Fix discretization discrepancy

[danielenricocahall]

Resolved minor bugs in #21514 to address #21468 to fix the discrepancy in discretization

[gemini-code-assist]

Summary of Changes

Hello @danielenricocahall, 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 addresses a minor bug in the Discretization layer by removing an unnecessary dtype coercion. The change simplifies the layer's initialization logic and resolves a reported issue. To ensure the stability and correctness of this fix, a new test has been introduced to verify consistent behavior between direct model calls and predictions for the Discretization layer.

Highlights

• Discretization Layer dtype Handling: The explicit dtype coercion logic within the Discretization layer's __init__ method has been removed. Previously, if dtype was None, it would be coerced to int64 or backend.floatx(). This change aligns with the fix for a minor bug related to dtype discrepancies.
• New Consistency Test: A new test case, test_model_call_vs_predict_consistency, has been added to discretization_test.py. This test ensures that the Discretization layer produces consistent outputs when called directly via model(input) and when using model.predict(input).

Using Gemini Code Assist

The 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 /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

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 .gemini/ folder in the base of the repository. Detailed instructions can be found here.

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

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request aims to fix a dtype discrepancy in the Discretization layer. While the proposed change of removing the dtype coercion logic does resolve the inconsistency, it introduces a new issue where output_mode="int" produces float outputs, which is counter-intuitive. My review provides a suggestion to fix the original discrepancy while preserving the expected integer output type. I've also recommended strengthening the new test case to include dtype checks for better validation.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.65%. Comparing base (47fcb39) to head (ebbb85c).
⚠️ Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #21769      +/-   ##
==========================================
- Coverage   82.69%   82.65%   -0.04%     
==========================================
  Files         573      577       +4     
  Lines       58888    59181     +293     
  Branches     9218     9273      +55     
==========================================
+ Hits        48696    48918     +222     
- Misses       7845     7887      +42     
- Partials     2347     2376      +29     

Flag	Coverage Δ
keras	82.46% <100.00%> (-0.04%)	⬇️
keras-jax	63.30% <100.00%> (+0.06%)	⬆️
keras-numpy	57.54% <100.00%> (-0.18%)	⬇️
keras-openvino	34.34% <50.00%> (-0.07%)	⬇️
keras-tensorflow	64.07% <100.00%> (+0.06%)	⬆️
keras-torch	63.62% <100.00%> (+0.06%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[danielenricocahall]

/gemini review

[gemini-code-assist]

Code Review

This pull request effectively resolves a dtype discrepancy in the Discretization layer, particularly when output_mode is set to "int". The changes correctly introduce an output_dtype property to centralize dtype logic, ensuring consistent behavior between symbolic and eager execution paths. The removal of the manual dtype handling in __init__ and the subsequent updates in compute_output_spec and call are well-implemented. The addition of test_model_call_vs_predict_consistency is a great way to safeguard against future regressions. I've only found one minor issue in the new test file, which is a duplicated assertion.

[fchollet]

LGTM, thanks for the PR!