Overview

Diffusion sampling starts with noise and repeatedly denoises it to generate data, using reverse stochastic processes such as DDPM or deterministic flows such as DDIM and probability-flow ODEs. Work improves this process with higher-order or adaptive solvers, learned timestep schedules, and methods that reuse or predict model computations; discrete and masked diffusion models also require choices about which tokens to reveal and when. Guidance, particle methods, and search can steer samples toward rewards, constraints, diversity, or solutions to inverse problems, with applications spanning image and text generation, molecules, and scientific modeling. A central tension is reducing steps and computation without losing fidelity, diversity, or correctness, while theory seeks sharper guarantees for error, stability, and sampling cost.

Latest papers

287 papers23 papers in the last four weeks, down 55% on the four weeks before. 0.2% of all new papers.

Jan 26, 2026

ART for Diffusion Sampling: A Reinforcement Learning Approach to Timestep Scheduling

We consider time discretization for score-based diffusion models to generate samples from a learned reverse-time dynamic on a finite grid. Uniform and hand-crafted grids can be suboptimal given a budget on the number of time steps. We introduce Adaptive Reparameterized Time (ART), which controls the clock speed of a reparameterized time variable to redistribute computation along the sampling trajectory while preserving the terminal time, with the objective of minimizing the aggregate Euler discretization error. We derive a randomized companion ART-RL that recasts ART as a continuous-time reinforcement learning problem with Gaussian policies, and prove a two-directional bridge between the two: the deterministic ART optimum lifts to an optimal Gaussian policy, and conversely any optimal Gaussian policy must recover the ART control through its mean. This bridge turns continuous-time actor--critic learning into a principled, rather than heuristic, route to the deterministic timestep optimum. Within the official EDM pipeline, ART-RL improves FID on CIFAR--10 across a wide range of budgets; after one-time offline training, the distilled deterministic schedule transfers without retraining to AFHQv2, FFHQ, and ImageNet at no extra inference cost.
Jan 20, 2026

A Gaussian Perspective for Distributional Discrepancy in Generative Diffusion Models

This paper introduces an analytical approach to quantifying and optimizing the distributional discrepancy in generative diffusion models. For a multivariate Gaussian source, we explicitly derive the closed-form evolution trajectory and the resulting Kullback-Leibler (KL) divergence between the distributions of the source data and the reversely sampled data. Asymptotic analysis via the Euler-Maclaurin expansion characterizes the convergence behavior of this KL divergence, extracting its dominant term as an explicit functional of the noise schedule. Minimizing this dominant term via the calculus of variations yields a noise schedule described by a tangent law, inherently determined by the source covariance spectrum. We further prove that the Gaussian source exhibits an extremal property for the KL divergence among general source distributions with a given covariance. We also utilize the analytical KL divergence as a principled metric to identify efficient time discretization strategies for pretrained diffusion models, and demonstrate via experiments over diverse datasets that the identified strategies consistently outperform established baselines, particularly under constrained function evaluation budgets.
Jan 6, 2026

Systematic Hazard Sampling: Minimal-Variance Inference for Discrete Diffusion and Flow Models

Uniform-noise discrete diffusion and flow models generate sequences non-autoregressively through iterative, context-dependent token replacements. However, these models are typically formulated as time-inhomogeneous continuous- or discrete-time Markov chains (CTMC/DTMC), sampled using independent Bernoulli change decisions per discretization step. This induces Poisson-binomial variance in per-position jump counts that grows with the number of required edits, leading to the common under-editing (residual noise) and over-editing (cascading substitutions) failure modes that degrade sample quality. We identify this sampler-induced variance as an orthogonal source of degradation, distinct from model-side errors and addressable purely at inference time. We propose Systematic Hazard Sampling (SHS), a training-free, drop-in, and hyperparameter-free inference principle for any sampler that admits a stay-vs.-replace decomposition. SHS models per-token edits as events driven by cumulative hazard (CTMC) or jump mass (DTMC) and triggers an edit whenever this quantity exceeds unit-spaced thresholds with a single random phase per position. For any fixed cumulative mass, this preserves the expected jump count while achieving the minimum conditional variance possible among unbiased integer estimators (at most 1/4), without altering per-jump destination sampling. Experiments on four uniform-noise discrete diffusion and flow language models spanning ~110M to ~3B parameters show that SHS consistently improves sample quality across numbers of function evaluations (NFE). Code is available at https://github.com/Jang-seunghwan/Systematic-Hazard-Sampling.
Dec 30, 2025

