Overview

Diffusion models learn a data distribution by gradually corrupting examples with noise and training a denoiser to reverse that process. Major approaches include DDPMs and score-based SDEs, latent diffusion and diffusion Transformers, flow and consistency methods, and discrete masked or continuous-time Markov diffusion; conditioning and guidance adapt them to specific tasks. They are used for images, video, audio, language, molecules, and scientific forecasting, with evaluations ranging from ImageNet FID to language perplexity and domain-specific measures of accuracy and uncertainty. A central tension is how to make sampling faster and generation more controllable without sacrificing quality, calibration, or diversity, while theory and ongoing audits address generalization, memorization, and privacy.

Latest papers

1,278 papers90 papers in the last four weeks, down 44% on the four weeks before. 0.7% of all new papers.

Oct 26, 2026

The Noise-Transmission Burden of Diffusion Network: Why Image Prediction Can Outperform Noise Prediction

Clean-image prediction (xx-prediction) and noise prediction yield equivalent population-optimal denoisers, yet can perform very differently in finite diffusion networks. For noisy input YtY_t with noise standard deviation σtσ_t, a noise predictor fε,tf_{\varepsilon,t} induces the clean-image estimate X^1=Yt−σtfε,t(Yt)\widehat X_1=Y_t-σ_t f_{\varepsilon,t}(Y_t). This reveals a noise-transmission burden: the predictor must preserve information needed to cancel the input corruption. Under Gaussian corruption of dd-dimensional data, a linear input bottleneck of rank r<dr<d imposes a denoising-error floor σt2(d−r)σ_t^2(d-r) even with unrestricted downstream computation, while the corresponding optimal risk for image prediction remains bounded. EDM attenuates this bottleneck contribution at high noise; exact preconditioning cancels it at every noise level. Without an input bottleneck, assuming exact affine recovery of the noisy patch from the model representation, we characterize how the covariance spectrum of corrections beyond linear denoising determines the width required above the patch dimension. ImageNet experiments with JiT show that measured denoising losses reflect the predicted bottleneck floor and its attenuation or cancellation through preconditioning. Further widening yields larger relative denoising gains at low noise, where corrections beyond linear denoising spread across more directions, consistent with our spectral characterization of the required model width. Together, these results provide a denoising-based account of practical generation-quality gaps across prediction targets, preconditioning choices, and model widths.
Oct 25, 2026

Cumulative Marginal Matching for One-Step Diffusion

The sampling procedures of diffusion and flow-matching models involve hundreds of denoising steps incurring substantial inference time, which has motivated a growing body of work on training schemes for few-step and one-step generation. Existing approaches primarily seek to align the generation trajectory of few-step and base models, yet their training objectives only involve discrepancies at a single point in the trajectory, encouraging only local point-level alignment. While recent methods such as Shortcut-CSL introduce cumulative objective that enforce alignment over the entire trajectory, this is suboptimal as they do not directly enforce alignment between the distributions induced by the two models throughout the generation process. We propose Cumulative Marginal Matching (CMM), a principled objective for few-step generation: by viewing flow-matching as a sequence of intermediate marginal distributions connecting the prior and data distributions, CMM directly optimizes the cumulative discrepancy between the base and few-step marginal distributions over the entire sequence. Specifically, we formulate this discrepancy using the Wasserstein-2 distance. By exploiting the pushforward representation of flow-matching, we transform an intractable cumulative Wasserstein objective into a tractable surrogate that can be optimized using samples from the intermediate marginal, which are readily generated through the conditional interpolation process of flow-matching. Furthermore, we show that the existing Shortcut-CSL objective arises as a first-order approximation of that of the CMM, establishing a theoretical connection between the two objectives. Experiments on standard benchmarks demonstrate improvement in generation quality over Shortcut-CSL and other state-of-the-art one- and few-step baselines under identical training budgets.
Oct 25, 2026

Conditional Diffusion Generation of DVL and Inertial Signals for Underwater Navigation

Autonomous underwater vehicles (AUVs) navigate without satellite positioning and rely on Doppler velocity log (DVL) and inertial sensors for motion estimation. Deep learning methods for processing these sensors require training data covering the motion the AUV encounters, commonly obtained through costly sea trials. To overcome this limitation, we present a conditional diffusion framework that generates synthetic DVL and inertial signals to replace real recordings in training. The generator employs an Elucidated diffusion model conditioned on three per-channel motion descriptors. Three configurations are evaluated: DVL beams alone, inertial channels alone, and all channels jointly from one model. Across thirteen trajectories of the real dataset, the generated signals followed the requested conditions with Pearson correlations between 0.72 and 0.997, transferred to trajectories withheld from generator training, and did not replicate any training window. A BeamsNet velocity estimator trained on generated signals alone, with no real recording retained, matched the velocity error obtained with the full real training set. Replacing up to all real windows with generated ones caused no measurable loss. Thus, this approach reduces the dependence of learned AUV navigation on sea-trial data, enables training when recordings are scarce, and facilitates developing and evaluating deep learning methods for underwater navigation.
Oct 25, 2026

