Sentence Transformers

Sentence Transformers, created by UKP Lab and now maintained by Hugging Face, is a free, open-source Python library for computing embeddings from text, images, audio, or video and calculating similarity between them — the standard toolkit behind most semantic search, similarity, and paraphrase-mining applications built in Python. It supports four model types (Sentence Transformers for embeddings, Cross-Encoders for reranking, Sparse Encoders, and Multi-Vector Encoders), handles multimodal inputs, and gives access to over 25,000 pre-trained models on the Hugging Face Hub — plus training and fine-tuning tools for building custom models when a pre-trained one isn't a good enough fit. Being a developer library rather than an end-user app, it requires Python and machine-learning familiarity to use effectively; it's not something a non-technical user picks up directly. Good fit for a developer building semantic search, RAG retrieval, or similarity-matching features who wants a proven, widely-adopted library rather than building embedding infrastructure from scratch.