Starstruk Logo MarkStarstruk

v0.5.0

MinishLab/model2vecv0.5.0Apr 30, 2025by stephantul

AI Summary

This release introduces quantization capabilities and faster inference for large vocabularies. It adds dimensionality configuration during loading and token provenance tracking.

Key Highlights

  • Added quantization support
  • Faster inference for large vocabularies
  • Token provenance tracking
  • Dimensionality configuration during loading
  • Added py.typed file for type hints

New Features

  • Quantization support
  • Faster inference for large vocab
  • Token provenance tracking
  • Dimensionality during loading
  • Save/load subfolder support
  • Quantization for from_sentence_transformers

Full Release Notes

## What's Changed
* fix: Updated semantic chunking tutorial by @bhavnicksm in https://github.com/MinishLab/model2vec/pull/205
* rewrite backend by @stephantul in https://github.com/MinishLab/model2vec/pull/207
* fix bibtex by @stephantul in https://github.com/MinishLab/model2vec/pull/208
* feat: Added py.typed file by @Pringled in https://github.com/MinishLab/model2vec/pull/214
* fix: pretokenize tokens before checking vocabulary by @stephantul in https://github.com/MinishLab/model2vec/pull/215
* feat: add dimensionality during loading by @stephantul in https://github.com/MinishLab/model2vec/pull/216
* feat: add quantization by @stephantul in https://github.com/MinishLab/model2vec/pull/217
* feat: save load subfolder by @stephantul in https://github.com/MinishLab/model2vec/pull/218
* feat: Added quantization for from_sentence_transformers by @Pringled in https://github.com/MinishLab/model2vec/pull/219
* feat: faster inference for large vocab by @stephantul in https://github.com/MinishLab/model2vec/pull/221
* feat: track token provenance by @stephantul in https://github.com/MinishLab/model2vec/pull/222
* fix: typing issues, bug in infernece by @stephantul in https://github.com/MinishLab/model2vec/pull/224
* fix: issues with unk and pad by @stephantul in https://github.com/MinishLab/model2vec/pull/225
* bug: fix 0 score in evaluate by @stephantul in https://github.com/MinishLab/model2vec/pull/226
* fix: precision during training by @stephantul in https://github.com/MinishLab/model2vec/pull/228
* fix: issue with unk in unigram by @stephantul in https://github.com/MinishLab/model2vec/pull/227
* docs: add info about quantization and dimensionality reduction by @stephantul in https://github.com/MinishLab/model2vec/pull/231
* increment version by @stephantul in https://github.com/MinishLab/model2vec/pull/232

## New Contributors
* @bhavnicksm made their first contribution in https://github.com/MinishLab/model2vec/pull/205

**Full Changelog**: https://github.com/MinishLab/model2vec/compare/0.4.1...v0.5.0