Sacred

Overview

Sacred is a Python library for configuring, organizing, logging, and reproducing machine learning experiments. It tracks hyperparameters, source code state, dependencies, and results, storing run details in a MongoDB database for later analysis. Developed at IDSIA, it is designed to help researchers ensure experiment reproducibility.

Best for

Best for
Researchers and engineers who want a lightweight, code-centric experiment tracking system

Use cases

• Track hyperparameters and metrics across training runs
• Reproduce past experiments with consistent configuration and source code
• Organize multiple experiment variants into a searchable database

Notes

Sacred is a Python library for configuring, organizing, logging, and reproducing machine learning experiments. It tracks hyperparameters, source code state, dependencies, and results, storing run details in a MongoDB database for later analysis. Developed at IDSIA, it is designed to help researchers ensure experiment reproducibility.

4,365 stars on GitHub. Last updated 2025-10-22. Licensed MIT.

Use cases

• Track hyperparameters and metrics across training runs
• Reproduce past experiments with consistent configuration and source code
• Organize multiple experiment variants into a searchable database

Pros

• Lightweight and integrates easily with existing Python code
• Automatically logs source code state and dependencies for reproducibility
• Active open source community with ongoing development

Cons

• Requires MongoDB for full functionality, adding infrastructure complexity
• Limited built-in visualization compared to dedicated experiment tracking platforms
• Configuration changes can require code refactoring for deeply nested experiments

Indexed from awesome-llmops and enriched against its public facts.