Intermediate · Research

Reproducibility

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. The ability for others to re-run an experiment and obtain the same results.

Technical Definition

The ability for others to re-run an experiment and obtain the same results.

How it works

Requires sharing code, data, environment, random seeds, and exact hyperparameters. ML's reproducibility crisis spurred tools like containers, lockfiles, dataset versioning, and model registries. Without it, scientific claims and production debugging both fall apart.

Related Concepts

• Data Versioning — Tracking changes to datasets the way Git tracks changes to code.
• Containerization — Packaging an application with all its dependencies so it runs consistently across environments.
• Experiment Tracking — Systematically recording the inputs, code, parameters, and outputs of every ML experiment.