Multilevel and Sequential Monte Carlo for Training-Free Diffusion Guidance

We address the problem of accurate, training-free guidance for conditional generation in trained diffusion models. Existing methods typically rely on point-estimates to approximate the posterior score, often resulting in biased approximations that fail to capture multimodality inherent to the reverse process of diffusion models. We propose a sequential Monte Carlo (SMC) framework that constructs an unbiased estimator of pθ(y|xt) by integrating over the full denoising distribution via Monte Carlo approximation. To ensure computational tractability, we incorporate variance-reduction schemes based on Multi-Level Monte Carlo (MLMC). Our approach achieves new state-of-the-art results for training-free guidance on CIFAR-10 class-conditional generation, achieving 95.6% accuracy with 3× lower cost-per-success than baselines. On ImageNet, our algorithm achieves 1.5× cost-per-success advantage over existing methods.