Freeze What Is Settled, Cache What Is Smooth: Spatial Freezing Keeps Step Caching in Its Low-Error Regime

Instruction-based image editors built on diffusion transformers run tens of denoising steps. Training-free acceleration either reuses features across steps (step caching) or stops computing the regions an edit leaves untouched (region freezing), and the two have been combined before. We ask how a high speed-up target should be split between them. On Qwen-Image-2.1, freezing does more than remove tokens: it lowers the cost of every step, so the step cache can reach the same overall speed-up with a conservative threshold that keeps its error small. A per-step cost model fitted on development runs predicts held-out latency with a median error of 4.0% and reproduces the speed ceilings of the region methods we re-implemented (2.73×\times predicted vs. 2.70×\times measured for RegionE). Driving frozen tokens to the model's one-step x0x_0 estimate beats returning them to the reference image on every development sample, and in 90% of measured cases the composed error stays below the sum of the two individual errors. The resulting method, FreezePlan, keeps five full steps and gives the rest of the budget to TaylorSeer. On a pre-registered test split it runs 5.05×\times faster end to end at 29.8 dB PSNR to the original output, with no significant drop in a VLM judge score; TaylorSeer at 5.02×\times reaches 23.5 dB. The curve extends to 5.99×\times at 28.0 dB. Swapping in the region selectors of prior methods under our schedule lowers median PSNR by only 0.5--0.6 dB, which places the gain in the schedule and budget split rather than in the selector.
Oct 23, 2026

Information-Theoretic Scaling Laws for Generative Diffusion Training

Diffusion models trained on increasingly large datasets sampled from the same data distribution will converge to the same learned distribution. However, in practice, a model trained on finite data depends on the particular training dataset; this dependence is reflected in properties of the model's outputs, such as memorization and sample quality. To understand this dependence through an information-theoretic lens, we study model information, a computationally tractable measure of how much samples from the trained model depend on the particular training dataset. We explore its scaling behavior as model size and training dataset size vary. Specifically, we uncover two qualitatively distinct scaling regimes: (1) a nonlinear model information regime marking the transition from pure memorization to imperfect generalization, and (2) a generalization regime characterized by linear scaling of the model information and progressively improved generalization and sample quality. We then posit an information-theoretically principled scaling law to quantify these two regimes, and show that this law yields induced scaling laws for memorization and sample quality. We finally demonstrate that these induced scaling laws extrapolate well outside the fit's domain. Our results demonstrate information-theoretic principles for memorization and sample quality scaling laws in diffusion models. Our code is publicly available on GitHub.
Oct 23, 2026

Support-guided OOD Removal with Diffusion

Many real-world signals are corrupted by localized, structured perturbations whose distribution is unknown: an intermittent background noise in a music recording, a watermark stamped on a photograph, a cast shadow. We study the task of recovering a clean signal xx from an observation yy containing such a localized out-of-distribution (OOD) corruption. We propose Support-guided OOD Removal with Diffusion, SORD, an iterative algorithm to recover xx using only two ingredients: a binary map of the corruption support and a single diffusion model pretrained on clean signals. SORD does not require any paired data or any statistical model of the corruption. We evaluate SORD on natural images and human faces with spatially localized corruptions including handwritten digits, other images, watermarks, and shadows, as well as audio recordings of different musical instruments with temporally localized corruptions including other instruments and human speech. SORD is competitive with, and often outperforms, posterior-sampling inpainting given the same diffusion prior and the same mask, source-separation baselines that require a separate diffusion model for each source, and, on several benchmarks, supervised networks trained on paired clean and corrupted signals. The code is released here: https://github.com/voilalab/SORD.
Oct 23, 2026

LLaCO: Continuous Diffusion for Large Language Models with Competition-Aware Semantic Scheduling

Continuous diffusion models have emerged as a promising non-autoregressive paradigm for text generation over continuous embedding spaces. Along the continuous trajectory, the residual token identity directly shapes the denoising objective. However, the geometric trajectory misaligns with temporal progression, triggering severe temporal compression. In this paper, we show that embedding norm decay causes Euclidean distance to yield spurious overtaking, indicating that token competition must be evaluated directionally. In this space, a candidate overtakes the source token only when the effective noise magnitude surpasses their semantic separation. We formalize these pairwise dynamics as the Competition Threshold Theory (CTT). Crucially, high-dimensional overtaking requires extreme noise levels, causing standard trajectories to concentrate identity decay into a narrow window near the noisy endpoint. To resolve this, we propose Semantic Retention Scheduling (SRS) to align time allocation with the progression of identity decay, and demonstrate that existing token identity metrics are fundamentally statistical aggregations of these pairwise crossings. Building on these theoretical insights, we introduce LLaCO, a generative framework powered by competition-aware semantic scheduling. LLaCO adapts a pretrained backbone to continuous representations via an embedding directional adapter and timestep-conditioned modulation. To operate within the directional space, LLaCO employs a spherical noising process formulated as a geodesic interpolation toward a mask-biased endpoint, eliminating norm decay while preserving accurate token competition. Guided by SRS, LLaCO reallocates the denoising trajectory proportionally to the logarithmic rank decay. Instantiated as an 8B model, LLaCO outperforms previous continuous diffusion baselines and rivals discrete diffusion models of comparable scale.
Oct 22, 2026

