Diffusion Model Inference Acceleration

Follow this topic Ask about this topic

Momentum

40 papers in the last four weeks, up 100% on the four weeks before. 0.4% of all new papers.

Jul 13Week of Sep 28

Latest papers 335

Oct 8, 2026cs.CV

BudgetPix: Compute-Adaptive Tokenization for Pixel-Space Image Diffusion

Most image generation models rely on uniform tokenization, allocating the exact same computational budget to equally-sized image patches. This static paradigm cannot adapt to different resource constraints at inference time, and yields suboptimal quality-cost tradeoff by devoting the same effort to both plain backgrounds and intricate details. We propose BudgetPix, an adaptive tokenization framework that dynamically allocates compute based on visual complexity and spatial layout, enabling flexible computational budgeting at inference time. BudgetPix comprises three key components: (1) an adaptive encoder that maps a fixed-size image to a variable-length token sequence using an entropy-guided quadtree alongside a multi-scale patch embedder; (2) a scale-aware decoder reconstructs fixed-resolution images from multi-scale token sets; and (3) a flexible training and sampling schedule that enables pixel-space denoisers to operate across variable token counts. BudgetPix seamlessly integrates with existing pixel-space diffusion architectures, enabling a single checkpoint to be operated at a wide range of compute budgets. Evaluated on text-to-image generation, BudgetPix matches the fidelity of MiniT2I-L at

512^2

and PixelDiT at

1024^2

using just 25% of the original compute budget. In class-conditional generation using a MeanFlow backbone, BudgetPix requires merely 60% of the full compute budget to produce images with near-zero quality degradation, observing a marginal 0.8-point increase in FID. Comprehensive assessments by human and VLM judges confirm that BudgetPix establishes a significantly improved quality-efficiency tradeoff over prior budget-adaptive baselines. More details are available at our project page: https://karaozgur.com/BudgetPix

Ozgur Kara, Yujia Chen, Daniel Watson +4

University of Illinois Urbana-Champaign · Google

Image Tokenization Diffusion Model Inference Acceleration Pixel-Space Diffusion Models T2I Generation

Oct 8, 2026cs.SD

DiffuPlex: Accelerating Full-Duplex Spoken Dialog Models via Rolling Masked Diffusion

Recent full-duplex spoken dialog models enable simultaneous listening and speaking, but fine-grained models still advance their backbone autoregressively at every interaction frame. We introduce DiffuPlex, a rolling masked diffusion framework that reduces this sequential computation by predicting multiple future user and assistant frames in a single backbone wake. DiffuPlex consumes only a confident prefix of each predicted future while interaction continues at the original frame rate. As user speech arrives, it checks the corresponding user predictions and, when the interaction diverges, preserves already played assistant content while revising only the unplayed future. We consider two inference policies over the same predictor: DiffuPlex-LISTEN consumes multiple future frames when they predict assistant silence, whereas DiffuPlex-SPEAK can also consume predicted assistant speech. Across full-duplex interaction and spoken-language evaluations, DiffuPlex substantially reduces sequential backbone computation while largely preserving interaction behavior and general capability. DiffuPlex-LISTEN and DiffuPlex-SPEAK achieve

1.46\times

and

1.59\times

deployment-path wall-clock speedups and

1.61\times

and

1.80\times

Core LM speedups, with all measured backbone invocations completing within the 80ms interaction interval. Human evaluation shows that LISTEN preserves speech naturalness and conversational quality, while SPEAK retains conversational quality with some degradation in speech naturalness.

Heeseung Kim

Department of Artificial Intelligence University of Seoul Seoul, Republic of Korea

Full-Duplex Spoken Dialogue Systems Diffusion Model Inference Acceleration Masked Diffusion Models

Oct 8, 2026cs.LG

Just Weather Scoring: Efficient End-to-end Nowcasting with Distributional Diffusion

Generative diffusion models are well-suited for probabilistic precipitation nowcasting, but existing approaches often rely on separately trained compression or deterministic forecasting components and remain costly at inference due to iterative denoising. We introduce Just Weather Scoring (JWS), a single-stage, end-to-end diffusion model which addresses both issues by forecasting directly in radar space and enabling few-step generation. Radar-space modeling greatly simplifies training and inference and eliminates uncertainty arising from lossy compression. JWS combines Masked Asynchronous Diffusion, a timestep-sampling scheme that preserves clean context while adapting diffusion training to high-dimensional spatio-temporal data, with a simple scoring-rule objective that aligns training with probabilistic forecasting and unlocks few-step generation. On the SEVIR and MeteoNet benchmarks, JWS achieves state-of-the-art probabilistic forecasting performance at reduced training and inference cost. Even our smallest model remains competitive using substantially fewer parameters and more than 17x faster inference.

