Why compositional diffusion fails: sampler, not model — MCMC samplers fix it
MarkNeumannnn · x · 2026-09-17
Mark Neumann shared two references on "conservative diffusion": the 2023 arXiv paper Reduce, Reuse, Recycle (Yilun Du, Sander Dieleman, Jascha Sohl-Dickstein, Will Grathwohl et al.), which shows the sampler—not the model—is responsible for compositional generation failures and proposes MCMC-inspired, energy-based parameterizations with Metropolis-corrected samplers that markedly improve compositional generation on ImageNet-style tasks; plus an apparently unpublished discussion of why this isn't done in practice for images.
Related event: Compositional Generation Failures in Diffusion Models Traced to Samplers(2 posts)→
More from Research
- Doctor: AI is crushing maths but has barely touched clinical trials and medicine — LaCaipirinha · 2026-09-17
- ColabFold 1.6.3 Released: 2.5x Faster, pip-Installable, Adds ipSAE+pDockQ2 Scores — sokrypton · 2026-09-17
- Dev claims Jev's non-autoregressive JSON-schema architecture mirrors his open-sourced year-old work — Nandakishor_ml · 2026-09-17
- Gaze as evidence for common grounding: cross-corpus analysis of MapTask and MUNDEX — cs-nlp-uu · 2026-09-17
- Microsoft's ProgramDistill: 4,063 tasks expose coding agents' interactive blind spots — MicrosoftResearch · 2026-09-17
- Fly brains simulated in virtual worlds spark debate on digital consciousness ownership — MightyGrilledSalmon · 2026-09-17