ParaLin++: Accelerating Parallel Diffusion Integrator via Intrinsic Partially Linear Structure

Training-free few-step samplers are often described as nearly quality-lossless under conventional FID/CLIP-style evaluation. It remains unclear whether that conclusion transfers to more realistic and complex generation-quality benchmarks. We first systematize the quality metrics used to evaluate training-free sampling acceleration for diffusion and flow-matching models. We find that representative training-free step-shortening methods often remain nearly lossless under FID/CLIP-class metrics, yet can suffer substantial quality degradation when evaluated with semantic-composition, dense-instruction, and long-text-rendering benchmarks such as GenEval, DPG-Bench, and LongTextBench. Parallel sampling is a notable exception: by retaining the full-step discretization and removing only the solver's temporal dependency, it can preserve near-lossless quality while accelerating sampling. Building on this observation, we propose ParaLin++, a step-parity parallel sampler that preserves the full-step discretization while accelerating the solver iterations. Experiments across Stable Diffusion-v2, LSUN Church, FLUX.1-dev, and Ming-Flash-Omni 2.0 show that ParaLin++ preserves the full-step Euler quality profile at 2.6×2.6\times--4.8×4.8\times, 1.68×1.68\times on FLUX, and 1.21×1.21\times on Ming. Project page: https://paralinplus.github.io/.
Oct 21, 2026

DiffLTM: Continuous Diffusions Large Tabular Models for Conditional Generation

Conditional generation is a key capability to be supported by Large Tabular Models, enabling a number of applications. Accurate conditional generation has to contend with two key challenges: (i)(i) Data comprise a mixture of numerical and categorical entries, the latter often with high cardinality and heavily unbalanced proportions; (ii)(ii) Conditional generation becomes increasingly difficult for rare sub-populations. The best models in this class use a hybrid of continuous and discrete diffusions. In this paper, we design a continuous diffusion model with strong generation quality. An important element is the use of cross-entropy loss for categoricals, for which we provide a first theoretical justification. We introduce metrics that focus on rare subpopulations. Further, we generalize conditional generation techniques to enable new types of conditioning. Finally, we show that our model (trained on labor data) can capture classical insights from econometrics. Code is available at https://github.com/granica-ai/DiffLTM
Oct 21, 2026

Seeing through Denoising: Frequency-guided One-step Denoising for Infrared Small Target Detection

Infrared small target detection (IRSTD) is a challenging task due to the extremely small size of targets, low signal-to-noise ratios, and heavy background clutter in infrared imagery. While the progress of deep learning has advanced this field, existing methods primarily adopt discriminative paradigms that directly map images to target masks, overlooking the potential of generative modeling. In this paper, we propose DiffIRSTD, a novel generative framework that reformulates infrared small target detection as a background reconstruction problem. Instead of explicitly predicting target regions, we reinterpret small targets as noise-like signals and leverage a diffusion-based model to recover a clean, target-free background from a perturbed input image. The residual between the reconstructed background and the original image is then exploited to highlight candidate target regions, providing explicit cues for subsequent detection. To address the computational cost typically associated with diffusion sampling, we introduce a one-step denoising strategy that enables efficient background reconstruction while maintaining practical inference speed. Furthermore, we present a frequency-guided perception loss to preserve high-frequency background details and focus more on the reconstruction quality of target regions. More importantly, our DiffIRSTD operates as a plug-and-play framework, compatible with existing IRSTD models without increasing the computational complexity of detectors. Extensive experiments on IRSTD-1K, NUAA-SIRST, and NUDT-SIRST datasets demonstrate the superiority of our method, achieving state-of-the-art performance across three benchmarks.
Oct 21, 2026

Internal Guidance as a Sampling Method for Tilted Diffusion Models

Reward guidance methods sample an exponential tilt of a base density using a diffusion model trained with only samples from the base distribution. Existing methods use what we call external guidance: additive terms in the drift or multiplicative terms on the transition probabilities. We propose internal guidance which leaves the output of the base denoiser unmodified and instead changes the point at which it is evaluated. For discrete distributions embedded in the hypercube {−1,1}d\{-1,1\}^{d}, the denoiser of any linearly tilted density is exactly the base denoiser evaluated at a shifted and rescaled input. In practice the base denoiser is a neural network trained on noised samples of the base distribution. The required evaluation points for sampling the tilted density would be out-of-distribution for basic noising of the base distribution. We modify the training algorithm with random diagonal rescalings of the data, so that the denoiser network learns both in- and out-of-distribution evaluations of the denoiser relative to the base distribution. A single denoiser network samples binary reward tilts, class labels, and infilling with no fine-tuning and no external guidance term.
Oct 20, 2026