Jannik Wiese, Johannes Schusterbauer, Tommaso Martorella +1

Precipitation Nowcasting Probabilistic Forecasting Spatiotemporal Forecasting Diffusion Models Diffusion Model Inference Acceleration Masked Diffusion Models

Oct 8, 2026cs.CV

Streaming-Aware Diffusion for Real-Time Video Super-Resolution via Cross-Step Attention

Real-time video super-resolution requires high spatio-temporal fidelity under strict latency constraints, challenging diffusion models due to their iterative sampling cost and limited temporal coordination. We propose a streaming-aware framework that adapts pretrained single-image latent diffusion models for efficient video super-resolution (VSR) by exploiting the sequential structure of video streams. Our Cross-Step Attention mechanism reuses intermediate denoising features across adjacent frames and diffusion steps, enabling temporal information exchange without explicit temporal modeling. We further introduce Trajectory-Coupled Diffusion Scheduling, which aligns adjacent diffusion states and provides cleaner intermediate representations for cross-step conditioning, improving temporal coherence. These components are integrated into a streaming inference pipeline that incrementally propagates latent states across frames, reducing the effective computational complexity from

O(N \cdot S)

to

O(N + S)

for

N

frames and

S

diffusion steps. Experiments on REDS4 and YouHQ40-Test demonstrate improved perceptual quality and temporal realism while maintaining frame-wise stability. Our method achieves over 40 FPS at

512 \times 512

resolution after cold start, enabling real-time VSR without explicit temporal modeling.

Harris Partaourides, Sotirios Chatzis

Ethical AI Novelties Limassol, Cyprus · Cyprus University of Technology Limassol, Cyprus

Video Super-Resolution Diffusion Model Inference Acceleration Temporal Consistency in Video Generation Diffusion Timestep Scheduling

Oct 8, 2026cs.CV

iCATS: Fast Video Generation via Interaction-Aware Sparse Attention and Timestep-Adaptive Sparsity

Training-free sparse attention offers a practical acceleration solution to Diffusion Transformers (DiTs) via reducing computations without fine-tuning. It typically involves estimating the importance of query-key regions and deriving sparse masks to compute only the important candidates, which inevitably introduces approximation errors that may degrade generation quality. To better balance the efficiency-quality trade-off, we propose iCATS, integrating improved importance estimation and sparse mask construction with an efficient hardware execution strategy. Specifically, for importance estimation, unlike previous works that perform independent clustering over query and key tokens based on feature similarity to estimate attention scores, iCATS demonstrates that clustering based on query-key dot-product interactions is more accurate and further reformulates this objective as a simple quadratic form for low-cost computation. For sparse mask construction, instead of using a fixed top-p rule, we observe that tolerance to sparse approximation errors varies across denoising timesteps and therefore introduce an SNR-guided sparsity schedule to adjust sparsity dynamically, leading to higher accuracy. Finally, for hardware execution, we devise a tail-merging strategy to reduce padding overhead caused by irregular cluster sizes, improving GPU kernel utilization. Extensive experiments show that iCATS achieves

2.03\times

acceleration with 31.017 dB PSNR on HunyuanVideo-T2V-13B and

1.55\times

acceleration with 29.301 dB PSNR on Wan2.1-T2V-14B, delivering a state-of-the-art efficiency-quality trade-off.

Chengfeng Han, Baole Ai, Xianlu Bian +4

Hangzhou Normal University · T-Head Semiconductor Co., Ltd. · Alibaba Group

Diffusion Model Inference Acceleration Video Diffusion Models Diffusion Transformer Diffusion Models

Oct 8, 2026cs.LG

The Lattice of Transition Laws

Diffusion and autoregression (AR) have long been seen as different categories of generative models, with diffusion specialising in continuous fields and AR specialising in discrete tokens. Recent work seeks to combine the advantages of the two models, and each hybrid fixes its decoding schedule by design. In this paper, we ask whether the performance of decoding schedules of one model can be predicted before decoding at a fixed number of steps. We describe diffusion, AR, and models in between as paths on one corruption lattice, and define the cost of a schedule as the dependence its parallel steps discard. The cost shows that the fewest steps of a zero-cost schedule are set by the geometry of the data, in the same way for tokens and for continuous fields. In particular, for data that are Markov on a graph and dependent along its paths, the fewest steps equal the graph's treedepth, which is logarithmic in the length of a sequence and linear in the side length of a grid. With fewer steps than the treedepth, every schedule pays a positive cost, whose ranking we predict before decoding with a kernel of pairwise dependence estimated from pretrained weights. Across text generation, image generation, and video generation, we verify most of the predictions about the rankings of different schedules under different metrics and benchmarks. This work therefore provides a design principle for decoding for future AR models, diffusion models, and anything in between. Our code is available at https://github.com/TSUITUENYUE/The-Lattice-of-Transition-Laws.

