Changelog

This document records all notable changes to EVoC.

Version 0.1.0 (TBD)

Initial release of EVoC.

Features:

  • Core clustering algorithm with hierarchical multi-layer support

  • Scikit-learn compatible API

  • Support for multiple embedding types (float, int8, uint8)

  • Optimized distance metrics (cosine, quantized cosine, bitwise Jaccard)

  • Numba-accelerated performance

  • Comprehensive parameter set for fine-tuning

  • Built-in duplicate detection

  • Extensive documentation and examples

API Reference:

  • EVoC - Main clustering class

  • evoc_clusters - Functional interface

  • build_cluster_layers - Multi-layer clustering construction

Performance:

  • Efficient processing of high-dimensional embeddings

  • Memory-optimized algorithms

  • Multi-threaded computation support

Documentation:

  • Complete API documentation with numpydoc formatting

  • Interactive Jupyter notebook examples

  • Comprehensive user guide and tutorials

  • ReadTheDocs integration