Croissant Baker: Metadata Generation for Discoverable, Governable, and Reusable ML Datasets

TL;DR

Croissant Baker is an open-source tool that generates validated Croissant metadata locally from datasets, enabling better discovery, governance, and reuse of ML datasets, especially for large or private repositories.

Contribution

It introduces Croissant Baker, a local-first command-line tool that automates Croissant metadata generation with high accuracy for diverse datasets.

Findings

Achieved 97-100% agreement with ground truth on multiple datasets.

Successfully scaled to MIMIC-IV with 886 million rows and 374 Parquet files.

Supports over 140 datasets with validated metadata generation.

Abstract

Croissant has emerged as the metadata standard for machine learning datasets, providing a structured, JSON-LD-based format that makes dataset discovery, automated ingestion, and reproducible analysis machine-checkable across ML platforms. Adoption has accelerated, and NeurIPS now requires Croissant metadata in every submission to its dataset tracks. Yet in practice Croissant generation usually starts with uploading data to a public platform, a path infeasible for governed and large local repositories that hold much of the high-value data ML increasingly relies on. We release Croissant Baker, a local-first, open-source command-line tool that generates validated Croissant metadata directly from a dataset directory through a modular handler registry. We evaluate Croissant Baker on over 140 datasets, scaling to MIMIC-IV at 886 million rows and 374 Parquet files. On held-out comparisons against producer-authored or standards-derived ground truth, Croissant Baker reaches 97-100% agreement across multiple domains.