T. Y. Tsui, Jiatao Gu, Lingjie Liu

University of Pennsylvania

Parallel Decoding Diffusion Sampling Autoregressive Generation Diffusion Model Inference Acceleration

Oct 7, 2026cs.CV

Omni-Diffusion-Distill: Few-Step Distillation of Unified Multimodal Diffusion Large Language Models

Unified multimodal diffusion large language models (dLLMs) offer a single architecture for both image generation and multimodal understanding, but their iterative decoding requires tens to hundreds of forward passes. Existing few-step distillation methods largely focus on either image generation or text generation, making it unclear how to compress a fully discrete multimodal dLLM into a single efficient student while preserving both generation and understanding. We introduce Omni-Diffusion-Distill, a unified two-stage distillation framework that retains strong generation and understanding capabilities while substantially reducing the inference cost of a unified multimodal dLLM. Omni-Diffusion-Distill aligns the distillation of both generation and understanding, for both images and text, in the discrete token space. In the first stage, the student is trained to skip decoding steps by replaying cached teacher trajectories, and in the second stage the student is refined on intermediate states along its own rollouts. We further remedy two sources of degradation in unified distillation with a pairwise collision penalty that reduces repetition under parallel text decoding, and entropy-matched guidance that prevents entropy collapse caused by fitting the sharpened teacher distribution in image generation. Omni-Diffusion-Distill achieves state-of-the-art trade-offs between decoding efficiency and generation and understanding performance for multimodal dLLMs, reducing image generation from 128 to 8 decoding steps and multimodal understanding from 512 to 64, giving 18.2x and 21.2x wall-clock speedups. Under these budgets, it scores 0.828 on GenEval and 83.0 on DPG-Bench for text-to-image generation, while reaching GPT judge scores of 20.0 on MM-Vet and 57.2 on COCO captioning (twice the teacher's 28.4 at the same steps) for multimodal understanding.

Hong Huang, Chenhongyi Yang, Junzhe Sun +3

Simon Fraser University · Meta

Diffusion Language Model Distillation Diffusion Model Inference Acceleration Multimodal Diffusion Models T2I Diffusion Models Multimodal Understanding

Oct 7, 2026cs.CV

GRACE: Generation-aware latent compression for efficient video generation

Highly compressed video autoencoders offer an effective way to accelerate video diffusion models, as the Diffusion Transformer (DiT) operates on far fewer tokens. However, such autoencoders are challenging to train, since a higher compression ratio degrades reconstruction quality and recovering it requires more channels, which is known to slow the convergence of the DiT. The compressed latent also differs from the one the DiT was trained on, so the pretrained DiT must be either retrained from scratch or adapted at considerable cost. Compressing the autoencoder the DiT was trained with appears to preserve compatibility, yet optimizing it for reconstruction alone still shifts the latent away from the distribution the DiT has learned. To address this, we propose Generation-Aware Latent Compression for Efficient Video Generation (GRACE), a two-stage framework that compresses a pretrained video autoencoder while keeping it compatible with the pretrained DiT. Specifically, we keep a frozen base latent from the pretrained encoder and learn a residual latent for the information lost under stronger compression, while aligning the compressed latent with the pretrained latent in the feature space of the frozen DiT so that the autoencoder is optimized for generation. We then adapt the DiT with lightweight fine-tuning and asymmetric denoising, where the base is denoised ahead of the residual. GRACE reduces the token count of Wan2.1-I2V-14B by 8x and its latency by 11.1x at 480x832x81, while matching the generation quality of the pretrained pipeline before compression on VBench.

Jiyoung Kim, Paul Hyunbin Cho, Jisu Nam +5

KAIST AI · Kakao Corp.

Diffusion Model Inference Acceleration Video Diffusion Models Video Token Compression Diffusion Transformer

Oct 7, 2026cs.CV

MORCA: Offline-to-Online Reinforcement Learning for Adaptive Cache Reuse in Video Diffusion Acceleration

Diffusion Transformers (DiTs) achieve remarkable performance in video synthesis, but their iterative denoising process suffers from high inference latency. To address this, caching has emerged as an effective acceleration strategy by capitalizing on inter-step redundancy during denoising. Existing dynamic caching methods typically estimate the error that cache reuse would introduce at each denoising step (step error) to guide cache decisions, whereas our concern is how much quality loss cache reuse would cause in the final generated video (terminal error). We show that step error does not directly correspond to terminal error and that latent information helps capture their relationship, thereby informing cache decisions. Moreover, existing threshold-based methods cannot provide precise speedup control, making it difficult to meet practical requirements for user-specified acceleration targets. To address these limitations, we introduce MORCA, a cache scheduling framework trained through offline-to-online reinforcement learning to make latent-aware reuse/recompute decisions under user-specified acceleration targets. Extensive experiments on different video generation models across multiple target acceleration ratios demonstrate that MORCA achieves better generation fidelity than state-of-the-art caching methods under comparable computational budgets. Code is available at https://github.com/x10ngyx/MORCA.

Yuxiang Xiong, Ruiyan Wang, Wenqiang Wang +5

Shanghai Jiao Tong University · Alibaba Cloud Computing · Alibaba Token Hub, Alibaba Group

Diffusion Model Caching Diffusion Model Inference Acceleration Video Diffusion Models

Oct 7, 2026cs.LG

Koopman Observers for Diffusion Acceleration: Correcting Feature Forecasts with Shallow Measurements

Feature caching accelerates diffusion sampling by replacing expensive network evaluations with predictions from previously computed activations. However, forecasts based only on past features cannot directly incorporate changes in the current denoising state. We investigate whether inexpensive, freshly computed features can serve as observations for correcting these predictions. We introduce an observation-corrected Koopman framework for accelerating frozen diffusion models. Using calibration trajectories, we identify finite-dimensional, time-dependent Koopman approximations that jointly describe the increments of shallow and deep network features. During accelerated sampling, these operators predict the evolution of expensive deep features, while innovations in the observed shallow features correct the predicted state. Periodic full evaluations refresh the observer, and all generative-model parameters remain unchanged. This formulation enables controlled comparisons of temporal prediction and observation correction. Across three 10,000-image runs per dataset, our method reduces paired Inception-feature MSE by

19.9\%

on CIFAR-10 and

11.9\%

on a ten-class ImageNet subset relative to channelwise affine prediction under the same four-partial-step schedule. Matched ablations attribute additional reductions of

4.54\%

and

4.67\%

to observation correction. The observer achieves

1.89\times

and

1.85\times

measured speedups over DDIM-50, supporting improved reference-sampler fidelity without retraining the denoiser.

