Model Card
A standardized documentation format for ML models covering intended use, performance metrics by demographic group, and limitations.
Model cards are structured documentation artifacts for machine learning models, introduced by Google researchers in 2018. The format includes four main sections: model details (architecture, training data, version), intended use (primary use cases and out-of-scope applications), performance metrics broken down by demographic group, and training data provenance. Currently, no mandatory requirements or established standards exist for model cards. They serve as transparency mechanisms for researchers comparing models, policymakers assessing impact, and downstream organizations making adoption decisions.
Also known as
ML model card, model documentation