What testing framework does this skill use?

This skill is designed to work with Pytest, mirroring the source layout and utilizing shared fixtures for maximum efficiency.

Does this skill help with slow ML tests?

Yes, it provides patterns for mock-based testing that allow you to verify network shapes and adapter logic without running real, time-consuming training sessions.

How does it handle permutation invariance in neural networks?

The skill includes specific IRT (Invariance/Equivariance) testing patterns that verify if permuting input data correctly affects (or doesn't affect) the output as intended.

Why is the conftest.py setup so critical for BayesFlow?

Keras 3 reads the KERAS_BACKEND environment variable only once during the first import. This skill ensures the backend is set before any imports occur, preventing accidental defaults to TensorFlow.

BayesFlow Testing Patterns

Name: BayesFlow Testing Patterns
Author: matthiaskloft

bymatthiaskloft

0•

Security & Testing

Standardizes and automates the creation of robust test suites for BayesFlow extension packages using Pytest and Keras 3.

The BayesFlow Testing skill provides specialized domain knowledge for developing high-quality tests for simulation-based inference frameworks. It enforces critical architectural patterns, such as the mandatory pre-import configuration of the Keras backend to prevent initialization failures. The skill guides developers through implementing shape verification for simulators, config roundtrip validation, and sophisticated invariance/equivariance testing for neural networks. By utilizing mock-based testing strategies included in this skill, developers can verify complex model logic and adapter pipelines without the overhead of full training cycles, leading to faster, more reliable CI/CD workflows.

Key Features

010 GitHub stars

02Automated shape and dtype validation for simulators and adapters

03Critical Keras 3 backend initialization setup via conftest.py

04Invariance and equivariance testing for symmetry-aware architectures

05Deterministic RNG management to eliminate flaky CI results

06Mock-based neural network testing for high-speed verification

Use Cases

01Setting up a new BayesFlow extension repository with standardized test structures.

02Debugging backend initialization errors and non-deterministic test failures in ML pipelines.

03Implementing unit tests for custom summary networks with specific permutation requirements.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add matthiaskloft/claude-skills bayesflow-testing

For use in Claude.ai and ChatGPT

Key Features

010 GitHub stars

02Automated shape and dtype validation for simulators and adapters

03Critical Keras 3 backend initialization setup via conftest.py

04Invariance and equivariance testing for symmetry-aware architectures

05Deterministic RNG management to eliminate flaky CI results

06Mock-based neural network testing for high-speed verification

Use Cases

01Setting up a new BayesFlow extension repository with standardized test structures.

02Debugging backend initialization errors and non-deterministic test failures in ML pipelines.

03Implementing unit tests for custom summary networks with specific permutation requirements.