Hanru Bai, Yuanchao Xu, Fengyi Li

ETH Zurich · Max Planck Institute for Intelligent Systems · Kyoto University +1

Diffusion Model Inference Acceleration Diffusion Model Caching Koopman Operator Learning

Oct 7, 2026cs.CV

Pooling Representation Autoencoders for Efficient Diffusion

Representation Autoencoders (RAEs) generate images from pre-trained visual fea- tures, but their dense token grids make generative modeling expensive. Motivated by local feature correlations, we introduce PoolDINO, a learned affine pooling operator that merges neighboring tokens. Training the pooling operator jointly with the RGB decoder preserves the standard two-stage RAE procedure without a separate feature auto-encoder. On ImageNet-256, 4x token compression retains comparable generation quality under internal guidance, while 16x compression trades some quality for greater efficiency. At a fixed budget of 100 sampling steps, latent-sampling throughput increases by 3.7x and 9.0x, respectively, relative to the unpooled baseline. Classification and dense prediction evaluations show that comparable guided generation quality can coexist with weaker performance on other tasks.

Ramón Calvo-González, Youssef Saied, François Fleuret

University of Geneva · University of Geneva and Meta

Autoencoders Diffusion-Based Image Generation Diffusion Model Inference Acceleration

Oct 6, 2026cs.LG

Consistent Distribution Matching for Data-Free Diffusion Distillation

Flow and diffusion models suffer from slow inference due to computationally expensive numerical integration. Distillation provides a promising way for a student model to learn from a teacher's dynamics, enabling one-step or few-step generation. However, existing methods often depend on curated distillation datasets, costly teacher rollouts, or auxiliary proxy networks, which complicate model training and scaling. In this work, we propose Consistent Distribution Matching, a simulation-free and data-free distillation method for accelerating diffusion and flow models while preserving strong generative capacity. Our key insight is to unify sample generation and score estimation with one student network. Thus, our framework uses only two models, a frozen teacher and a trainable student, and optimizes one objective. We prove that minimizing our objective indicates Wasserstein convergence of the student flow-map pushforwards to the teacher marginals. On ImageNet 256

\times

256, our method attains an FID of 2.04 with a single function evaluation (1-NFE) and a 4-NFE FID of 1.37 within 40 epochs of training, surpassing the state-of-the-art distillation baselines without data. Our code code and model are available at https://consistentdmd.github.io/.

Yuxiang Fu, Qi Yan, Zike Wu +4

University of British Columbia · Vector Institute for AI · Canada CIFAR AI Chair

Diffusion Model Distillation Diffusion Model Inference Acceleration Distribution Matching Distillation Data-Free Knowledge Distillation

Oct 6, 2026cs.LG

SNR-Gated LSTM-Conditioned Diffusion Model for MIMO Channel Estimation

Accurate and low latency channel estimation is critical for modern MIMO systems, particularly under mobility, where channels exhibit structured sparsity and strong temporal correlation. This paper proposes a time-series conditioned diffusion framework for channel estimation that performs denoising in the angular domain. Starting from least squares (LS) observations, we train a diffusion denoiser whose conditioning information is encoded by a long short-term memory (LSTM) network over a short observation sequence, enabling the model to exploit temporal dynamics beyond per-snapshot estimation. To robustly balance observation fidelity and learned generative priors across a wide signal-to-noise ratio (SNR) range, we introduce a learnable SNR-gated late-fusion shortcut that injects the network input into the final decoding stage through a sigmoid gate with trainable center and scale. To reduce inference latency, we adopt deterministic denoising diffusion implicit model (DDIM) style reverse updates with SNR-adaptive truncation and step allocation, which significantly reduces the number of reverse diffusion steps at high SNR while maintaining strong performance in low SNR regimes. Simulations on time-evolving standardized channel models demonstrate that the proposed method achieves consistent performance gains over existing diffusion-based channel estimation baselines, while retaining low latency through SNR-adaptive inference.

