sentence-transformers

Sentence/image embeddings, retrieval, and reranking via Transformers

Sentence-Transformers (SBERT) provides sentence, image, and code embeddings via Transformer networks for semantic search, clustering, semantic textual similarity, and retrieval-augmented generation. Bundled training utilities support fine-tuning on contrastive losses; bundled inference utilities cover bi-encoders, cross- encoders (rerankers), and sparse encoders. Embedding-based RAG pipelines downstream — including sci-ml/amd-gaia's ui module and most LLM-orchestration frameworks — depend on this library.

homepage ↗homepage ↗homepage ↗ github: huggingface/sentence-transformers pypi: sentence-transformers