This directory contains tests for the Cross-Layer Transcoder (CLT) library. The testing strategy is divided into two main categories:

Located in tests/unit/, these tests verify that individual components of the library work correctly in isolation. Unit tests are designed to:

• Test specific functionality of a single class or function
• Mock dependencies to isolate the component being tested
• Run quickly and provide good coverage
• Help identify where issues originate

Located in tests/integration/, these tests verify that multiple components work together correctly. Integration tests are designed to:

• Test realistic user workflows
• Verify that components interact properly
• Use real (but small-scale) components instead of mocks
• Test data flow between components

To run all tests:

pytest

To run unit tests only:

pytest tests/unit

To run integration tests only:

pytest tests/integration

The tests/integration/data/ directory contains fixtures for integration tests, including:

• Sample activation data
• Pre-trained model files
• Helper scripts to generate test data

These fixtures enable deterministic testing of model loading, inference, and training without requiring external data.

When adding new functionality to the library:

1. Add unit tests for the new component in tests/unit/
2. Add integration tests in tests/integration/ for any new workflows

Use the @pytest.mark.integration marker for integration tests:

@pytest.mark.integration
def test_my_integration_test():
    # Test code here
    pass