Jixing Zhou, Xinming Huang

Worcester Polytechnic Institute Worcester, MA, USA

Channel Estimation Conditional Diffusion Models Diffusion Models Diffusion Model Inference Acceleration

Oct 6, 2026cs.CV

Backend-Agnostic Sparse Attention for Fast High-Resolution Visual Generation

Diffusion Transformers (DiTs) have achieved strong performance in image and video generation, but the quadratic complexity of full attention makes high-resolution generation computationally expensive. Window attention offers an efficient alternative, yet existing methods face a practical trade-off: partitioned window attention typically achieves computational efficiency consistent with its theoretical complexity. However, isolated windows block cross-window interaction, often introducing visible grid-like artifacts in the generated results. Fine-grained sliding-window attention effectively restores interactions across neighboring windows and improves visual quality. However, its irregular computation patterns create a substantial gap between theoretical and practical speedups and require specialized kernels tailored to each hardware backend. To tackle these challenges, we propose BASA, a backend-agnostic sparse attention, which brings the best of both worlds: visual quality and practical acceleration. Specifically, BASA replaces visual self-attention with shifted local-window attention. By introducing a structured window-shifting scheme across DiT blocks, we allow tokens divided by window boundaries in one layer to communicate in the following layers, thereby achieving global information exchange and eliminating window-induced visual artifacts. Notably, our design introduces no additional irregular operators or customized kernels, making it readily deployable on existing attention backends and closing the gap between theoretical sparsity and practical acceleration. Experiments demonstrate that BASA achieves measured speedups exceeding 90% of the theoretical estimates on FLUX and delivers a 4.52

\times

attention speedup on Wan while maintaining competitive generation quality. Codes are publicly available at: https://github.com/lama0110/BASA.

Liao Ma, Jiayi Song, Yunfeng Wu +2

School of Artificial Intelligence, Shanghai Jiao Tong University · School of Data Science, Fudan University · School of Computing and Data Science, The University of Hong Kong

Video Diffusion Models Diffusion Transformer Diffusion-Based Image Generation Sparse Attention Sliding-Window Attention Diffusion Model Inference Acceleration

Oct 6, 2026cs.CV

RefRoute: Decoupling Conditioning Cost from References via Compact Residual Conditioning and Spatial Routing

Multi-reference image generation requires preserving the appearance of multiple subjects while composing them into a coherent scene. However, existing diffusion transformers commonly encode references as dense visual token grids and jointly process them with global attention, making conditioning increasingly expensive as the number and resolution of references grow. We present RefRoute, a framework that addresses both reference representation cost and attention overhead through two complementary mechanisms. Compact residual conditioning combines low-resolution latent tokens with lightweight residual features extracted from full-resolution pixels, reducing reference token counts while retaining fine-grained appearance cues. Condition routing and attention routing align reference tokens with their assigned target regions and restrict cross-reference interactions, while allowing selective reference access beyond region boundaries for scene integration. We further introduce RefRoute-Data for training many-reference generation models and ManyRef100, a benchmark spanning human, object, and mixed compositions with 10-17 references. After many-reference fine-tuning, RefRoute achieves an overall Weighted-Ref-VIEScore of 36.06 on ManyRef100, compared with 8.88 for FLUX.2-Klein-9B. Separate inference-cost evaluations show substantially slower latency growth as the reference count increases: at 16 references, our 50-step and 4-step configurations achieve

18.3\times

and

14.2\times

speedups over their corresponding FLUX baselines, respectively. These results establish compact reference representations and spatially routed attention as an effective approach to scalable many-reference image generation.

Wanning He, Yuyao Zhang, Yu-Wing Tai

Dartmouth College

Diffusion Transformer Image Generation Compositional Image Generation Attention Control in Diffusion Models Diffusion Model Inference Acceleration

Oct 5, 2026cs.CV

MC-Sparse: Deconstructing and Closing the Dense-Sparse Attention Gap in Diffusion Transformers

Sparse attention is a primary approach to reducing the latency of diffusion transformers in long-sequence generation tasks, such as video and high-resolution 3D asset generation. However, existing methods can degrade generation quality and fidelity at high sparsity levels. Through controlled oracle comparisons, we trace this degradation to three sources: constraints imposed by token grouping, inaccurate interaction selection, and the attention contributions lost when tokens are discarded. Guided by this analysis, we propose Meta-Cached Sparse Attention (MC-Sparse), a training-free framework that selects individual key-value (KV) tokens while organizing similar queries into tile-aligned groups for efficient GPU execution. MC-Sparse caches metadata comprising query groups, KV indices selected using exact attention probabilities, and residuals between dense and sparse attention outputs, and reuses them across subsequent denoising steps. Across video and 3D generation models, MC-Sparse achieves higher fidelity to dense-attention outputs and larger denoising speedups than existing sparse-attention baselines, without visible quality degradation. Relative to dense attention, it delivers a

1.80\times

denoising speedup on Minimax-H3-Base and a

2.32\times

speedup on 3D asset generation, both with negligible quality loss.

