FlagOpen/FlagEmbedding

Project overview

This toolkit provides embedding models, rerankers, fine-tuning, and evaluation utilities for text retrieval and retrieval-augmented generation. Developers can install it locally as a Python library to build and assess search pipelines.

Project type

RAG · AI Search

Use cases

Knowledge Q&A · Search & Research

Deployment

Refer to project documentation

License

MIT

Best for

• Developers, AI engineers, and researchers who need a self-hosted Python library to build, fine-tune, and evaluate text search and RAG systems.

Key capabilities

• Provides lightweight and LLM-based cross-encoder rerankers to improve search ranking performance.
• Supports fine-tuning embedding and reranker models via dedicated dependencies and examples.
• Provides examples and utilities for evaluating embedder and reranker performance.

Limitations and risks

• Users must install a dedicated, separate dependency to access and execute model fine-tuning capabilities.

Getting started

• Install the toolkit using a standard pip command and import FlagAutoModel to begin encoding sentences. No complex infrastructure is required for setup.

Evidence and sources

• README: ⚡️BGE: One-Stop Retrieval Toolkit For Search and RAG
• README: pip install -U FlagEmbedding
• README: from FlagEmbedding import FlagAutoModel model = FlagAutoModel.from_finetuned('BAAI/bge-base-en-v1.5', query_instruction_for_retrieval="Represent this sentence for searching releva…
• README: For more details, you can refer to [embedder inference](./examples/inference/embedder), [reranker inference](./examples/inference/reranker), [embedder finetune](./examples/finetun…
• README: BAAI/bge-reranker-v2-m3 | Multilingual | a lightweight cross-encoder model

Reviewed at: 2026-07-19T18:54:11Z