Conversation
Summary of ChangesHello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request enhances the Highlights
🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console. Changelog
Activity
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request introduces keras.ops.fmod, a valuable addition for consistency with NumPy's API. The implementation across different backends is well-structured, and the accompanying tests are thorough. I have a couple of suggestions to enhance the PR: one is to include a code example in the fmod docstring as per the Keras API design guidelines, and the other is to address a correctness issue in the TensorFlow backend implementation to ensure it correctly handles integer dtypes.
Note: Security Review did not run due to the size of the PR.
| quotient = x1 / x2 | ||
| truncated = tf.sign(quotient) * tf.math.floor(tf.math.abs(quotient)) | ||
| return x1 - truncated * x2 |
There was a problem hiding this comment.
The current implementation can be simplified by using tf.trunc. More importantly, it returns a float tensor for integer inputs, which is inconsistent with other backends and numpy.fmod. The result should be cast back to the original integer dtype if the inputs were integers.
| quotient = x1 / x2 | |
| truncated = tf.sign(quotient) * tf.math.floor(tf.math.abs(quotient)) | |
| return x1 - truncated * x2 | |
| quotient = x1 / x2 | |
| result = x1 - tf.trunc(quotient) * x2 | |
| if "int" in dtype: | |
| return tf.cast(result, dtype) | |
| return result |
| Returns: | ||
| Output tensor, element-wise remainder with truncation. | ||
| """ |
There was a problem hiding this comment.
Per the Keras API design guidelines, all docstrings should include code examples. Adding an example here would help users understand the function's behavior, especially its difference from keras.ops.mod.
Returns:
Output tensor, element-wise remainder with truncation.
Examples:
>>> x1 = keras.ops.array([-3., -2., -1., 1., 2., 3.])
>>> x2 = keras.ops.array([2., 2., 2., 2., 2., 2.])
>>> keras.ops.fmod(x1, x2)
array([-1., -0., -1., 1., 0., 1.], dtype=float32)
>>> x1 = keras.ops.array([1, 2, 3, 4, 5])
>>> x2 = keras.ops.array([-2, -2, -2, -2, -2])
>>> keras.ops.fmod(x1, x2)
array([1, 0, 1, 0, 1], dtype=int32)
"""References
- All docstrings should include code examples. (link)
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## master #22369 +/- ##
==========================================
+ Coverage 82.93% 82.96% +0.02%
==========================================
Files 595 596 +1
Lines 66083 66323 +240
Branches 10312 10326 +14
==========================================
+ Hits 54808 55025 +217
- Misses 8659 8674 +15
- Partials 2616 2624 +8
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
hertschuh
left a comment
hertschuh
left a comment
There was a problem hiding this comment.
Thanks for adding this!
| quotient = x1 / x2 | ||
| truncated = tf.sign(quotient) * tf.math.floor(tf.math.abs(quotient)) | ||
| truncated = tf.cast(truncated, dtype) | ||
| return x1 - truncated * x2 |
There was a problem hiding this comment.
I think this would be a faster implementation:
return tf.sign(x1) * tf.math.floormod(tf.abs(x1), tf.abs(x2))
4 participants
Add
keras.ops.fmod— the C-style remainder of division (truncation toward zero).Unlike
keras.ops.modwhich returns a remainder with the same sign as the divisor,fmodreturns a remainder with the same sign as the dividend, matching the behavior of C'sfmodandnumpy.fmod.Implements across all backends (numpy, jax, tensorflow, torch). OpenVINO raises
NotImplementedError.Includes dynamic shape, static shape, correctness, and dtype tests.