What AI says about MemoRAG

MemoRAG is an open-source memory-based retrieval-augmented generation framework that leverages a long-memory model to recall query clues and deliver more accurate, context-rich responses across all-purpose applications. It features Lite mode for scalable memory-augmented RAG, supports memory models such as Meta-Llama-3.1-8B-Instruct and Llama3.1-8B-Chinese-Chat, and ships training scripts, datasets, demos, and documentation. The project aims to advance RAG through memory-inspired knowledge discovery, with its accompanying paper accepted by WebConf 2025.