Fast Score-Based Sampling via Log-Concave Reductions

Sampling based on score diffusions has led to striking empirical results, and has attracted considerable attention from various research communities. It depends on availability of (approximate) Stein score functions for various levels of additive noise. We show how in some generality, the availability of scores allows the general problem to be ``reduced'' to sampling from an adaptively constructed sequence of KK strongly log-concave (SLC) sub-problems. The reduction is simple, constructive and algorithm-independent, so that any SLC sampler can be used as a subroutine. Various bounds on score-based sampling complexity follow directly: for instance, high-accuracy SLC samplers yield O~(Kdpolylog⁡(1/ε))\tilde{\mathcal{O}}(K \sqrt{d} \operatorname{polylog}(1/\varepsilon)) guarantees for accuracy ε\varepsilon in dimension dd, where randomized midpoint SLC schemes yield O~(Kd1/3poly⁡(1/ε))\tilde{\mathcal{O}}(K d^{1/3} \operatorname{poly}(1/\varepsilon)) guarantees. When the original distribution itself is SLC, we prove that K≤1+log⁡2(κ)K \leq 1 + \log_2(κ), thereby obtaining the first efficient procedure with logarithmic dependence on condition number κκ; for general distributions, the quantity KK depends on the geometry of score Hessian across the trajectory. Our analysis is direct and simple, involving techniques and insights complementary to those in standard analyses of discretized diffusions.
Nov 11, 2025

VEDA: Generation of 3D Molecules via Variance-Exploding Diffusion with Annealing

Diffusion models show promise for 3D molecular generation, but face a fundamental trade-off between sampling efficiency and conformational accuracy. While flow-based models are fast, they often produce geometrically inaccurate structures, as they have difficulty capturing the multimodal distributions of molecular conformations. In contrast, denoising diffusion models are more accurate but suffer from slow sampling, a limitation attributed to sub-optimal integration between diffusion dynamics and SE(3)-equivariant architectures. To address this, we propose VEDA, a unified SE(3)-equivariant framework that combines variance-exploding diffusion with annealing to efficiently generate conformationally accurate 3D molecular structures. Specifically, our key technical contributions include: (1) a VE schedule that enables noise injection functionally analogous to simulated annealing, improving 3D accuracy and reducing relaxation energy; (2) a novel preconditioning scheme that reconciles the coordinate-predicting nature of SE(3)-equivariant networks with a residual-based diffusion objective, and (3) a new arcsin-based scheduler that concentrates sampling in critical intervals of the logarithmic signal-to-noise ratio. On the QM9 and GEOM-DRUGS datasets, VEDA matches the sampling efficiency of flow-based models, achieving state-of-the-art valency stability and validity with only 100 sampling steps. More importantly, VEDA's generated structures are remarkably stable, as measured by their relaxation energy during GFN2-xTB optimization. The median energy change is only 1.72 kcal/mol, significantly lower than the 32.3 kcal/mol from its architectural baseline, SemlaFlow. Our framework demonstrates that principled integration of VE diffusion with SE(3)-equivariant architectures can achieve both high chemical accuracy and computational efficiency.
Oct 9, 2025

Guided Star-Shaped Masked Diffusion

The performance of pre-trained masked diffusion models is often constrained by their sampling procedure, which makes decisions irreversible and struggles in low-step generation regimes. We introduce a novel sampling algorithm that works with pre-trained models and, after a lightweight fine-tuning of a single layer, significantly improves sample quality and efficiency. Our method reformulates the generation process using a star-shaped paradigm, which inherently allows for error correction. To make this process effective, we augment it with a learnable re-masking scheduler that intelligently identifies and revises likely errors. This approach yields a substantial quality boost, particularly when using a small number of sampling steps. We extensively ablate key components of our approach and show its usability in different scenarios. In comprehensive experiments on text, and code generation, our sampling algorithm outperforms or matches existing methods.
Oct 6, 2025

Demystifying MaskGIT Sampler and Beyond: Adaptive Order Selection in Masked Diffusion

