Conversation
jzakirov
added
the
refactoring
|
@zakajd are you still working on this one? |
|
@snk4tr Yeah, I'm waiting for the documentation PR to be merged so that I can update it with -o flag usage example. |
|
@snk4tr This one is ready for merge. |
Codecov Report
@@ Coverage Diff @@
## master #234 +/- ##
==========================================
+ Coverage 93.75% 95.18% +1.43%
==========================================
Files 28 28
Lines 1826 1766 -60
==========================================
- Hits 1712 1681 -31
+ Misses 114 85 -29
Flags with carried forward coverage won't be shown. Click here to find out more.
|
snk4tr
left a comment
snk4tr
left a comment
There was a problem hiding this comment.
Nice one, @zakajd.
Let me add a few tiny fixes and merge.
| Args: | ||
| kernel_size: By default, the mean and covariance of a pixel is obtained | ||
| by convolution with given filter_size. | ||
| by convolution with given filter_size. Must be an odd value. |
There was a problem hiding this comment.
We require kernel_size to be odd but never check that even in the debug mode. Add will add an assertion for that.
README.rst
Outdated
| Assertions | ||
| ---------- | ||
| In PIQ we use assertions to raise meaningful messages when some component doesn't receive an input of the expected type. | ||
| This makes prototyping and debugging easier, but it might hurt a little the performance. |
There was a problem hiding this comment.
Grammatically incorrect. I will fix that in the commit later.
| This makes prototyping and debugging easier, but it might hurt a little the performance. | |
| This makes prototyping and debugging easier, but it might hurt the performance. |
| y: A target tensor. Shape :math:`(N, C, H, W)`. | ||
| reduction: Specifies the reduction type: | ||
| ``'none'`` | ``'mean'`` | ``'sum'``. Default:``'mean'`` | ||
| data_range: Maximum value range of images (usually 1.0 or 255). |
There was a problem hiding this comment.
Good that you removed the note about default here. It duplicates info and adds nothing besides potential confusion
| Args: | ||
| kernel_size: By default, the mean and covariance of a pixel is obtained | ||
| by convolution with given filter_size. | ||
| by convolution with given filter_size. Must be an odd value. |
Signed-off-by: Sergey Kastryulin <snk4tr@gmail.com>
|
Kudos, SonarCloud Quality Gate passed!
|
2 participants
Closes #212
Another semantic update =)
Proposed Changes
_reductionfunction to unify reduction process._validate_featuresand_validate_inputinto one function and simplify it's interfaceWith this PR we make requirements on input tensors more strict.
Drop support for 2D and 3D tensors
Drop support for scale weights as lists
Add section to README with description of using
-oflag after Documentation: Include README.md #232 is merged.