Ben Strawbridge DOT COM CONSULTING

Testing

A Theme Meets a Stranger

I clicked the “Latest release” badge in Ryder’s README and landed on v0.0.1-alpha, dated January 2024. That would have been unremarkable if it were the latest release. It wasn’t. Ryder had shipped v0.1.0 through v0.2.4, but the release workflow marked every automated release as a prerelease. GitHub excludes prereleases from that badge, so two years of releases were present, downloadable, and effectively invisible from the project’s front door.

Deepchecks is an open-source platform for validating machine learning data and models, from tabular data to NLP and computer vision. It groups individual validation checks into suites covering data integrity, distribution drift, and model performance.

Great Expectations is an open-source Python framework for validating, profiling, and documenting data quality. Assertions about a dataset, called Expectations, are expressed as human-readable Python methods and run against data sources ranging from CSV files to databases.

TensorFlow's guide to TensorFlow Model Analysis (TFMA), a library for evaluating TensorFlow models at scale within a TFX pipeline. It computes metrics over data slices and can visualize the results in a Jupyter notebook.