PEEK: One-Step-Look-Ahead Exposure Bias Correction for Diffusion Sampling

Diffusion image generators owe much of their practical success to accurate denoising networks and strong numerical samplers. However, during sampling, the denoiser predicts the next step from flawed inputs that differ substantially from real data. This train-test gap, or exposure bias, caps image generation quality (i.e., blurry images, missing details, or distorted shapes). To mitigate the problem, we propose PEEK, a training-free deterministic sampler plugin. At each eligible denoising step, PEEK first performs a provisional one-step update, evaluates the denoiser once at the look-ahead state, and interpolates the current clean prediction with this one-step-later prediction to form the next noisy state. By taking the additional step, we obtain a generally more image like prediction from the higher-signal noisy state, mitigating exposure bias. To optimize the interpolation positions, PEEK searches for an interpolation schedule with a greedy metric-driven search made efficient by the PEEK prefix and baseline noise-schedule-rebasing caches. Across pixel-space and latent-space diffusion models with unconditional, label-guided, and text-guided settings on CIFAR-10, LSUNBedroom,I mageNet, and MS-COCO,the look-ahead correction consistently improves over baselines matched by number of function evaluations (NFE), with Fréchet Inception Distance (FID) reductions up to 49.3%. The one-time per-model calibration, the only extra cost, takes at most 2.2 minutes in our experiments. Project website is available at https://dachengruoque.github.io/PEEK-Video/.
Oct 20, 2026

Policy-Mask: Budget-Aware Adaptation of Diffusion Features to Discriminative Tasks

Selecting intermediate features from a frozen diffusion model for discriminative tasks is non-trivial because the useful signal is distributed across the denoising trajectory, network depth, and channels. In practice, this requires choosing a timestep, a block, and a channel subset. Existing methods either fix these choices by hand or optimize only a few extraction hyperparameters, most often the timestep, then reuse the same rule across inputs and budgets. We introduce Policy-Mask, a budget-conditioned policy trained on the downstream objective that jointly selects all three under a prescribed channel budget. Without normalization, the channel-score contribution grows with the number of selected channels and dominates the timestep and block updates. We normalize it by subset cardinality, keeping the three policy heads on comparable scales and stabilizing joint training. At matched budgets, Policy-Mask improves over fixed extraction and feature-filter baselines across segmentation, classification, and semantic correspondence, reaching 38.9 mIoU on ADE20K, 75.4% top-1 on CIFAR-100 at 128 dimensions, and 76.3 [email protected] on SPair-71k. Component-wise ablations show that block selection materially affects downstream accuracy, and the highest recorded channel-budget score occurs at 256 retained channels, identifying a practical operating point for compact diffusion representations. We further adapt Policy-Mask to diffusion transformers, including PixArt-ΣΣ and SD3.5 MMDiT, by replacing U-Net channels with image-token hidden dimensions.
Oct 20, 2026

Continual Concept Personalization in Diffusion Models via Residual-Space Adaptation

Text-to-image personalization aims to equip pretrained diffusion models with the ability to learn user-specific concepts from only a few reference images. Existing methods, however, typically train each concept independently or assume that all concepts are available simultaneously. In practical applications, concepts often arrive sequentially, requiring models to continually learn new subjects, objects, or styles while minimizing interference with previously acquired concepts. We address this continual concept personalization setting by extending residual-space adaptation to the stochastic adaptation process of pretrained diffusion models. Our work achieves the highest concept fidelity among competing continual personalization methods while maintaining near-zero forgetting. Beyond these empirical gains, the learned residual-space dynamics reveal where adaptation capacity is concentrated within the diffusion model. Restricting adaptation to the identified layers by our method preserves nearly the same performance while reducing the number of trainable parameters by up to 72%, demonstrating that residual-space dynamics can identify the layers that contribute most to continual concept personalization and directly guide more parameter-efficient adaptation.
Oct 20, 2026

