Increase test coverage for IntegerLookup layer

[MarcosAsh]

Description

Addresses TODO in integer_lookup_test.py to increase test coverage.

Changes

Added tests for:

• max_tokens - vocabulary truncation
• mask_token - masking behavior
• invert - inverse lookup
• pad_to_max_tokens - output padding
• num_oov_indices - multiple OOV buckets
• num_oov_indices=0 - no OOV handling
• get_vocabulary() - vocabulary retrieval after adapt
• adapt() with steps parameter
• sparse=True - sparse tensor output
• vocabulary from file path
• Invalid argument validation (max_tokens, num_oov_indices, vocabulary_dtype)

[gemini-code-assist]

Summary of Changes

Hello @MarcosAsh, 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 significantly enhances the test coverage for the IntegerLookup preprocessing layer within Keras. By introducing a comprehensive suite of new tests, it addresses a previous TODO item, ensuring the layer's various functionalities, edge cases, and argument validations behave as expected. This improvement contributes to the overall reliability and robustness of this critical data preprocessing component.

Highlights

• Vocabulary Truncation: Added tests to verify the max_tokens parameter correctly truncates the vocabulary.
• Masking Behavior: Included tests to ensure the mask_token parameter functions as expected for masking values.
• Inverse Lookup: Implemented tests for the invert functionality, allowing inverse mapping of integer indices back to original values.
• Output Padding: Added tests to confirm pad_to_max_tokens correctly pads the output to the specified maximum token length.
• Out-Of-Vocabulary (OOV) Handling: Covered scenarios for num_oov_indices, including multiple OOV buckets and the case where num_oov_indices=0 (no OOV handling).
• Vocabulary Retrieval: Verified the get_vocabulary() method retrieves the correct vocabulary after the layer has been adapted.
• Adaptation with Steps: Added tests for the adapt() method when used with the steps parameter for controlled adaptation.
• Sparse Tensor Output: Included tests to confirm the layer can produce sparse tensor output when sparse=True.
• Vocabulary from File: Tested the ability to initialize the vocabulary from a file path.
• Invalid Argument Validation: Added validation tests for incorrect max_tokens, num_oov_indices, and vocabulary_dtype parameters, ensuring appropriate ValueError exceptions are raised.

🧠 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.

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 does a great job of increasing the test coverage for the IntegerLookup layer, addressing a previous TODO. The new tests cover a wide range of features and edge cases. I have a few suggestions to further improve the quality and style of the new tests.

[gemini-code-assist]

For better test clarity and robustness, please use the self.assertSparse(output) helper method from the base test case instead of self.assertTrue(hasattr(output, 'indices')). assertSparse provides more specific, backend-aware checks for sparse tensors.

Suggested change

	output = layer([1, 2])
	self.assertTrue(hasattr(output, 'indices')) # SparseTensor check
	output = layer([1, 2])
	self.assertSparse(output)

[gemini-code-assist]

This test correctly checks the behavior for in-vocabulary tokens when num_oov_indices=0. However, it's missing a check for the documented behavior that an out-of-vocabulary (OOV) token should raise an error in this configuration. Please add an assertion to verify this behavior.

Suggested change

	def test_num_oov_indices_zero(self):
	layer = layers.IntegerLookup(
	vocabulary=[1, 2, 3],
	num_oov_indices=0,
	output_mode="int",
	)
	output = layer([1, 2, 3])
	self.assertAllClose(output, np.array([0, 1, 2]))
	def test_num_oov_indices_zero(self):
	layer = layers.IntegerLookup(
	vocabulary=[1, 2, 3],
	num_oov_indices=0,
	output_mode="int",
	)
	output = layer([1, 2, 3])
	self.assertAllClose(output, np.array([0, 1, 2]))
	from keras.src.utils.module_utils import tensorflow as tf
	with self.assertRaises(tf.errors.InvalidArgumentError):
	layer([1, 2, 3, 4])

[gemini-code-assist]

This test can be improved for better style and consistency with the project's testing practices:

1. Local imports of tempfile and os should be moved to the top of the file.
2. The tmp_path=None argument is unused and should be removed from the method signature.
3. Instead of tempfile.TemporaryDirectory(), please use self.get_temp_dir() which is provided by the testing.TestCase base class for creating temporary directories.

Suggested change

	def test_vocabulary_from_file(self, tmp_path=None):
	import tempfile
	import os
	with tempfile.TemporaryDirectory() as tmp_dir:
	vocab_file = os.path.join(tmp_dir, "vocab.txt")
	with open(vocab_file, "w") as f:
	f.write("10\n20\n30\n")
	layer = layers.IntegerLookup(
	vocabulary=vocab_file,
	output_mode="int",
	)
	output = layer([10, 20, 30, 999])
	self.assertAllClose(output, np.array([1, 2, 3, 0]))
	def test_vocabulary_from_file(self):
	import os # This import should be at the top of the file.
	tmp_dir = self.get_temp_dir()
	vocab_file = os.path.join(tmp_dir, "vocab.txt")
	with open(vocab_file, "w") as f:
	f.write("10\n20\n30\n")
	layer = layers.IntegerLookup(
	vocabulary=vocab_file,
	output_mode="int",
	)
	output = layer([10, 20, 30, 999])
	self.assertAllClose(output, np.array([1, 2, 3, 0]))

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.78%. Comparing base (7dda29f) to head (77a5f18).

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #22022      +/-   ##
==========================================
+ Coverage   82.75%   82.78%   +0.02%     
==========================================
  Files         592      592              
  Lines       62172    62172              
  Branches     9738     9738              
==========================================
+ Hits        51450    51466      +16     
+ Misses       8196     8186      -10     
+ Partials     2526     2520       -6     

Flag	Coverage Δ
keras	82.60% <ø> (+0.02%)	⬆️
keras-jax	62.43% <ø> (+0.02%)	⬆️
keras-numpy	56.56% <ø> (+0.04%)	⬆️
keras-openvino	37.48% <ø> (ø)
keras-tensorflow	63.67% <ø> (+0.02%)	⬆️
keras-torch	62.44% <ø> (+0.02%)	⬆️

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.

[hertschuh]

Thank you for adding all these!

[hertschuh]

This line is failing on GPU with Torch. I think you need to change it to:

self.assertTrue(all(o in [0, 1] for o in backend.convert_to_numpy(output[3:])))

Unit test is failing

[MarcosAsh]

Hi, thanks for the feedback. Just pushed the changes

[hertschuh]

This line is failing on GPU with Torch. I think you need to change it to:

self.assertTrue(all(o in [0, 1] for o in backend.convert_to_numpy(output[3:])))

[hertschuh]

Now the code formatter is complaining about this line, which is too long.

      self.assertTrue(
          all(o in [0, 1] for o in backend.convert_to_numpy(output[3:]))
      )

You can always run ruff --config pyproject.toml format . locally.