Vibe Coding Discoverfastembed
View on GitHubFast, Accurate, Lightweight Python library to make State of the Art Embedding
FastEmbed is a lightweight Python library for generating dense, sparse, image, and late-interaction embeddings, plus reranking scores. It runs models with ONNX Runtime and supports CPU and GPU inference for retrieval pipelines.
Use Cases
Generate text embeddings for semantic searchCreate sparse embeddings for lexical retrievalEmbed images for multimodal searchGenerate late-interaction embeddings for document retrievalRerank search resultsBuild retrieval pipelines for QdrantRun embedding generation in lightweight serverless environments
Built With
- Language
- Python
- Frameworks
- ONNX Runtime · Hugging Face Hub · NumPy · Qdrant
Tags
embeddings · dense embeddings · sparse embeddings · reranking · vector search · retrieval · multimodal · ONNX · CPU inference · GPU inference