Binary Discrete Diffusion: Candidate-Set Diffusion for Logic and Planning Tasks

Discrete diffusion models iteratively denoise a noisy sequence of categorical data such as text, enabling any-order generation with self-correction and in-filling capabilities. This makes them promising candidates for reasoning, planning, and constrained-generation tasks like coding and math. Yet masked and uniform diffusion erase a token's identity in a single corruption event, so their noisy states cannot explicitly represent partially resolved uncertainty in which several alternatives remain plausible. We close this gap with Binary Discrete Diffusion, a family of continuous-time discrete-diffusion processes whose noisy state is a set of candidate tokens per position, represented as a binary membership vector. Binary diffusion changes candidate sets through additions and removals, allowing information to be lost and recovered gradually while the diffusion state itself represents partial uncertainty. On Sudoku-Extreme, Boxoban, and graph coloring, the symmetric formulation of binary diffusion consistently outperforms strong, well-tuned masked and uniform baselines under matched architectures and controlled training, sampling, and tuning budgets, with its largest gains on the hardest instances, while remaining competitive on language modeling (Text8, TinyStories, OpenWebText). These results support candidate sets as an effective intermediate representation for iterative reasoning and show that the corruption kernel can materially change the computation learned by the reverse model.
Oct 20, 2026

MoDDiT: Simple Token Routing in Diffusion Transformers

Real-time video generation, interactive world models, and on-device image synthesis are almost a reality. Diffusion Transformers (DiT) achieve excellent sample quality, shifting the bottleneck to the throughput of the models themselves. Natural images are inherently sparse, as large portions of an image are often background, contribute little to the scene content, and are easy to denoise. DiTs fail to exploit this sparsity, as they process image tokens statically and uniformly. We propose MoDDiT, a simple architecture modification that learns to route tokens through both MLP and self-attention operations. By routing tokens through both blocks independently, MoDDiT allows for dynamic compute allocation in MLPs that varies across tokens, images, and denoising timesteps. Using MoDDiT to train image generation models from scratch significantly reduces computational costs of training and inference, processing as few as 12.5% of the tokens. By increasing sparsity and model depth, MoDDiT reduces costs and improves generation performance over unrouted baselines. The learned allocation is interpretable, revealing how computation varies with visual content and noise level. We also use MoDDiT to train video generation models, where allocation correlates with motion, and to distill a dense bidirectional video model into a sparse autoregressive student.
Oct 20, 2026

Detection of Memorization in Diffusion Models Through Thermodynamic Singularity

In diffusion models, proper distributional learning entails reversing a diffusion process over underlying density functions, whereas a key failure is memorization, in which the reverse process instead learns diffusion from individual training samples. We provide an analytical characterization of such memorized reverse processes, along with a criterion that detects the associated thermodynamic singularity. The proposed criterion reliably distinguishes between properly learned and memorizing networks across a range of neural network models designed to promote or mitigate memorization.
Oct 19, 2026

Diffusion Score Estimation with Shallow Neural Networks: Regularity, Approximation, and Finite-Sample Rates

We study the sample complexity of estimating Ornstein--Uhlenbeck diffusion scores in Rd\mathbb R^d using denoising score matching (DSM) with finite-width shallow ReLU networks, and its implications for reverse-process sampling. Under weak differentiability and polynomial-growth conditions on normalized derivatives of the initial density, we combine complementary estimates from the Gaussian transition kernel and the initial density to obtain explicit Sobolev bounds that are uniform in time on each fixed ball. Combining these bounds with existing shallow-network approximation theory gives explicit control of the approximation error and a sufficient path-norm budget. We then use truncation and localized Rademacher complexity to establish high-probability risk bounds for path-norm-constrained empirical DSM minimizers with unbounded inputs and labels. Under additional bounded-density and sub-Gaussian assumptions, suitable choices of network width and path-norm budget yield squared L2(Pt)L^2(P_t) score risk O~(n−(d+3)/(2d+3))\widetilde O(n^{-(d+3)/(2d+3)}) at each fixed time t>0t>0, where nn is the number of independent training samples and PtP_t is the forward marginal. The nonasymptotic bounds retain explicit dependence on diffusion time and the regularity and tail parameters of the initial distribution. Aggregating the score errors over a time grid and applying existing sampling theory together with an early-stopping bound yields a high-probability total-variation error O~(n−(d+3)/(4d+6))\widetilde O(n^{-(d+3)/(4d+6)}) between the generated and original data distributions under suitable sampling parameters. Here O~\widetilde O suppresses logarithmic factors for fixed dimension, confidence level, and initial distribution.
Oct 19, 2026

RECAP: Reciprocal Sparse Attention and Hierarchical Caching for Efficient Video Diffusion

Diffusion transformers generate high-quality videos but remain computationally expensive. Sparse attention and temporal caching reduce computation within and across denoising steps, respectively, offering a natural composition. However, predictions produced with sparse attention become sources for temporal reuse, coupling their errors along the sampling trajectory. We observe that improving the accuracy of these sources recovers fidelity without changing the reuse schedule, and that sparsification errors remain directionally aligned across nearby steps, suggesting that correction information can be shared rather than recomputed. These observations motivate RECAP, a training-free method that couples sparse attention with hierarchical caching through reciprocal correction, adaptively coordinated by online error feedback. Cached dense refinement corrects omitted interactions in current sparse estimates, while fresh sparse statistics track input changes and regulate the aging correction. The resulting predictions then renew the sources for whole-step prediction reuse, allowing prediction history to advance between dense refreshes. We evaluate RECAP on multiple video diffusion models against sparse-attention methods, caching methods, and their direct compositions. On HunyuanVideo at 720p 5-second generation, RECAP achieves a 3.05x speedup and outperforms all tested direct compositions in both fidelity and inference speed.
Oct 19, 2026

