Research map / Image generation / Fast diffusion sampling
Blackout Diffusion: Generative Diffusion Models in Discrete-State Spaces
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use binomial bridge sampling instead of Poisson tau-leaping sampling (ablation) | Blackout Diffusion · CIFAR-10 | FID: 4.77 vs 4.92 (instantaneous); 4.83 vs 5.01 (finite-time) | Helped |
| Switch loss from instantaneous to finite-time (ablation) | Blackout Diffusion · CIFAR-10 | FID: 4.92 vs 5.01 (Poisson); 4.77 vs 4.83 (binomial) | No clear effect |
| Add 2 corrector steps (ablation) | Blackout Diffusion · CIFAR-10 | FID: 4.58 vs 4.77 | Helped |
| Blackout Diffusion (pure-death forward process, no corrector) vs discrete baselines | Blackout Diffusion · CIFAR-10 | FID: 4.77 vs D3PM Absorbing 30.97, D3PM Gauss 7.34, tauLDR-0 8.10; tauLDR-10 3.74 is better | Helped |
Closest papers by meaning
- Why Masking Diffusion Works: Condition on the Jump Schedule for Improved Discrete Diffusion (NeurIPS 2025 · similarity 0.90)
- Towards Non-Asymptotic Convergence for Diffusion-Based Generative Models (ICLR 2024 · similarity 0.90)
- Simplified and Generalized Masked Diffusion for Discrete Data (NeurIPS 2024 · similarity 0.90)
- Distributional Diffusion Models with Scoring Rules (ICML 2025 · similarity 0.90)
- Bridging discrete and continuous state spaces: Exploring the Ehrenfest process in time-continuous diffusion models (ICML 2024 · similarity 0.89)
- Scalable Discrete Diffusion Samplers: Combinatorial Optimization and Statistical Physics (ICLR 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.