用20GB显存实现百亿知识图谱的高效注入，让大模型秒级调用海量知识。
AtlasKV: Augmenting LLMs with Billion-Scale Knowledge Graphs in 20GB VRAM
- 将知识图谱转为可直接嵌入的键值对，用注意力机制动态关联知识
- 百亿三元组知识注入仅需不到20GB显存，推理延迟低于传统检索方法
- 无需外部检索器或重新训练，适合快速更新知识的大模型应用
检索增强生成（RAG）虽能提升大语言模型（LLM）的外部知识能力，但依赖外部检索模块和长上下文，大规模知识引入时会带来显著推理延迟。本文提出一种参数化知识融合方法AtlasKV，可在不超过20GB VRAM条件下，高效集成百亿规模知识图谱（如10亿三元组）。通过KG2KV与HiKVP技术，实现亚线性的时间与内存复杂度，利用LLM自身注意力机制保持强知识对齐与泛化性能，且无需外部检索器、长上下文先验或重训练即可适配新知识。
原文摘要 · Abstract (English)
Retrieval-augmented generation (RAG) has shown some success in augmenting large language models (LLMs) with external knowledge. However, as a non-parametric knowledge integration paradigm for LLMs, RAG methods heavily rely on external retrieval modules and the retrieved textual context prior. Especially for very large scale knowledge augmentation, they would introduce substantial inference latency due to expensive searches and much longer relevant context. In this paper, we propose a parametric knowledge integration method, called \textbf{AtlasKV}, a scalable, effective, and general way to augment LLMs with billion-scale knowledge graphs (KGs) (e.g. 1B triples) using very little GPU memory cost (e.g. less than 20GB VRAM). In AtlasKV, we introduce KG2KV and HiKVP to integrate KG triples into LLMs at scale with sub-linear time and memory complexity. It maintains strong knowledge grounding and generalization performance using the LLMs' inherent attention mechanism, and requires no external retrievers, long context priors, or retraining when adapting to new knowledge.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。