SWIFT: Self-guided Weak-to-strong dIFfusion models via Target extrapolation

Training high-quality diffusion models is computationally expensive, and inference-time guidance often adds further sampling cost. In this work, we introduce Swift, a simple yet effective self-guided training method that uses a frozen weak denoiser to improve a stronger model without adding inference overhead. For each noisy input, Swift extrapolates the clean regression target away from the weak denoiser's prediction, amplifying components that remain underfit. The weak guide can be an early training checkpoint, requires no backpropagation during main-model training, and is discarded afterward. The resulting model retains the original sampler and supports both conditional and unconditional generation without label dropout or an unconditional branch. We explain this mechanism through a spectral analysis of finite-time population gradient flow in an RKHS. With an early-stopped weak guide, we prove that a nonempty range of extrapolation strengths yields strictly lower posterior-mean error than matched standard training, despite shifting the regression target away from the posterior mean. Experiments on CIFAR-10 and ImageNet-256 demonstrate improvements across model scales and representations. In one ImageNet-256 configuration, Swift surpasses the baseline's best FID using roughly half the total training compute, including guide preparation and training overhead. On RAEv2, it achieves FID 1.15 without inference-time guidance and 0.99 with guidance. These results establish target extrapolation as an effective approach to weak-to-strong diffusion training with efficient single-model inference.
Oct 19, 2026

Hihaze: One-Step Diffusion for High-Resolution Real-World Image Dehazing

Diffusion models have recently achieved encouraging results in real-world image dehazing. However, restoring high-resolution images remains challenging due to spatially non-uniform haze and inconsistent predictions across local patches. To address these issues, we propose Hihaze, a one-step diffusion framework for high-resolution real-world dehazing. Hihaze adopts two haze-aware modulation modules: attentive skip compensation (ASC) and a spatial degradation adapter (SDA). ASC selectively transfers reliable encoder information to the variational autoencoder (VAE) decoder, preserving structural details while suppressing haze-degraded features. SDA extracts multi-scale spatial degradation representations and injects them into the U-Net to guide restoration according to local haze variations. Furthermore, we design a training-free global-guided patch fusion (GPF) method. It uses a global prediction to align patch appearance and downweight unreliable predictions, followed by structure-guided detail compensation. Notably, this training-free design also makes GPF a plug-and-play module for other dehazing models. Experiments across four real-world benchmarks show that Hihaze surpasses state-of-the-art approaches in both visual quality and inference speed. Specifically, it delivers PSNR improvements of 3.62dB on Dense-Haze and 3.69dB on NH-HAZE compared with the strongest baselines. Code: https://github.com/zhengchen1999/Hihaze.
Oct 19, 2026

PoSTCache: From Forecasting to Postdictive Smoothing for Revisable Diffusion Caching

Diffusion and flow matching models achieve impressive visual generation quality but remain computationally expensive due to iterative sampling. Caching-based acceleration provides a practical training-free solution by reusing or predicting intermediate features. However, existing methods largely enforce a forward-only information flow: approximations produced for skipped model evaluations are immediately committed to the sampling trajectory and cannot benefit from feedback generated by later full-computation steps. We propose PoSTCache (Postdictive Smoothing Trajectory Cache), a training-free and plug-and-play framework that reformulates diffusion caching as revisable trajectory inference. PoSTCache first uses an existing caching strategy to construct forward proposals for skipped steps. At a later full-computation step, the discrepancy between the refreshed prediction and its forward proposal yields a correction signal, which PoSTCache propagates backward to refine and reintegrate the preceding predictions. Viewing full-computation steps as sources of corrective feedback also changes how they should be selected. We therefore introduce the Posterior Variance Reduction (PVR) strategy, which estimates cross-timestep dependencies from offline denoising trajectories and selects anchors according to their expected reduction in surrogate trajectory uncertainty rather than local approximation error alone. Experiments on representative image and video generation models show that PoSTCache consistently improves the quality--efficiency trade-off over representative caching methods.
Oct 18, 2026

T(R,O) Grasp++: Cross-Embodiment Dexterous Grasping in Cluttered Environments at Scale

Dexterous grasping in cluttered scenes remains a central challenge in robotic manipulation, as it requires high-dimensional hand coordination under complex object-environment constraints. To address this problem, we introduce one of the largest simulation benchmarks to date, comprising around 0.5B grasp annotations across eight dexterous hands and approximately 50K objects. Furthermore, we propose T(R,O) Grasp++, a scalable diffusion-based framework for cross-embodiment dexterous grasp synthesis in cluttered environments. Extensive experiments demonstrate that our model significantly outperforms existing baselines, achieving an average success rate of 88% across eight dexterous embodiments. Meanwhile, our pipeline achieves only 0.3s inference latency while improving inference sample efficiency by over 10 times.
Oct 18, 2026

