Research
Discrete Diffusion Inference-Time Control with Nested Sequential Monte Carlo
arXiv:2608.20123v1 Announce Type: cross Abstract: We study inference-time control for text generation in discrete diffusion language models, where the goal is to steer sampling toward sequence-level r
arXiv:2608.20123v1 Announce Type: cross Abstract: We study inference-time control for text generation in discrete diffusion language models, where the goal is to steer sampling toward sequence-level rewards without retraining. Prior work in this domain has focused on particle-based methods such as best-of-n sampling and bootstrap sequential Monte Carlo, which may suffer from overoptimism and weight degeneracy, respectively. We address these limitations using nested sequential Monte Carlo methods. We formulate nested SMC (NSMC) and fully-adapted nested SMC (FA-NSMC) for Feynman--Kac steering, identifying and correcting errors in prior formulations that lead to biased final estimates. We evaluate these methods on toxicity and fluency steering tasks, showing that NSMC and FA-NSMC consistently outperform best-of-n and bootstrap SMC.
Related
- A Mean-Field Framework for Inference-Time Distributional Control of Diffusion Models
- Sampling for Quality: Training-Free Reward-Guided LLM Decoding via Sequential Monte Carlo
- Accelerated Markov Chain Monte Carlo Algorithms on Discrete States
- Discrete Diffusion Language Models Are Training-Free Multi-Label Classifiers
- Discrete Tilt Matching
Source: arXiv cs.LG | 2026-08-21