MinishLab/tokenlearn

Pre-train Static Word Embeddings

40
/ 100
Emerging

Implements a two-stage pipeline extracting mean token embeddings from sentence transformers (featurize step) then training lightweight Model2Vec static embeddings against those targets. Provides CLI tools to process HuggingFace datasets end-to-end and integrates with MTEB evaluation framework for downstream task assessment. Powers the Potion model family with configurable support for multilingual and multi-scale embeddings.

No commits in the last 6 months. Available on PyPI.

Stale 6m
Maintenance 2 / 25
Adoption 9 / 25
Maturity 18 / 25
Community 11 / 25

How are scores calculated?

Stars

94

Forks

8

Language

Python

License

MIT

Last pushed

Sep 09, 2025

Commits (30d)

0

Dependencies

5

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/embeddings/MinishLab/tokenlearn"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.