PolarVis: A Dataset and Diffusion Framework for RGB-to-Polarization Perception in Adverse Driving Scenes

Reflections, glare, transparent structures, and material-dependent appearance variations can introduce ambiguity into RGB-based visual perception in vehicle-relevant environments. Polarization provides complementary reflection- and material-sensitive information, but its acquisition requires dedicated sensing hardware. This paper presents RGB-P Wild, a real-world dataset containing 10,000 paired RGB and four-angle linear-polarization image sets, together with PolarVis, a diffusion-based framework for recovering polarization representations from a single RGB image. PolarVis combines hierarchical RGB feature extraction, diffusion-based feature refinement, four angle-specific decoders, and multi-scale attention recalibration to reconstruct polarization intensity images at 0∘0^\circ, 45∘45^\circ, 90∘90^\circ, and 135∘135^\circ. On RGB-P Wild, PolarVis achieves a normalized-domain MSE of 0.0465, a PSNR of 27.83 dB, and an SSIM of 0.8597. Compared with the released PolarAnything model under the adopted evaluation protocol, PolarVis obtains an AoLP error of 25.66∘25.66^\circ and a DoLP MAE of 0.0404, compared with 46.03∘46.03^\circ and 0.1084, respectively. Cross-dataset evaluation on vehicle-mounted imagery further examines transfer beyond the training domain. In downstream evaluation, a frozen PCDNet achieves AP50/AP50:95 of 86.3%/59.1% on 990 RGBP-Car test images using PolarVis-generated polarization, compared with 87.0%/59.8% using measured polarization. These results indicate that RGB-derived polarization retains useful polarimetric information for the evaluated perception tasks, while remaining distinct from calibrated physical polarization measurements.
Oct 18, 2026

LiftDiff: Guided Discrete Diffusion for Quantum Code Construction

The realization of fault-tolerant quantum computation relies on Quantum Error-Correcting Codes (QECC). Of particular interest are codes that combine a high encoding rate with sufficient distance. Existing algebraic constructions, such as lifted product (LP) codes, build a large quantum low-density parity-check (qLDPC) code from a classical base matrix. In this work we introduce LiftDiff, a generative approach to the discovery of QECC. A discrete denoising diffusion model is trained to generate base matrices over the group algebra F2[Zℓ]\mathbb{F}_2[\mathbb{Z}_\ell], which are then expanded into LP codes. The reverse diffusion process is guided by an energy function that evaluates quantum code properties, including girth, degree, sparsity and decoding performance, steering the sampler toward matrices that yield LP codes with favorable parameters. Our method discovers codes with parameters such as [[169,13,11]][[169,13,11]], [[143,15,9]][[143,15,9]], and [[260,24,10]][[260,24,10]], including a [[247,19,14]][[247,19,14]] code that achieves the highest kd2/n{kd^2}/{n} among the baselines. Moreover, several discovered codes outperform the Gross code across a broad range of the tested physical error rates under both code-capacity and circuit-level noise.
Oct 16, 2026

Rethinking the design space of Vanilla Pixel Diffusion Transformers

Pixel-space diffusion models have recently re-emerged as a compelling alternative to latent diffusion, avoiding the lossy latent bottleneck and separately trained autoencoder. JiT shows that a vanilla diffusion transformer can generate directly in pixel space by replacing the commonly used velocity prediction (u\boldsymbol{u}-prediction) with clean image prediction (x0\boldsymbol{x_0}-prediction). In this work, we revisit this choice and introduce MFlow, an adaptive parameterization that learns to interpolate between x0\boldsymbol{x_0}- and u\boldsymbol{u}-prediction across principal directions of the data, learned end-to-end with diffusion training. We find that the preferred parameterization varies systematically with the data representation and model capacity -- notably, increasing model capacity shifts preference toward velocity prediction. We evaluate MFlow across ImageNet, UCF-101 video, and large-scale text-to-image generation, demonstrating competitive or improved generative performance across diverse settings. In addition, we find that vanilla pixel diffusion transformers, irrespective of parameterization, retain a persistent high-frequency deficit relative to real images. To address this, we introduce ConvRefiner, a lightweight convolutional module trained post-hoc with an adversarial loss while keeping the diffusion backbone frozen. With only a small amount of additional training, ConvRefiner restores the high-frequency statistics of generated images while preserving FID. Project webpage: https://soumik-kanad.github.io/mflow-diffusion/.
Oct 16, 2026