Jiarui Chen, Zeqiang Lai, Jiangshan Wang +5

Fudan University · Tencent HY · Shanghai Innovation Institute +4

Video Diffusion Models Diffusion Transformer Sparse Attention Diffusion Model Caching Diffusion Model Inference Acceleration

Oct 5, 2026cs.CV

Level-of-Token Diffusion

Image and video diffusion models allocate equal computation to every region, even when the intended scene calls for varying levels of detail. The spatial distribution of detail can often be anticipated before generation, indicating where computation can be reduced. We introduce Level-of-Token (LoT) Diffusion, a framework that turns this knowledge into an explicit multiresolution token layout (Level-of-Token layout) for adaptive and efficient generation. Tokens represent rectangular patches of varying sizes and shapes, allocating finer tokens where detail is needed and coarser tokens elsewhere. We adapt pretrained diffusion transformers to LoT layouts through a patch-wise asymmetric flow parametrization and embeddings for multiresolution tokens, preserving full-resolution flow prediction at every denoising step while processing only a reduced token sequence. LoT Diffusion enables layout-adaptive generation while preserving pretrained generative priors. We demonstrate LoT with layouts derived from semantic masks, bounding boxes, texture variance, and depth-of-field cues, as well as agentic plans. Across image and video generation, LoT offers favorable quality-efficiency tradeoffs, with significant speedups determined by the layout's token budget. Our project website is at https://georgenakayama.github.io/lotdiffusion/.

Kiyohiro Nakayama, Brian Chao, Jan Ackermann +5

Stanford University · Google

Video Diffusion Models Diffusion Transformer Diffusion-Based Image Generation Diffusion Model Inference Acceleration Visual Token Compression

Oct 5, 2026cs.RO

The Unexpired Plan: A Free Monitor for Accelerated Diffusion Policies

Training-free acceleration of a diffusion policy is accepted when an internal similarity signal reports that the shortcut changed nothing. We price every monitor a control loop can afford in closed-loop success rather than feature distance, over

114

accelerator configurations and four policy families: two accelerators each clear their own gate's bar and log every reuse as certified, while on one task one finishes every episode and the other none. A monitor is two designs, not one --- the statistic it reads, and what it does when that statistic fires. Published gates re-arm after every rejection, and under that response even an oracle handed every forward pass and the exact local action error loses twenty points; absorbing the same statistic costs one point, and most of its speed. What makes a response that never forgets affordable is a statistic that rarely fires, and what it must measure is deviation from the policy the accelerator replaced --- which a chunked policy has already paid for, its last plan not yet expired and free to read. Guarding every call this way cuts per-call compute by

1.55

--

3.09\times

, where any reference-requiring check at the same coverage would have to stop accelerating altogether. On three of our four families the schedule alone already holds the pre-stated

\pm2

-point margin. What the monitor is measurably worth shows in three places: on the fourth family, where it rescues the candidate selection landed on; on four configurations it did not select; and on a contact-rich fifth family, chosen where the schedule was expected to fail and run after every design choice was frozen, where no unmonitored arm at its speed holds the margin and the monitored one does.

Yi Zhao, Sebastian Scherer

AirLab, The Robotics Institute Carnegie Mellon University, Pittsburgh, PA, USA

Robotics Diffusion Policy Diffusion Model Inference Acceleration Runtime Verification

Oct 4, 2026cs.CV

Hybrid-Basis Feature Forecasting for Diffusion Sampling Acceleration

We propose Hybrid-Basis Feature Forecasting (HybridFF), a training-free, plug-and-play framework for accelerating diffusion sampling. To capture local smoothness, long-range trends, and complex non-monotonic variations when modeling feature evolution, HybridFF first estimates coefficients using moving least squares (MLS) for each of multiple complementary basis families and then combines the corresponding predictors using fusion weights. In addition to the choice of basis functions, the fusion weights also play a critical role. We introduce two strategies to balance quality and speedup. HybridFF (Fixed) prioritizes efficiency with model-specific fusion weights calibrated on a small set and held constant during inference. HybridFF (Adaptive) updates the fusion weights online using branch reliability scores computed from an exponential moving average of full-step prediction errors, improving prediction fidelity and generation quality under aggressive caching while retaining substantial acceleration. Experiments across DiT-XL/2, FLUX.1-dev, SD3.5-Large, and HunyuanVideo demonstrate a favorable speedup--quality trade-off over representative single-basis forecasters and caching baselines.

Kai-Liang Cheng, Yuan-Yuan Cheng, Yu-fan Jin +1

University of Science and Technology of China, China

Diffusion Sampling Diffusion Model Caching Diffusion Model Inference Acceleration Diffusion Models

Oct 1, 2026cs.LG

Clock Diffusion: Efficient Semi-Autoregressive Continuous Diffusion Language Models

Recent works on continuous diffusion for discrete data have demonstrated performance on par with comparable discrete diffusion models. However, these continuous counterparts lack key features that are essential to practical use as language models, namely variable-length generation and support for a key-value cache, and they still lag behind the frontier of autoregressive and discrete diffusion quality. In this work, we address these limitations. We do so by introducing a model parameterization that uses position-dependent noise schedules to define semi-autoregressive (SAR) continuous diffusion language models (DLMs). Together with efficient training and sampling algorithms, we call this framework Clock Diffusion, and we present two special cases of our method: block and sliding window generation. We then define ClockDLMs, a family of Gaussian DLMs based on sliding window Clock Diffusion that attain state-of-the-art diffusion likelihood bounds on OpenWebText, even beating the performant block SAR discrete diffusion models. ClockDLMs trained on TinyGSM also substantially outperform continuous baselines on the GSM8K benchmark and match and exceed comparable SAR discrete diffusion models. Finally, building on our parameterization, we propose more efficient samplers that we dub Cache Grab, which adapt techniques from accelerated inference in discrete diffusion, such as committing tokens whose probabilities exceed a confidence threshold and self-speculative decoding, further improving our models' quality and efficiency.

