# epfml/sent2vec

General purpose unsupervised sentence representations

Repository: https://github.com/epfml/sent2vec
Canonical: https://ross.abutalabs.com/products/sent2vec
Language: C++
License: NOASSERTION
License Family: other
Last push: 2022-08-03T18:18:45+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 3458, "days_push": 1491, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1201, forks 257 (observed 2026-08-28T04:03:58.164750+00:00)

## What it is
Sent2vec is a C++ library with a Cython/Python interface that trains unsupervised distributed representations of sentences and short texts, built on Facebook's FastText. It includes pre-trained models for generating sentence and word embeddings usable as features in downstream machine learning tasks.

## Use cases
- generate sentence embeddings for text classification
- get numerical features for short texts and sentences
- train an unsupervised sentence embedding model
- find nearest neighbour sentences and word analogies
- extract word embeddings from pre-trained models
- embed sentences in python while keeping the model in memory

## When to choose
- you need fast, lightweight unsupervised sentence embeddings without a GPU
- you want to train custom embeddings on your own corpus
- you need embeddings as input features for classical ML models

## When to avoid
- you need state-of-the-art contextual embeddings from transformer models
- you need cross-lingual sentence embeddings (use Bi-sent2vec instead)
- you need an actively developed project with recent updates

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, nlp
- domain: machine-learning
- platform: python, cpp, cross-platform
- tags: sentence-embeddings, word-embeddings, unsupervised-learning, fasttext, cython, natural-language-processing

## Member repositories
- epfml/sent2vec (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:58.164750+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:20:40.853433+00:00, confidence not recorded.
  - readme: https://github.com/epfml/sent2vec (fetched 2026-08-28T04:03:58.164750+00:00, sha 0b600416d032)
- Data as of 2026-08-30T08:39:29.467469+00:00.