model-card

Overview

Based on "Interpretable Machine Learning" by Christoph Molnar, with structure informed by the Google Model Cards framework (Mitchell et al., 2019). The core principle: a model without documentation is a liability. Interpretability is not a post-hoc courtesy - it is a prerequisite for responsible deployment. A model card makes the model's intended use, performance limits, and failure modes explicit before the model reaches production.

Workflow

Step 1: Write the model overview section

## Model Overview

**Model name:** [descriptive name, not a variable name]
**Version:** [semantic version, e.g., v1.2.0]
**Model type:** [e.g., XGBoost binary classifier, LSTM time-series forecaster]
**Owner:** [team or individual responsible for this model]
**Last updated:** [date]