Megadose AI progress, ranked and analyzed.

Spectral Prior for Reducing Exposure Bias in Diffusion Models

· HF Daily Papers ·
SPA targets diffusion sampling errors by matching intermediate frequency spectra to a learned training-data prior.

The paper says exposure bias shows up as frequency-dependent SNR mismatch between training and inference. That mismatch changes by model and timestep, so the authors argue fixed correction rules do not travel well. Their Spectral Alignment method fits a spectrum model offline, then applies FFT-based guidance during inference with reported overhead of 3-4%. They report gains across DDPM, ADM, SD2.0, SDXL, SD3.5, and FLUX, and say the method is complementary to CFG. HF Daily Papers' note

score 5

Categories: Research