Masked diffusion models have shown promising performance in generating high-quality samples in a wide range of domains, but accelerating their sampling process remains relatively underexplored. To investigate efficient samplers for masked diffusion, this paper theoretically analyzes the MaskGIT sampler for image modeling, revealing its implicit temperature sampling mechanism. Through this analysis, we show that MaskGIT is asymptotically equivalent to a choose-then-sample (CTS) formulation, instantiated as the "moment sampler," which explicitly separates index selection from token sampling. This CTS reformulation is essential: it yields unbiased token sampling and exposes an algorithmic design space for index selection, both of which are inaccessible in MaskGIT's original formulation. Regarding token sampling, we reveal that MaskGIT implicitly adopts a low-temperature sampler, which explains why MaskGIT often degrades with more sampling steps. The CTS reformulation of MaskGIT allows us to fix the temperature sampling to ensure unbiasedness. We also improve the index selection in CTS through two key innovations: a partial caching technique for transformers that approximates longer sampling trajectories without proportional computational cost, and a hybrid approach formalizing the exploration-exploitation trade-off in adaptive unmasking. Experiments in image and text domains demonstrate our theory as well as the efficiency of our proposed methods, advancing both theoretical understanding and practical implementation of masked diffusion samplers.
Sep 15, 2025

IS-Diff: Improving Diffusion-Based Inpainting with Better Initial Seed

Diffusion models have shown promising results in free-form inpainting. Recent studies based on refined diffusion samplers or novel architectural designs led to realistic results and high data consistency. However, random initialization seed (noise) adopted in vanilla diffusion process may introduce mismatched semantic information in masked regions, leading to biased inpainting results, e.g., low consistency and low coherence with the other unmasked area. To address this issue, we propose the Initial Seed refined Diffusion Model (IS-Diff), a completely training-free approach incorporating distributional harmonious seeds to produce harmonious results. Specifically, IS-Diff employs initial seeds sampled from unmasked areas to imitate the masked data distribution, thereby setting a promising direction for the diffusion procedure. Moreover, a dynamic selective refinement mechanism is proposed to detect severe unharmonious inpaintings in intermediate latent and adjust the strength of our initialization prior dynamically. We validate our method on both standard and large-mask inpainting tasks using the CelebA-HQ, ImageNet, and Places2 datasets, demonstrating its effectiveness across all metrics compared to state-of-the-art inpainting methods.
Jul 29, 2025

MixGRPO: Unlocking Flow-based GRPO Efficiency with Mixed ODE-SDE

Although GRPO substantially enhances flow matching models in human preference alignment of image generation, methods such as FlowGRPO and DanceGRPO still exhibit inefficiency due to the necessity of sampling and optimizing over all denoising steps specified by the Markov Decision Process (MDP). In this paper, we propose MixGRPO\textbf{MixGRPO}, a novel framework that leverages the flexibility of mixed sampling strategies through the integration of stochastic differential equations (SDE) and ordinary differential equations (ODE). This streamlines the optimization process within the MDP to improve efficiency and boost performance. Specifically, MixGRPO introduces a sliding window mechanism, using SDE sampling and GRPO-guided optimization only within the window, while applying ODE sampling outside. This design confines sampling randomness to the time-steps within the window, thereby reducing the optimization overhead, and allowing for more focused gradient updates to accelerate convergence. Additionally, as time-steps beyond the sliding window are not involved in optimization, higher-order solvers are supported for faster sampling. So we present a faster variant, termed MixGRPO-Flash\textbf{MixGRPO-Flash}, which further improves training efficiency while achieving comparable performance. MixGRPO exhibits substantial gains across multiple dimensions of human preference alignment, outperforming DanceGRPO in both effectiveness and efficiency, with nearly 50% lower training time. Notably, MixGRPO-Flash further reduces training time by 71%.
Jun 25, 2025

Ctrl-Z Sampling: Scaling Diffusion Sampling with Controlled Random Zigzag Explorations

