubdtype

9249ee6

make test lighter to avoid test timeout

89e90b4

mergify bot merged commit f6dbcd4 into chainer:master Aug 11, 2019

knorth55 deleted the useissubdtype branch August 12, 2019 00:51

toslunar added this to the v7.0.0b3 milestone Aug 12, 2019

kmaehashi changed the title ~~use numbers for input check in roi_{average|max}_{pooling|align}_2d.py~~ Use numbers for input check in roi_{average|max}_{pooling|align}_2d.py Aug 22, 2019

kmaehashi added the cat:enhancement Implementation that does not break interfaces. label Aug 22, 2019

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Use numbers for input check in `roi_{average|max}_{pooling|align}_2d.py` #5636

Use numbers for input check in `roi_{average|max}_{pooling|align}_2d.py` #5636

knorth55 commented Nov 7, 2018 •

edited

stale bot commented Feb 19, 2019

knorth55 commented Feb 19, 2019

mitmul commented Feb 21, 2019

knorth55 commented Feb 21, 2019

mitmul commented Feb 28, 2019

chainer-ci commented Feb 28, 2019

chainer-ci commented Apr 9, 2019

knorth55 commented Apr 9, 2019

knorth55 commented Apr 24, 2019

Hakuyume commented May 7, 2019

knorth55 commented May 7, 2019

chainer-ci commented Jun 11, 2019

mitmul commented Jul 16, 2019

chainer-ci commented Jul 16, 2019

mitmul commented Jul 17, 2019

knorth55 commented Jul 17, 2019

chainer-ci commented Jul 20, 2019

mitmul commented Jul 21, 2019

chainer-ci commented Jul 21, 2019

knorth55 commented Jul 22, 2019 •

edited

chainer-ci commented Jul 26, 2019

chainer-ci commented Jul 29, 2019

chainer-ci commented Aug 1, 2019

chainer-ci commented Aug 4, 2019

toslunar commented Aug 5, 2019

chainer-ci commented Aug 5, 2019

knorth55 commented Aug 6, 2019

chainer-ci commented Aug 9, 2019

toslunar commented Aug 11, 2019

chainer-ci commented Aug 11, 2019

chainer-ci commented Aug 22, 2019

Use numbers for input check in roi_{average|max}_{pooling|align}_2d.py #5636

Use numbers for input check in roi_{average|max}_{pooling|align}_2d.py #5636

Conversation

knorth55 commented Nov 7, 2018 • edited

stale bot commented Feb 19, 2019

knorth55 commented Feb 19, 2019

mitmul commented Feb 21, 2019

knorth55 commented Feb 21, 2019

mitmul commented Feb 28, 2019

chainer-ci commented Feb 28, 2019

chainer-ci commented Apr 9, 2019

knorth55 commented Apr 9, 2019

knorth55 commented Apr 24, 2019

Hakuyume commented May 7, 2019

knorth55 commented May 7, 2019

chainer-ci commented Jun 11, 2019

mitmul commented Jul 16, 2019

chainer-ci commented Jul 16, 2019

mitmul commented Jul 17, 2019

knorth55 commented Jul 17, 2019

chainer-ci commented Jul 20, 2019

mitmul commented Jul 21, 2019

chainer-ci commented Jul 21, 2019

knorth55 commented Jul 22, 2019 • edited

chainer-ci commented Jul 26, 2019

chainer-ci commented Jul 29, 2019

chainer-ci commented Aug 1, 2019

chainer-ci commented Aug 4, 2019

toslunar commented Aug 5, 2019

chainer-ci commented Aug 5, 2019

knorth55 commented Aug 6, 2019

chainer-ci commented Aug 9, 2019

toslunar commented Aug 11, 2019

chainer-ci commented Aug 11, 2019

chainer-ci commented Aug 22, 2019

Use numbers for input check in `roi_{average|max}_{pooling|align}_2d.py` #5636

Use numbers for input check in `roi_{average|max}_{pooling|align}_2d.py` #5636

knorth55 commented Nov 7, 2018 •

edited

knorth55 commented Jul 22, 2019 •

edited