FinanceBench RAG Eval
What it is
A production-grade Retrieval-Augmented Generation (RAG) system for the FinanceBench financial dataset, built with LangGraph, Qdrant, and LLM-as-judge evaluation. It features a CRAG pipeline with hybrid retrieval, reranking, and rigorous multi-tier evaluation calibrated against human labels.
The problem it solves
Lack of rigorous, calibrated evaluation methods for RAG systems in numerical-heavy domains like finance, where 'vibe checks' are insufficient and hallucinations are costly.
What you could use it for
- Evaluating RAG systems on financial data with objective ground truth
- Implementing CRAG pipelines with hybrid retrieval and reranking
- Calibrating LLM-as-judge evaluations for numerical precision in financial contexts
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.