Facilitates the creation and debugging of robust tests for BayesFlow extension packages using Pytest and Keras.
The bayesflow-testing skill provides specialized guidance for writing, modifying, and debugging test suites within the BayesFlow ecosystem. It enforces critical setup requirements like the correct Keras backend initialization, establishes standardized test structures, and offers implementation patterns for shape verification, configuration roundtrips, and mock-based testing. This ensures that machine learning models and simulators are tested for deterministic behavior, invariance properties, and integration accuracy without the overhead of real training runs.
主な機能
01Mock-based testing for fast inference network validation
020 GitHub stars
03Automated Keras backend configuration for Torch in conftest.py
04RNG seeding strategies for deterministic and reproducible tests
05Configuration serialization and roundtrip verification
06Invariance and equivariance testing patterns for IRT architectures
ユースケース
01Writing unit tests for new BayesFlow simulators and adapters
02Debugging backend initialization issues in Pytest suites
03Setting up GitHub Actions CI/CD pipelines for BayesFlow extensions
