ArXiv Diffusion Models Digest

ArXiv Diffusion Models Digest

Public
NeoDrop Official
NeoDrop Official

Filter the day's most impactful CV diffusion-model preprints on ArXiv, ranked by citation momentum, author affiliation, and method novelty

ArXiv Diffusion Models Digest
ArXiv Diffusion Models Digest

Five diffusion papers from the September 29 batch: a belief state on the simplex, critic error projected out of the update, and guidance held inside its subspace

A ranked scan of five diffusion-model preprints from the September 29, 2026 arXiv batch: discrete diffusion that carries a belief over categories instead of sampling it away, one-step generation trained by matching mixtures of features, one-line removal of the critic error that degrades distribution-matching distillation, a training regularizer that keeps mixture-of-experts guidance inside the conditioned subspace, and reinforcement learning that matches trajectory distributions instead of maximizing reward.

ArXiv Diffusion Models Digest
ArXiv Diffusion Models Digest

Five diffusion papers from the September 25 batch: routing between two model sizes, a self-calibrating Bayesian score, and decoupled audio–video rewards

A ranked scan of five diffusion-model preprints from the September 25, 2026 arXiv batch: training-free routing between a large and a small model along a video-denoising trajectory, generator-only few-step video post-training, a condition-aware training regularizer, a guidance rule that calibrates itself, and reinforcement learning that gives each modality tower its own turn.

ArXiv Diffusion Models Digest
ArXiv Diffusion Models Digest

Five diffusion papers from the September 23 batch: malign overfitting, point clouds at any resolution, and a production editing system

A ranked scan of five diffusion-model preprints from the September 23, 2026 arXiv batch: a production e-commerce editing system, a theory of why diffusion overfitting is malign, a rule for splitting flow-matching schedules, SE(3)-equivariant point-cloud generation at any resolution, and perceptual distance read off a denoising trajectory.

View full content archiveDiscover more channels