Diffusion models generate conditional samples by progressively denoising Gaussian noise, yet the denoising trajectory can stall at visually plausible but low-quality outcomes with conditional misalignment or structural artifacts. We interpret this behavior as local optima in a surrogate quality landscape: Once early denoising commits to a suboptimal global structure, later steps mainly sharpen details and seldom correct the underlying mistake. While existing inference-time approaches explore alternative diffusion states via re-noising with fixed strength or direction, they exhibit limited capacity to escape steep quality plateaus. We propose Controlled Random Zigzag Sampling (Ctrl-Z Sampling),a scalable sampling strategy that detects plateaus in quality landscape via a surrogate score, and allocates exploration only when a plateau is detected. Upon detection, Ctrl-Z Sampling rolls back to noisier states, samples a set of alternative continuations, and updates the trajectory when a candidate improves the score, otherwise escalating the exploration depth to escape the current plateau. The proposed method is model-agnostic and broadly compatible with existing diffusion frameworks. Experiments show that Ctrl-Z Sampling consistently improves generation quality over other inference-time scaling samplers across different NFE budgets, offering a scalable compute-quality trade-off. Code available at: https://github.com/ShunqiM/Ctrl-Z-Sampling.
Jun 16, 2025

Fast Convergence for High-Order ODE Solvers in Diffusion Probabilistic Models

Diffusion probabilistic models generate samples by learning to reverse a noise-injection process that transforms data into noise. A key development is the reformulation of the reverse sampling process as a deterministic probability flow ordinary differential equation (ODE), which allows for efficient sampling using high-order numerical solvers. Unlike traditional time integrator analysis, the accuracy of this sampling procedure depends not only on numerical integration errors but also on the approximation quality and regularity of the learned score function, as well as their interaction. In this work, we present a rigorous convergence analysis of deterministic samplers derived from probability flow ODEs for general forward processes with arbitrary variance schedules. Specifically, we develop and analyze pp-th order (exponential) Runge-Kutta schemes, under the practical assumption that the first and second derivatives of the learned score function are bounded. We prove that the total variation distance between the generated and target distributions can be bounded as \begin{align*} O\bigl(d^{\frac{7}{4}}\varepsilon_{\text{score}}^{\frac{1}{2}} +d(dH_{\max})^p\bigr), \end{align*} where εscore2\varepsilon^2_{\text{score}} denotes the L2L^2 error in the score function approximation, dd is the data dimension, and Hmax⁡H_{\max} represents the maximum solver step size. Numerical experiments on benchmark datasets further confirm that the derivatives of the learned score function are bounded in practice.
Jun 13, 2025

The Effect of Stochasticity in Score-Based Diffusion Sampling: a KL Divergence Analysis

Sampling in score-based diffusion models can be performed by solving either a reverse-time stochastic differential equation (SDE) parameterized by an arbitrary stochasticity function or a probability flow ODE, corresponding to setting this stochasticity function to zero. In this work, we investigate the effect of this stochasticity on the generation process through the evolution of Kullback-Leibler (KL) divergences, obtaining general KL divergence bounds and a novel analysis of the impact of the time-profile of the score error on model performance. For exact score functions, stochasticity has a contractive effect, decreasing KL divergence along the sampling trajectory. For approximate scores, however, a trade-off arises between correcting accumulated errors and amplifying current score errors, meaning stochasticity can either improve or degrade generation performance. Theoretical considerations indicate that the gain from stochasticity depends on the time-localization of the trained model error. We test this in experiments on both toy and benchmark data sets, also comparing the KL divergence evolution with the obtained bounds. We also present a fully analytical example, where all the relevant quantities can be computed, and the optimal stochasticity function can be characterized via an optimal control analysis.
Feb 11, 2025

DiffRatio: Training One-Step Diffusion Models Without Teacher Supervision

Score-based distillation methods train one-step diffusion models in two stages: they first train a teacher score model, then distill it into a one-step student model. However, this distillation process can introduce bias from two sources: errors in the teacher score model and student score estimate. We propose DiffRatio, a new approach for training one-step diffusion models without teacher supervision based on a density-ratio formulation that connects score-based distribution matching and GAN-style adversarial training. Instead of using a teacher score model to provide training targets, DiffRatio directly learns a log density ratio between the student and data distributions across diffusion time steps. As a pre-trained score model is only used to initialize the one-step generator and does not supervise training, this design simplifies the training pipeline, mitigates gradient estimation bias, and reduces the size of auxiliary networks. The learned density ratio can be used as a verifier, enabling a principled inference-time parallel scaling method that further improves sample quality without external rewards or extra sequential computation. DiffRatio achieves strong one-step generation results on CIFAR-10 and ImageNet (64x64 and 512x512), outperforming most teacher-supervised distillation methods.
Jan 22, 2025

