What AI says about CRUD-RAG

CRUD-RAG is the official codebase and Chinese benchmark for evaluating retrieval-augmented generation (RAG) systems, including datasets and tutorials to run experiments. The prompts are designed for ChatGPT, with guidance that smaller models (e.g., 7B) may struggle with complex prompts, and the evaluation uses a GPT-based RAGQuestEval metric for QA and generation. A one-time process to build a vector index for the text is required when you first run the code.