Skip some Convolution2D tests for older numpy versions
#8458
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
I suggest avoid skipping but instead omit numerical checks (#7946 (comment)). |
emcastillo
force-pushed
the
fix-conv2-numpy
branch
from
83490a9
to
06d4e9e
Compare
|
Comment addressed, thanks a lot for the feedback. |
toslunar
reviewed
Nov 19, 2019
tests/chainer_tests/functions_tests/connection_tests/test_convolution_2d.py
Outdated
Show resolved
Hide resolved
emcastillo
force-pushed
the
fix-conv2-numpy
branch
from
06d4e9e
to
2bf1b9e
Compare
takagi
requested changes
Nov 21, 2019
tests/chainer_tests/functions_tests/connection_tests/test_convolution_2d.py
Show resolved
Hide resolved
tests/chainer_tests/functions_tests/connection_tests/test_convolution_2d.py
Show resolved
Hide resolved
|
pfnCI, test this please. |
|
Jenkins CI test (for commit 2bf1b9e, target branch master) succeeded! |
takagi
added
cat:test
|
LGTM! The CI failure on Travis is not related to this change. |
takagi
added a commit
to takagi/chainer
that referenced
this pull request
Nov 23, 2019
Skip some `Convolution2D` tests for older numpy versions
Merged
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Closes #7946
The reason behind the flakiness is the use of a numpy version < 1.17
1.17 fixed a bug in the float16 implementation.
What happens is that forward_expected is comparing all the backends against the cpu implementation using numpy. And when using CUDA the results in float16 might sightly differ in some corner cases dut to the cpu-gpu difference (numpy bug) in fp16 implementation. Other tests that rely on numpy for forward_expected are most likely to be affected by this.
This failure only happens with grouped convolutions.