Yair Schiff, Omer Belhasin, Roy Uziel +7

NVIDIA · Cornell University

Diffusion Language Model Inference Language Modeling Diffusion Language Models Continuous Diffusion Language Models Diffusion Model Inference Acceleration

Sep 30, 2026cs.CV

Looped Diffusion Transformer

Improving text-to-image models has traditionally relied on increasing model size or the number of denoising steps. In this work, we explore an alternative way to scale computation by repeatedly running shared Transformer blocks within each denoising step, effectively increasing computational depth while keeping the parameter count fixed. This looped computation enables iterative refinement of internal representations without explicit reasoning tokens. However, naive looping fails to consistently improve image quality. We trace this problem to weak supervision across intermediate loops and unregulated attention updates that progressively erode local information. To overcome these challenges, we propose Looped Diffusion Transformer (Looped-DiT), which combines deep supervision across intermediate loops with self-modulating attention to stabilize looped feature updates. Under matched-parameter and matched-compute settings, Looped-DiT consistently outperforms non-looped baselines. Notably, a 260M-parameter looped model can surpass a model 6.5x larger across multiple text-to-image benchmarks while requiring 4.9x lower inference compute. Beyond this performance gain, we find that looped computation can offer a more effective form of iterative computation for diffusion models, with increasing loop depth yielding larger gains than adding more denoising steps under a fixed inference budget. Furthermore, deeper loops can progressively correct mistakes made in earlier loops, exhibiting behaviors suggestive of latent reasoning. Together, these results show that looped computation offers a promising way to scale visual generation models.

Yong Xien Chng, Tianyi Chen, Wenwen Tong +7

SenseTime Research · LeapLab, Tsinghua University · Nanyang Tehnological Univesity

Diffusion Transformer Recurrent Transformers Diffusion Model Inference Acceleration T2I Diffusion Models

Sep 30, 2026cs.LG

Distribution Matching Distillation for Continuous Diffusion Language Models

Continuous diffusion language models generate all tokens in parallel, yet high-quality generation can still require hundreds of network evaluations (NFEs). We study how distributional distillation can reduce this cost by exploiting the student's probabilistic token outputs. Our unified formulation connects the student's output parameterization to the resulting gradient estimators and yields two methods with the same student architecture and reverse-KL matching objective: Simplex-DMD uses continuous token relaxations and pathwise gradients, while Reinforce-DMD uses categorical sampling and REINFORCE with a learned density ratio. We develop both methods for multi-step generation and investigate the training and sampling choices associated with each parameterization. On OpenWebText, for sequences of 1,024 tokens, Simplex-DMD achieves a generative perplexity of 45.6 at a unigram entropy of 5.44 nats in just 4 NFEs, a 49% reduction relative to the strongest evaluated diffusion baseline at matched entropy and sampling budget. Reinforce-DMD improves the frontier at larger budgets, reaching a generative perplexity of 14.9 at an entropy of 5.00 nats with 256 NFEs, a 20% reduction under the same comparison protocol.

Paul Le Van Kiem, Dario Shariatian, Umut Simsekli +1

Inria, PSL Research University · Cohere · CMAP, Ecole Polytechnique

Diffusion Model Distillation Diffusion Language Models Continuous Diffusion Language Models Distribution Matching Distillation Diffusion Model Inference Acceleration Diffusion Language Model Distillation

Sep 30, 2026cs.LG

Fork-dLLM: Avoiding the Flexibility Trap in Diffusion Language Models

Masked diffusion language models (dLLMs) have shown strong potential for faster inference through parallel token generation when combined with confidence-based samplers. However, recent work has shown that such methods can defer unmasking high-entropy fork positions at which multiple plausible continuations exist. This results in reduced generation diversity, as shown by worse pass@k scaling, and limits gains obtainable from RL post-training. To avoid this flexibility trap, prior work advocated for autoregressive (AR) sampling. Here, we show that discarding confidence-based sampling is unnecessary and, once inference cost is taken into account, wasteful. We first propose Fork-dLLM, a simple hybrid sampler that uses AR-style ordering only at uncertain fallback steps while retaining parallel generation otherwise. We then extend the same principle to post-training with ForkGRPO, which uses Fork-dLLM rollouts and applies the GRPO objective only at fallback steps, preserving exact policy-likelihood ratios while substantially reducing rollout and optimization cost. In our experiments, Fork-dLLM matches the strong pass@k scaling of AR sampling while being 2-3x more efficient, and ForkGRPO achieves downstream performance comparable to or better than AR-based GRPO baselines at a substantially lower training cost.

