Library / SDK
Retrieval & Memory
sentence-transformers
1State-of-the-art 2text 3embeddings 4and 5semantic 6search
the six Ws · specification
W1
Who
Created at UKPLab (TU Darmstadt), now maintained with Hugging Face.
W2
What
A Python library computing dense embeddings of text and images for search and similarity.
W3
Where
A library on CPU or GPU, self-hosted in your application.
W4
When
Reach for it when you need embeddings for retrieval, clustering, or RAG.
W5
Why
Easy access to many high-quality embedding and reranker models with a simple API.
W6
With
PyTorch and Transformers; downloads embedding model weights; GPU optional.
W7
Watch
Originated at UKPLab; open-source (Apache-2.0); runs fully locally and offline once models are cached.
works with