Low-dimensional adaptation of diffusion models: Convergence in total variation

This paper investigates how diffusion generative models leverage (unknown) low-dimensional structure to accelerate sampling. Focusing on two mainstream samplers -- the denoising diffusion implicit model (DDIM) and the denoising diffusion probabilistic model (DDPM), we prove that their iteration complexities under exact score functions are at most the order of k/εk/\varepsilon (up to log factor), where ε\varepsilon is the precision in total variation distance and kk is some intrinsic dimension of the target distribution. We further extend these convergence guarantees to the setting in which the score functions are learned from data rather than known exactly, showing that the convergence performance degrades gracefully under suitable score estimation assumptions. We then show that these assumptions are attainable via kernel-based score estimators with finite-sample guarantees that also adapt to the low-dimensional structure. Our results apply to a broad family of target distributions without requiring smoothness or log-concavity. Our findings provide the first rigorous evidence for the adaptivity of the DDIM-type samplers to unknown low-dimensional structure, and improve over the state-of-the-art DDPM theory regarding total variation convergence.
Nov 26, 2024

ContrastiveCFG: Guiding Diffusion Sampling by Contrasting Positive and Negative Concepts

As Classifier-Free Guidance (CFG) has proven effective in conditional diffusion model sampling for improved condition alignment, many applications use a negated CFG term as a Negative Prompting (NP) to filter out unwanted features from samples. However, simply negating CFG guidance creates an inverted probability distribution, often distorting samples away from the marginal distribution. Inspired by recent advances in conditional diffusion models for inverse problems, here we present a novel method to achieve guidance toward the given condition using contrastive loss. Specifically, our guidance term aligns or repels the denoising direction based on the given condition through contrastive loss, achieving a similar guiding effect to traditional CFG for positive conditions while overcoming the limitations of existing negative guidance methods. Experimental results demonstrate that our approach effectively injects or removes the given concepts while maintaining sample quality across diverse scenarios, from simple class conditions to complex and overlapping text prompts.
Jul 5, 2024

Speed-accuracy relations for diffusion models: Wisdom from nonequilibrium thermodynamics and optimal transport

We discuss a connection between a generative model, called the diffusion model, and nonequilibrium thermodynamics for the Fokker-Planck equation, called stochastic thermodynamics. Using techniques from stochastic thermodynamics, we derive the speed-accuracy relations for diffusion models, which are inequalities that relate the accuracy of data generation to the entropy production rate. This relation can be interpreted as the speed of the diffusion dynamics in the absence of the non-conservative force. From a stochastic thermodynamic perspective, our results provide quantitative insight into how best to generate data in diffusion models. The optimal learning protocol is introduced by the geodesic of space of the 2-Wasserstein distance in optimal transport theory. We numerically illustrate the validity of the speed-accuracy relations for diffusion models with different noise schedules and different data. We numerically discuss our results for optimal and suboptimal learning protocols. We also demonstrate the applicability of our results to data generation from the real-world image datasets.
May 22, 2024

Adaptive Momentum Benefits Score-based Generative Models

Score-based Generative Models (SGMs) have achieved remarkable success in generation tasks by establishing a diffusion process that gradually perturbs real data into Gaussian noise and then learning a reverse process to reconstruct the data from noise. However, conventional denoising methods, such as Langevin dynamics and numerical stochastic differential equation solvers, rely on randomness but suffer from slow generation speeds due to the high number of score function evaluations. To address this, we propose a plug-and-play Markov adaptive momentum sampling method, inspired by the strong connection between the SGMs' sampling process and stochastic gradient descent optimization. Our approach accelerates the transformation process without introducing additional hyperparameters. Theoretically, we establish a rigorous foundation by analyzing the relationship between the distribution of samples generated by our method and the original data distribution. This provides a novel perspective on the analysis of sampling distributions. Empirically, our method demonstrates superior efficiency, generating high-quality images/graphs in significantly fewer sampling steps, achieving a speedup of 2x to 5x while maintaining or surpassing baseline performance on various image and graph generation benchmarks. These results underscore the effectiveness of adaptive momentum in enhancing the efficiency of SGMs.