From RAG to Memory: Non-Parametric Continual Learning for Large Language Models

2025-02-20

AI Summary

To address the challenge of emulating the dynamicity and associative nature of human long-term memory in large language models (LLMs) during continual learning, this paper proposes HippoRAG 2—a non-parametric continual learning framework. It pioneers the integration of personalized PageRank into the retrieval-augmented generation (RAG) paradigm, jointly optimizing dynamic paragraph embedding fusion and LLM online inference to holistically enhance factual memory, comprehension memory, and associative memory. Compared to conventional RAG, HippoRAG 2 achieves significant improvements in semantic construction and factual retrieval; on associative memory tasks, it outperforms state-of-the-art embedding models by 7% in accuracy. Its core contribution lies in establishing a neurocognitively inspired knowledge updating and association modeling paradigm—mimicking human long-term memory mechanisms—and thereby overcoming the performance bottlenecks inherent in static retrieval and fixed representation schemes.

Technology Category

• Machine Learning: Life-Long and Continual Learning
• Natural Language Processing: (Large) Language Models
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Our ability to continuously acquire, organize, and leverage knowledge is a key feature of human intelligence that AI systems must approximate to unlock their full potential. Given the challenges in continual learning with large language models (LLMs), retrieval-augmented generation (RAG) has become the dominant way to introduce new information. However, its reliance on vector retrieval hinders its ability to mimic the dynamic and interconnected nature of human long-term memory. Recent RAG approaches augment vector embeddings with various structures like knowledge graphs to address some of these gaps, namely sense-making and associativity. However, their performance on more basic factual memory tasks drops considerably below standard RAG. We address this unintended deterioration and propose HippoRAG 2, a framework that outperforms standard RAG comprehensively on factual, sense-making, and associative memory tasks. HippoRAG 2 builds upon the Personalized PageRank algorithm used in HippoRAG and enhances it with deeper passage integration and more effective online use of an LLM. This combination pushes this RAG system closer to the effectiveness of human long-term memory, achieving a 7% improvement in associative memory tasks over the state-of-the-art embedding model while also exhibiting superior factual knowledge and sense-making memory capabilities. This work paves the way for non-parametric continual learning for LLMs. Our code and data will be released at https://github.com/OSU-NLP-Group/HippoRAG.

Problem

• Enhance continual learning in large language models
• Improve retrieval-augmented generation for factual memory
• Mimic human long-term memory in AI systems

Innovation

• HippoRAG 2 framework
• Personalized PageRank algorithm
• Non-parametric continual learning