Notebook Quick Start

This guide walks through using Kongming HV in a Jupyter notebook, cell by cell.

Section	Description
Notebook Platforms	Setup differences between Jupyter, Colab, and Binder
Interactive Notebooks	Links to existing notebooks
Walkthrough	Step-by-step: vocabulary, similarity, learning, binding

Tips

Reproducibility: Use fixed seeds in SparseOperation for deterministic results across reruns.
Visualization: Use pandas DataFrames for overlap matrices — they render nicely in Jupyter.
Performance: The Rust backend is fast. Building 10,000 vectors takes under a second on MODEL_64K_8BIT.
Model choice: Start with MODEL_64K_8BIT for exploration. Switch to MODEL_1M_10BIT or larger for production workloads.