Open SourceRAG16k
Sentence Transformers
The standard toolkit for text and image embeddings and rerankers.
Sentence Transformers (SBERT) is a Python framework for computing dense vector embeddings of sentences, paragraphs, and images, plus training and using cross-encoder rerankers. It exposes thousands of pretrained models and a simple API for encoding, semantic similarity, and fine-tuning on custom data. It is the most widely used library for generating the embeddings that feed vector databases and RAG retrieval.
Repository
UKPLab/sentence-transformersLanguage
PythonWhat you'd build with it
- Generating document and query embeddings to index into a vector store for semantic search and RAG
- Fine-tuning an embedding model on domain data to improve retrieval relevance
- Adding a cross-encoder reranking stage to reorder top-k retrieved chunks before passing them to an LLM
Tags
embeddingsrerankingsemantic-searchnlp