sentence-transformers

Sentence Transformers is a Python framework for computing dense, sparse, and multi-vector embeddings for texts, as well as for reranking passages. It provides easy-to-use APIs for loading pre-trained models from Hugging Face, encoding sentences, and computing similarity scores. The framework supports four model types: Sentence Transformer (dense embeddings), Cross-Encoder (reranker), Sparse Encoder (SPLADE-style sparse embeddings), and Multi-Vector Encoder (ColBERT-style late interaction). It includes over 15,000 pre-trained models, supports training and fine-tuning with various loss functions, and offers extensive documentation and examples for applications like semantic search, clustering, and paraphrase mining.