Make TableConfig and FeatureConfig unhashable.

[hertschuh]

TableConfig contains a keras Optimizer and a keras Initializer, which are not hashable and even mutable in the case of the optimizer. The "unsafe hash" was actually unsafe. Instead of using TableConfigs as keys directly, we use the id of the TableConfig, which is correct because we are detecting reused instances.

Also:

• renamed TensorFlow config conversion functions to be more consistent with the JAX ones
• added unit tests for the TensorFlow config conversion functions
• fix optimizer conversion with the Torch backend

[gemini-code-assist]

Summary of Changes

Hello @hertschuh, 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 resolves a critical issue where TableConfig and FeatureConfig objects, which encapsulate unhashable Keras Optimizer and Initializer instances, were being used as dictionary keys due to an unsafe_hash flag. The solution involves making these dataclasses unhashable and adapting internal data structures to use Python's id() function for object identity when these configurations are used in mappings. Additionally, the pull request includes a significant refactoring of TensorFlow configuration conversion functions to align their naming with JAX conventions and introduces new unit tests to validate these conversions, enhancing the robustness and clarity of the codebase.

Highlights

• Unhashable Configs: TableConfig and FeatureConfig dataclasses are now explicitly unhashable by removing unsafe_hash=True, addressing issues with mutable Keras objects they contain.
• Object Identity for Mappings: Internal mappings that previously used TableConfig objects as keys now correctly use their unique id() to ensure proper object identity tracking and avoid hashing errors.
• TensorFlow Config Conversion Refactoring: TensorFlow configuration conversion functions have been renamed for better consistency with their JAX counterparts, improving code readability and maintainability.
• New Unit Tests: Comprehensive unit tests have been added for the TensorFlow config conversion logic, covering optimizer, table, and feature config translations to ensure correctness.

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 correctly addresses the issue of using mutable TableConfig and FeatureConfig objects as dictionary keys by making them unhashable and using their id() for identity checks instead. The changes are consistently applied across the codebase. The renaming of TensorFlow configuration conversion functions improves clarity and consistency with the JAX counterparts. Furthermore, the addition of unit tests for these conversion functions is a great enhancement to the test suite. I've identified one potential issue in the JAX implementation where non-unique table names could lead to incorrect behavior and have provided a suggestion for a more robust solution. Overall, this is a strong pull request that improves the correctness and maintainability of the code.