KVSmooth: Mitigating Hallucination in Multi-modal Large Language Models through Key-Value Smoothing

TL;DR

KVSmooth is a training-free, plug-and-play method that reduces hallucinations in multimodal large language models by adaptively smoothing attention states based on entropy, improving factual consistency without retraining.

Contribution

We introduce KVSmooth, a novel inference-time technique that mitigates hallucination in MLLMs through attention-entropy-guided smoothing, without requiring retraining or model modification.

Findings

Significantly reduces hallucination scores (CHAIR) from 41.8 to 18.2.

Improves F1 score from 77.5 to 79.2, enhancing accuracy and recall.

Operates efficiently during inference without additional training.

Abstract

Despite the significant progress of Multimodal Large Language Models (MLLMs) across diverse tasks, hallucination -- corresponding to the generation of visually inconsistent objects, attributes, or relations -- remains a major obstacle to their reliable deployment. Unlike pure language models, MLLMs must ground their generation process in visual inputs. However, existing models often suffer from semantic drift during decoding, causing outputs to diverge from visual facts as the sequence length increases. To address this issue, we propose KVSmooth, a training-free and plug-and-play method that mitigates hallucination by performing attention-entropy-guided adaptive smoothing on hidden states. Specifically, KVSmooth applies an exponential moving average (EMA) to both keys and values in the KV-Cache, while dynamically quantifying the sink degree of each token through the entropy of its attention distribution to adaptively adjust the smoothing strength. Unlike computationally expensive retraining or contrastive decoding methods, KVSmooth operates efficiently during inference without additional training or model modification. Extensive experiments demonstrate that KVSmooth significantly reduces hallucination ($\mathit{CHAIR}_{S}$ from $41.8 \rightarrow 18.2$) while improving overall performance ($F_1$ score from $77.5 \rightarrow 79.2$), achieving higher precision and recall simultaneously. In contrast, prior methods often improve one at the expense of the other, validating the effectiveness and generality of our approach.