Try out NBTest: A smarter tool for testing jupyter notebooks

[ElaineYao]

Hi pydra developers,

We’re researchers at Cornell University focused on quality assurance for machine learning workflows (our group). We noticed you’ve previously used nbval, a pytest plugin for validating Jupyter notebooks in CI pipelines.

We’d love to introduce you to NBTest, a new tool we’ve developed that builds on nbval’s core ideas—but goes significantly further for ML-specific use cases.

While nbval checks for output consistency, NBTest automatically detects key machine learning metrics (like dataset statistics and model performance) and generates statistically sound assertions. This helps catch subtle regressions and inconsistencies that traditional tools may miss.

Getting started is simple—you can follow our installation and usage instructions and start integrating it into your notebooks in minutes.

We’d be excited if you gave it a try, and we’d love to hear your thoughts. If you're interested, we’d also be happy to set up a quick chat to talk more!

Best regards,
Elaine

[effigies]

Moved to pydra-ml, as this repository is intended to provide ML tasks and workflows.

[satra]

thank you @ElaineYao - we will probably update this repo to work with the latest release of pydra at an upcoming hackathon at the end of june. we will take a look at this evaluation component then. it will also likely be of interest in other projects.