TAG: Tangential Amplifying Guidance for Hallucination-Resistant Sampling

Hyunmin Cho, Donghoon Ahn, Susung Hong, Jee Eun Kim, Seungryong Kim, Kyong Hwan Jin
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:19672-19698, 2026.

Abstract

Diffusion models achieve state-of-the-art image generation but often produce semantic inconsistencies, or hallucinations. Existing inference-time guidance methods rely on external signals or architectural modifications, adding computational overhead. We propose Tangential Amplifying Guidance (TAG), a training-free, architecture-agnostic, plug-and-play guidance method that operates purely on trajectory signals. TAG uses an intermediate sample as a projection basis and amplifies the tangential components of the estimated score to correct the sampling trajectory. A first-order Taylor analysis shows that this steers the state toward higher-probability regions of the data manifold, reducing inconsistencies and improving fidelity while adding negligible overhead to existing samplers. Code is available at our Project Page (https://hyeon-cho.github.io/TAG/).

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-cho26a, title = {{TAG}: Tangential Amplifying Guidance for Hallucination-Resistant Sampling}, author = {Cho, Hyunmin and Ahn, Donghoon and Hong, Susung and Kim, Jee Eun and Kim, Seungryong and Jin, Kyong Hwan}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {19672--19698}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/cho26a/cho26a.pdf}, url = {https://proceedings.mlr.press/v306/cho26a.html}, abstract = {Diffusion models achieve state-of-the-art image generation but often produce semantic inconsistencies, or hallucinations. Existing inference-time guidance methods rely on external signals or architectural modifications, adding computational overhead. We propose Tangential Amplifying Guidance (TAG), a training-free, architecture-agnostic, plug-and-play guidance method that operates purely on trajectory signals. TAG uses an intermediate sample as a projection basis and amplifies the tangential components of the estimated score to correct the sampling trajectory. A first-order Taylor analysis shows that this steers the state toward higher-probability regions of the data manifold, reducing inconsistencies and improving fidelity while adding negligible overhead to existing samplers. Code is available at our Project Page (https://hyeon-cho.github.io/TAG/).} }
Endnote
%0 Conference Paper %T TAG: Tangential Amplifying Guidance for Hallucination-Resistant Sampling %A Hyunmin Cho %A Donghoon Ahn %A Susung Hong %A Jee Eun Kim %A Seungryong Kim %A Kyong Hwan Jin %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-cho26a %I PMLR %P 19672--19698 %U https://proceedings.mlr.press/v306/cho26a.html %V 306 %X Diffusion models achieve state-of-the-art image generation but often produce semantic inconsistencies, or hallucinations. Existing inference-time guidance methods rely on external signals or architectural modifications, adding computational overhead. We propose Tangential Amplifying Guidance (TAG), a training-free, architecture-agnostic, plug-and-play guidance method that operates purely on trajectory signals. TAG uses an intermediate sample as a projection basis and amplifies the tangential components of the estimated score to correct the sampling trajectory. A first-order Taylor analysis shows that this steers the state toward higher-probability regions of the data manifold, reducing inconsistencies and improving fidelity while adding negligible overhead to existing samplers. Code is available at our Project Page (https://hyeon-cho.github.io/TAG/).
APA
Cho, H., Ahn, D., Hong, S., Kim, J.E., Kim, S. & Jin, K.H.. (2026). TAG: Tangential Amplifying Guidance for Hallucination-Resistant Sampling. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:19672-19698 Available from https://proceedings.mlr.press/v306/cho26a.html.

Related Material