DRL trains a discriminator on data versus base-model samples in pretrained representation space and uses its logit as reward in KL-regularized RL, cutting guidance-free FID from 9.38 to 2.62 on SiT and similar gains on other backbones.
Input perturbation reduces exposure bias in diffusion models
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
years
2026 2verdicts
UNVERDICTED 2roles
background 1polarities
background 1representative citing papers
A cold diffusion model with direct and delta-normalized reverse processes, using UNet and transformer backbones, outperforms diffusion baselines for dereverberating acoustic and electronic drum stems on in-domain and out-of-domain tests.
citing papers explorer
-
The Reward Was in Your Data All Along: Correcting Flow Matching with Discriminator-Guided RL
DRL trains a discriminator on data versus base-model samples in pretrained representation space and uses its logit as reward in KL-regularized RL, cutting guidance-free FID from 9.38 to 2.62 on SiT and similar gains on other backbones.
-
A Cold Diffusion Approach for Percussive Dereverberation
A cold diffusion model with direct and delta-normalized reverse processes, using UNet and transformer backbones, outperforms diffusion baselines for dereverberating acoustic and electronic drum stems on in-domain and out-of-domain tests.