EPISODE · Oct 14, 2025 · 21 MIN
TAG:Tangential Amplifying Guidance for Hallucination-Resistant Diffusion Sampling
from Daily Paper Cast · host Jingwen Liang, Gengyu Wang
🤗 Upvotes: 38 | cs.CV Authors: Hyunmin Cho, Donghoon Ahn, Susung Hong, Jee Eun Kim, Seungryong Kim, Kyong Hwan Jin Title: TAG:Tangential Amplifying Guidance for Hallucination-Resistant Diffusion Sampling Arxiv: http://arxiv.org/abs/2510.04533v1 Abstract: Recent diffusion models achieve the state-of-the-art performance in image generation, but often suffer from semantic inconsistencies or hallucinations. While various inference-time guidance methods can enhance generation, they often operate indirectly by relying on external signals or architectural modifications, which introduces additional computational overhead. In this paper, we propose Tangential Amplifying Guidance (TAG), a more efficient and direct guidance method that operates solely on trajectory signals without modifying the underlying diffusion model. TAG leverages an intermediate sample as a projection basis and amplifies the tangential components of the estimated scores with respect to this basis to correct the sampling trajectory. We formalize this guidance process by leveraging a first-order Taylor expansion, which demonstrates that amplifying the tangential component steers the state toward higher-probability regions, thereby reducing inconsistencies and enhancing sample quality. TAG is a plug-and-play, architecture-agnostic module that improves diffusion sampling fidelity with minimal computational addition, offering a new perspective on diffusion guidance.
What this episode covers
🤗 Upvotes: 38 | cs.CV Authors: Hyunmin Cho, Donghoon Ahn, Susung Hong, Jee Eun Kim, Seungryong Kim, Kyong Hwan Jin Title: TAG:Tangential Amplifying Guidance for Hallucination-Resistant Diffusion Sampling Arxiv: http://arxiv.org/abs/2510.04533v1 Abstract: Recent diffusion models achieve the state-of-the-art performance in image generation, but often suffer from semantic inconsistencies or hallucinations. While various inference-time guidance methods can enhance generation, they often operate indirectly by relying on external signals or architectural modifications, which introduces additional computational overhead. In this paper, we propose Tangential Amplifying Guidance (TAG), a more efficient and direct guidance method that operates solely on trajectory signals without modifying the underlying diffusion model. TAG leverages an intermediate sample as a projection basis and amplifies the tangential components of the estimated scores with respect to this basis to correct the sampling trajectory. We formalize this guidance process by leveraging a first-order Taylor expansion, which demonstrates that amplifying the tangential component steers the state toward higher-probability regions, thereby reducing inconsistencies and enhancing sample quality. TAG is a plug-and-play, architecture-agnostic module that improves diffusion sampling fidelity with minimal computational addition, offering a new perspective on diffusion guidance.
NOW PLAYING
TAG:Tangential Amplifying Guidance for Hallucination-Resistant Diffusion Sampling
No transcript for this episode yet
Similar Episodes
No similar episodes found.