Knowledge Capsules: Structured Nonparametric Memory Units for LLMs

📅 2026-04-22
📈 Citations: 0
Influential: 0
📄 PDF
🤖 AI Summary
Updating knowledge in large language models is costly, and existing retrieval-augmented approaches exhibit instability in long-context and multi-hop reasoning scenarios. This work proposes “Knowledge Capsules”—structured, non-parametric memory units—and introduces an external Key-Value Injection (KVI) framework that directly integrates external knowledge into the model’s attention mechanism rather than merely appending it as additional context. By elevating knowledge integration from the contextual level to the memory level, this approach enables efficient and stable knowledge injection while keeping the base model parameters frozen. Experimental results demonstrate that the method significantly outperforms both RAG and GraphRAG across multiple question-answering benchmarks, achieving notably higher accuracy and robustness, particularly in tasks involving long contexts and multi-hop reasoning.

Technology Category

Application Category

📝 Abstract
Large language models (LLMs) encode knowledge in parametric weights, making it costly to update or extend without retraining. Retrieval-augmented generation (RAG) mitigates this limitation by appending retrieved text to the input, but operates purely through context expansion, where external knowledge competes as tokens within the attention mechanism. As a result, its influence is indirect and often unstable, particularly in long context and multi hop reasoning scenarios. We propose Knowledge Capsules, structured nonparametric memory units that represent normalized relational knowledge and can be constructed directly from document corpora using a frozen base model. Instead of injecting knowledge as text, we introduce an External Key Value Injection (KVI) framework that compiles capsules into attention-compatible key value representations, enabling external knowledge to directly participate in the model's attention computation. By shifting knowledge integration from context-level augmentation to memory level interaction, the proposed framework consistently outperforms RAG and GraphRAG across multiple QA benchmarks, with improved stability and accuracy in long context and multi hop reasoning, while requiring no parameter updates.
Problem

Research questions and friction points this paper is trying to address.

Large Language Models
Knowledge Updating
Retrieval-Augmented Generation
Long Context Reasoning
Multi-hop Reasoning
Innovation

Methods, ideas, or system contributions that make the work stand out.

Knowledge Capsules
Nonparametric Memory
Key-Value Injection
Retrieval-Augmented Generation
Attention Mechanism
🔎 Similar Papers
B
Bin Ju
Zhejiang Angel Medical AI Technology Co., Ltd., Hangzhou, China
S
Shenfeng Weng
Miti AI Technology Co., Ltd., Hangzhou, China
D
Danying Zhou
Zhejiang Angel Medical AI Technology Co., Ltd., Hangzhou, China
K
Kunkai Su
China-Singapore Belt and Road Joint Laboratory on Translational Infection Biology for Diagnostics and Therapies, State Key Laboratory for Diagnosis and Treatment of Infectious Diseases, The First Affiliated Hospital
R
Rongkai Xu
Zhejiang Angel Medical AI Technology Co., Ltd., Hangzhou, China