1,088 stars, zero hype: a Spark NLP cookbook that just works

What it does This is the official companion kitchen for John Snow Labs’ Spark NLP library: Jupyter notebooks, Colab-ready tutorials, and Databricks notebooks covering annotation pipelines, model training, and evaluation. The setup instructions are refreshingly explicit — Java 8, PySpark 3.1.2, pip install, done. There’s even a one-liner shell script for Colab that downloads and configures everything behind the scenes.

The interesting bit The “old_generation_notebooks” folder in tutorials suggests this repo has been through enough iterations to accumulate historical baggage, yet someone is still maintaining backward compatibility. That’s either admirable diligence or a warning about API churn — the README doesn’t clarify which.

Key highlights

• Python and Scala examples side by side (rare in notebook-land)
• Dedicated Databricks folder for enterprise Spark deployments
• One-shot Colab setup via wget | bash — convenient, if you trust it
• Explicit dependency pinning (PySpark 3.1.2) rather than “latest and hope”
• Apache 2.0 licensed, with a Slack community linked for support

Caveats

• The “evalulation” typo in the table of contents has survived at least one README revision
• No topics tagged on GitHub, making discovery harder than it should be
• “Old generation” notebooks are still prominently linked; unclear if they’re deprecated or merely archived

Verdict Grab this if you’re already committed to Spark NLP and need working starter code. Skip it if you’re looking for a general NLP tutorial — the Spark dependency and JVM tooling make this a niche affair.