Continuously Steering LLMs Sensitivity to Contextual Knowledge with Proxy Models

2025-08-27

AI Summary

Large language models (LLMs) suffer from conflicts between parametric knowledge and contextual knowledge during generation, yet existing methods struggle to efficiently and continuously modulate context sensitivity—particularly for black-box LLMs. Method: We propose CSKS, a lightweight framework that employs two small proxy models to characterize output distribution discrepancies, enabling weight-free, continuous, and controllable adjustment of the base LLM’s context sensitivity without accessing or modifying its parameters. Contribution/Results: CSKS supports black-box deployment and achieves, for the first time, fine-grained, differentiable control over context dependency. Experiments on both synthetic and real-world knowledge-conflict benchmarks demonstrate significant improvements in model adaptability: CSKS flexibly amplifies or attenuates contextual influence while preserving parametric knowledge integrity.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Search and Optimization: Learning to Search

Application Category

• 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
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

In Large Language Models (LLMs) generation, there exist knowledge conflicts and scenarios where parametric knowledge contradicts knowledge provided in the context. Previous works studied tuning, decoding algorithms, or locating and editing context-aware neurons to adapt LLMs to be faithful to new contextual knowledge. However, they are usually inefficient or ineffective for large models, not workable for black-box models, or unable to continuously adjust LLMs' sensitivity to the knowledge provided in the context. To mitigate these problems, we propose CSKS (Continuously Steering Knowledge Sensitivity), a simple framework that can steer LLMs' sensitivity to contextual knowledge continuously at a lightweight cost. Specifically, we tune two small LMs (i.e. proxy models) and use the difference in their output distributions to shift the original distribution of an LLM without modifying the LLM weights. In the evaluation process, we not only design synthetic data and fine-grained metrics to measure models' sensitivity to contextual knowledge but also use a real conflict dataset to validate CSKS's practical efficacy. Extensive experiments demonstrate that our framework achieves continuous and precise control over LLMs' sensitivity to contextual knowledge, enabling both increased sensitivity and reduced sensitivity, thereby allowing LLMs to prioritize either contextual or parametric knowledge as needed flexibly. Our data and code are available at https://github.com/OliveJuiceLin/CSKS.

Problem

• Resolving knowledge conflicts between LLMs' internal and contextual knowledge
• Enabling continuous sensitivity adjustment to contextual knowledge
• Achieving lightweight control without modifying LLM weights

Innovation

• Uses proxy models to adjust sensitivity
• Shifts LLM output without weight modification
• Enables continuous knowledge sensitivity control