Search papers, labs, and topics across Lattice.
This paper investigates the phenomenon of hallucinations in diffusion models, identifying score smoothness as a key contributor to the generation of implausible samples. By linking the probability of hallucinations to the Lipschitz constant of the learned score function, the authors introduce a Variance-Guided Score Modulation (VSM) strategy that effectively reduces score smoothness. Empirical evaluations reveal that VSM can decrease hallucinations by up to 25% while preserving high fidelity and diversity in generated images, marking a significant advancement in the reliability of diffusion-based image generation.
Hallucinations in diffusion models can be reduced by up to 25% without sacrificing image quality, thanks to a novel score modulation technique.
Diffusion models have emerged as the backbone of modern generative AI, powering advances in vision, language, audio and other modalities. Despite their success, they suffer from hallucinations, implausible samples that lie outside the support of true data distribution, which degrade reliability and trust. In this work, we first empirically confirm previously proposed hypothesis that score smoothness causes hallucinations in Image Generation diffusion models and provide a density-based perspective. We further formalize this notion by linking the hallucinations probability mass to lipschitz constant of the learned score function. Motivated by this, we introduce a Variance-Guided Score Modulation (VSM) strategy that controls the score Jacobian, in turn reducing score smoothness and better approximating the ground truth score that decreases hallucinations. Empirical results on synthetic and real-world datasets demonstrate that our approach reduces hallucinations (up to ~25%) while maintaining high fidelity and diversity, providing a principled step toward more reliable diffusion-based image generation. We also propose two benchmark datasets with extreme semantic variation for systematic hallucination evaluation. Code and Datasets are publicly available at https://github.com/bhosalems/VSM.
