从上下文感知到冲突感知：泛化对比解码以应对LLMs中的知识冲突
From Context-Aware to Conflict-Aware: Generalizing Contrastive Decoding for Knowledge Conflict in LLMs
AI总结：
针对大语言模型生成时外部上下文与参数先验之间的知识冲突，提出冲突感知范式，通过动态分配先验与上下文的权重，并设计自适应机制解决不同冲突状态下的不对称问题。
摘要：
当大语言模型从检索或增强的上下文中生成时，外部上下文与参数先验之间的冲突仍然是核心可靠性瓶颈。现有的对比解码方法遵循一种\emph{上下文感知}范式，单方面放大上下文而压制参数先验，当上下文错误时会覆盖正确的先验。我们将其泛化为\textbf{冲突感知}范式，基于冲突信号动态分配先验与上下文的权威，而非预设上下文的可信度。我们证明，先验和上下文logits的仿射组合产生一个\textbf{幂族}，具有固有的\textbf{状态不对称性}：当先验正确时外推会无界放大错误，当上下文正确时内插会纠正不足，且没有静态状态能同时覆盖两者。现有的对比解码方法是该族实例，大多为外推型。为评估两种冲突方向，我们提出TriState-Bench，一种模型感知的评估协议，校准每个模型的先验知识以测量三种冲突状态：纠正、抵抗和一致。为解决不对称性，我们提出自适应状态路由（ARR），在每一步在状态间路由，将抵抗EM从低于6提升至16-33，且不牺牲纠正或一致。我们的代码可在该https URL获取。
英文摘要：
When large language models generate from retrieved or augmented contexts, conflicts between external context and parametric priors remain a central reliability bottleneck. Existing contrastive decoding methods follow a \emph{context-aware} paradigm that unilaterally amplifies context over parametric priors, overwriting correct priors when the context is erroneous. We generalize this to the \textbf{conflict-aware} paradigm that dynamically allocates authority between prior and context based on conflict signals, rather than presupposing context trustworthiness. We show that the affine combination of prior and context logits yields a \textbf{power family} with an inherent \textbf{regime asymmetry}: extrapolation amplifies errors unboundedly when the prior is correct, interpolation under-corrects when the context is correct, and no static regime covers both. Existing contrastive decoding methods are instances of this family, mostly extrapolative. To evaluate both conflict directions, we propose TriState-Bench, a model-aware evaluation protocol that calibrates per-model prior knowledge to measure three conflict states: correction, resistance, and agreement. To resolve the asymmetry, we propose Adaptive Regime Routing (ARR), which routes between regimes at each step, lifting resistance EM from below 6 to 16--33 without sacrificing correction or agreement. Our code is available at https://github.com/keith-Jiang/conflict-aware-decoding.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
