experiment-audit

What problem does it solve?

This Skill rigorously validates Machine Learning experiments, ensuring correctness, reproducibility, and integrity, thereby preventing costly errors and building trust in results.

Core Features & Use Cases

• Comprehensive Validation: Audits data integrity, pipeline execution, metric correctness, and reproducibility.
• Leakage Detection: Specifically checks for and identifies data leakage across splits and through preprocessing.
• Use Case: A research team has just completed a complex ML experiment. They need to ensure their findings are robust and reproducible before publication. This Skill will systematically review their experiment setup, data handling, and results to provide a confidence score and identify any potential issues.

Quick Start

Run a full audit on the experiment located in the '/path/to/experiment/results' directory.