Z²-Sampling implicitly realizes zero-cost zigzag trajectories for curvature-aware semantic alignment in diffusion models by reducing multi-step paths via operator dualities and temporal caching while synthesizing a directional derivative penalty.
Smoothed energy guidance: Guiding diffusion models with reduced energy curvature of attention
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
citation-role summary
baseline 1
citation-polarity summary
years
2026 3verdicts
UNVERDICTED 3roles
baseline 1polarities
baseline 1representative citing papers
Introduces a training-free inference-time method for aesthetic refinement in unconditional diffusion models using degradation concept vectors, bottleneck patching, and classifier-free guidance to steer away from degraded outputs.
Diffusion models overfit denoising loss at intermediate noise but generalize in inference as model error smooths the flow field and sampling paths avoid memorized noisy training data.
citing papers explorer
-
$Z^2$-Sampling: Zero-Cost Zigzag Trajectories for Semantic Alignment in Diffusion Models
Z²-Sampling implicitly realizes zero-cost zigzag trajectories for curvature-aware semantic alignment in diffusion models by reducing multi-step paths via operator dualities and temporal caching while synthesizing a directional derivative penalty.
-
Guidance for Low-Level Perceptual Editing in Unconditional Diffusion Models
Introduces a training-free inference-time method for aesthetic refinement in unconditional diffusion models using degradation concept vectors, bottleneck patching, and classifier-free guidance to steer away from degraded outputs.
-
Diffusion Models Memorize in Training -- and Generalize in Inference
Diffusion models overfit denoising loss at intermediate noise but generalize in inference as model error smooths the flow field and sampling paths avoid memorized noisy training data.