FlagEmbedding LLM

LLM-based reranking via BAAI FlagEmbedding.

Overview

Field	Value
Type	LLM-based
Library	FlagEmbedding
Default Model	BAAI/bge-reranker-v2-gemma

Installation

pip install "autorag-research[gpu]"
# or
uv add "autorag-research[gpu]"

Configuration

_target_: autorag_research.rerankers.flag_embedding_llm.FlagEmbeddingLLMReranker
model_name: BAAI/bge-reranker-v2-gemma

Options

Option	Type	Default	Description
model_name	str	BAAI/bge-reranker-v2-gemma	FlagEmbedding LLM model name
use_fp16	bool	False	Use FP16 for inference
batch_size	int	64	Batch size for multiple queries

Usage

from autorag_research.rerankers import FlagEmbeddingLLMReranker

reranker = FlagEmbeddingLLMReranker()
results = reranker.rerank("What is RAG?", ["doc1", "doc2", "doc3"], top_k=2)

When to Use

Good for higher quality reranking when compute resources are available. Uses LLM-scale models (e.g., Gemma) for more nuanced relevance scoring.