SAVE: Sparse Autoencoder-Driven Visual Information Enhancement for Mitigating Object Hallucination

TL;DR

SAVE is a framework that reduces object hallucination in multimodal large language models by enhancing visual understanding through autoencoder-driven feature steering, leading to improved accuracy and robustness.

Contribution

We introduce SAVE, a novel autoencoder-based method that identifies and steers models along visual features to mitigate hallucination without additional training.

Findings

Outperforms state-of-the-art methods on standard benchmarks.

Achieves 10% improvement in CHAIR_S metric.

Enhances attention to image tokens and reduces uncertain object token generation.

Abstract

Although Multimodal Large Language Models (MLLMs) have advanced substantially, they remain vulnerable to object hallucination caused by language priors and visual information loss. To address this, we propose SAVE (Sparse Autoencoder-Driven Visual Information Enhancement), a framework that mitigates hallucination by steering the model along Sparse Autoencoder (SAE) latent features. A binary object-presence question-answering probe identifies the SAE features most indicative of the model's visual information processing, referred to as visual understanding features. Steering the model along these identified features reinforces grounded visual understanding and effectively reduces hallucination. With its simple design, SAVE outperforms state-of-the-art training-free methods on standard benchmarks, achieving a 10\%p improvement in CHAIR\_S and consistent gains on POPE and MMHal-Bench. Extensive evaluations across multiple models and layers confirm the robustness and generalizability of our approach. Further analysis reveals that steering along visual understanding features suppresses the generation of uncertain object tokens and increases attention to image tokens, mitigating hallucination. Code is released at https://github.com/wiarae/SAVE.