Information Estimation with Discrete Diffusion

Information-theoretic measures, such as Mutual Information (MI), play a crucial role in understanding non-linear relationships between random variables and are widely used across scientific disciplines. Yet, their use on real-world discrete data remains challenging. Existing methods typically rely on embedding discrete data into a continuous space and apply neural estimators originally designed for continuous distributions. This process requires careful engineering for both the embedding model and estimator architecture, but suffers from issues related to high data dimensionality. In this work, we introduce InfoSEDD, a discrete diffusion-based approach that bridges information-theoretic estimation and generative modeling such that they can be used to compute Kullback-Leibler divergences. Backed by Continuous Time Markov Chains theory principles, the design of InfoSEDD is lightweight and scalable and allows seamless integration with pretrained models. We showcase the versatility of our approach through applications on motif discovery in genetic promoter data, semantic-aware model selection in text summarization, and entropy estimation in Ising models. Finally, we construct consistency tests on real-world textual and genomics data. Our experiments demonstrate that InfoSEDD outperforms alternatives that rely on the "embedding trick". Our results position InfoSEDD as a robust and scalable tool for information-theoretic analysis of discrete data.
Oct 16, 2026

Data-Efficient Classifier Pretraining Through Diffusion Models

Machine learning with limited data is susceptible to overfitting, especially with modern deep neural networks. Prior to the introduction of AlexNet, reconstruction-based pretraining with auto-encoder and denoising auto-encoder were found to be beneficial, partially due to the limited size of the training sets at the time. Motivated by these traditional approaches, we propose DiffPT, a simple diffusion modeling approach to pretrain a given classifier backbone for limited domain-specific training data. We experiment on time-series data, standard natural-image benchmarks, and medical imaging datasets, and contrast the diffusion approach to a multitude of modern and traditional un/self-supervised methods. While contrastive pretraining performs better on CIFAR-10 and Tiny ImageNet in low-data regimes, DiffPT is particularly effective on medical imaging and time-series tasks, where it outperforms self-supervised baselines across a range of low-data regimes. Our results suggest that DiffPT is a simple and effective approach for data-efficient classifier pretraining on different modalities.
Oct 16, 2026

Large Diffusion Models Need More Than EMA

Large-scale diffusion models are commonly trained with a constant learning rate (LR) and exponential moving average (EMA), despite the strong performance of LR decay and Schedule-Free optimization in training large language models. We study whether EMA is uniquely suited to diffusion training, or whether stronger alternatives have been underexplored. Our theory shows that averaging a constant-step trajectory has an inherent optimization limitation, whereas LR decay and Schedule-Free optimization admit vanishing stationarity guarantees under standard nonconvex assumptions. Empirically, LR decay and EMA provide complementary benefits: combining decay with EMA further improves the standard baseline at most tested starting points, although the gain remains sensitive to when decay begins. Schedule-Free avoids this timing requirement and consistently outperforms EMA with a constant LR across training stages, achieving both lower fixed-evaluation objectives and better generation quality. Our analysis further shows that all three strategies substantially reduce parameter variability, yet stronger stabilization is not always better: EMA with a constant LR retains comparatively higher parameter variability, while aggressive decay combined with EMA achieves the lowest checkpoint variability but also weakens continued optimization. Schedule-Free substantially reduces parameter variability while preserving more late-stage progress, providing a better balance between parameter stability and continued optimization. Finally, the benefits of Schedule-Free extend beyond pretraining to ReFL and generator-side DMD, demonstrating its transferability across distinct diffusion optimization settings. Overall, our results suggest that large diffusion models need more than parameter averaging alone: effective optimization requires balancing variance reduction with continued optimization.
Oct 15, 2026

HilbertGen-3D:Hilbert--Multifractal Conditioning for Topology-Aware 3D Generation

High-fidelity 3D generative modeling underpins game asset creation, medical imaging, and VR/AR environmental design, where geometric structure directly influences perception and decision-making. Diffusion models represent a powerful paradigm for point cloud generation. However, existing approaches rely heavily on local voxel or patch features and often miss global structural organization, and much less work has studied how global conditioning can be used to improve generation quality. In this paper, we introduce HilbertGen-3D, a novel topology-aware conditional diffusion framework based on Hilbert-Multifractal Conditioning (HMC). HMC computes multi-scale voxel measures, serializes them via a 3D Hilbert curve to expose structured spatial dependencies, and estimates the generalized fractal dimensions from scale-dependent partition functions. We encode these signals as global embeddings and multi-scale tokens, and inject them into a diffusion denoiser through an efficient bottleneck fusion module. Across the Chair, Airplane, and Car categories of ShapeNet, our two conditioning variants reduce the absolute deviation of 1-Nearest Neighbor Accuracy (1-NNA) from 50% by up to 8.90 percentage points and increase coverage by up to 5.88 percentage points relative to TopoDiT-3D. This shows that combining local density cues with global Hilbert-multifractal structure provides a simple and effective recipe for high-fidelity and diverse 3D generation.