Stipe Frković, Metod Jazbec, Christian A. Naesseth

University of Amsterdam · UvA-Bosch Delta Lab, University of Amsterdam

Masked Diffusion Models Speculative Decoding for Diffusion Language Models Diffusion Language Model Decoding RL for Diffusion Models Diffusion Model Inference Acceleration

Sep 30, 2026cs.LG

Graph Residual Conjugate Diffusion: SNR-Equalized Heat Flow for Graph Signals

Diffusion models generate data by reversing a forward corruption process that typically approaches a simple Gaussian prior. Recent work has extended this framework to signals supported on fixed graphs, e.g., road-network traffic and sensor-network measurements. Many graph signals have nonuniform spectral energy, whereas isotropic corruption adds the same conditional noise variance to every graph-frequency mode. Driving all modes to near-zero terminal signal-to-noise ratio (SNR) requires strong corruption, which increases the noise range that must be covered under a fixed sampling budget. We introduce Graph Residual Conjugate Diffusion (GRCD), which replaces the shared clock of graph heat diffusion with a mode-dependent clock that gives every graph-Fourier mode the same conditional SNR. GRCD fits a zero-mean graph-spectral Gaussian reference on the training split and stops at a finite terminal SNR at which the propagated reference still carries the fitted spectral variances. The Gaussian component has an exact modewise propagator in the probability-flow ODE, so sampling advances it analytically and integrates only the learned residual score numerically. We evaluate GRCD on five settings (METR-LA traffic, Molene weather, and three stochastic block models) against seven comparators under a matched protocol: Graph-Aware Diffusion (GAD), EDM (graph backbone), two adaptations of Whitened Score Diffusion (WSD), and three preconditioning controls. At four function evaluations (NFEs), GRCD lowers averaged maximum mean discrepancy (aMMD) by 22 to 36 times over the best comparator on all five settings, reaching 0.054 on METR-LA, where it clears an aMMD 0.1 target with 87% less sampling wall-clock time than the cheapest comparator that reaches it. Fitting the terminal reference reduces aMMD by 2.7 to 7.3 times at finite terminal SNR, while the factors shrink to 1.00 to 1.01 near zero.

Jinwei Li, Daniel Tenbrinck

Department of Data Science Friedrich-Alexander-Universität Erlangen-Nürnberg Erlangen, Germany

Graph Signal Processing Diffusion Model Inference Acceleration Graph Diffusion

Sep 30, 2026cs.AI

The Golden Path Hypothesis: Reusable Schedules in Diffusion Caching

Diffusion caching accelerates generation by replacing transformer computation with cached or predicted features at selected denoising steps. We introduce the Golden Path Hypothesis (GPH): under fixed inference conditions, prompt-independent cache schedules can achieve final-output quality comparable to the best prompt-specific schedules across prompts. We investigate the GPH across ten caching methods, four image and video models, and three cache ratios. Prompt-adaptive methods repeatedly select a small number of schedules, and reusing their most frequent schedules on new prompts closely matches the quality of prompt-specific choices. Exhaustive evaluation of 1.4 million schedules on four examples further identifies prompt-independent schedules that remain competitive on unseen prompts. To explain this transfer, we analyze denoising trajectories and the accumulation of caching errors. Latent-state trajectories exhibit similar structures across datasets and seeds, while an exact error decomposition shows that accumulated effects of earlier errors predict final latent-state error better than local approximation errors. This motivates searching for end-to-end schedules using final-output quality. With only a small set of examples, the resulting golden paths transfer across prompts and datasets, and can be tuned to the desired quality objective, including reconstruction fidelity or perceptual similarity.

Dong Wang, Wenwu Tang, Francesco Corti +3

Graz University of Technology, Austria · Swiss Data Science Center, Switzerland · ETH Zurich, Switzerland +1

Diffusion Model Caching Diffusion Timestep Scheduling Diffusion Model Inference Acceleration

Sep 30, 2026cs.CV

DeCoPrune: Efficient KV-Cache Pruning for Autoregressive Video Diffusion via Denoising Consistency

Autoregressive video diffusion supports streaming generation and interactive control, but its KV cache grows continuously with the generated history. Existing compression strategies either discard history using fixed windows or select tokens through local attention and similarity signals, which do not directly measure whether the current chunk contributes information beyond the retained context. We introduce DeCoPrune, a training-free method that treats cache compression as a denoising-consistency problem. We find empirically that denoising difficulty provides a useful proxy for a token's value in long-term retention: tokens with larger step-to-final discrepancies tend to carry visual evidence that is less predictable from the retained context. DeCoPrune measures each current-chunk token's denoising difficulty using the discrepancy between its intermediate clean prediction and final denoised value, retaining high-discrepancy tokens in the long-term cache while pruning those with low discrepancy. To evaluate information retention, we introduce CMBench, comprising 58 approximately one-minute generated or real-world context episodes and 116 Reappear or Revisit continuation tasks that require recalling specific previously observed objects or scenes. Experiments with LingBot World v2 show that DeCoPrune preserves near-FullKV long-range recall while pruning over 85% of historical KV tokens and accelerating continuation generation by over

4\times

, substantially outperforming the evaluated compression baselines at comparable budgets. These results indicate that denoising consistency can serve as a model-intrinsic signal for retaining long-range information while reducing autoregressive inference cost. Our project homepage is https://decoprune.github.io. The code is available at https://github.com/DeCoPrune/CMBench, and the benchmark at https://huggingface.co/datasets/Aoraku/CMBench.

Zeqi Xiao, Qingle Liu, Kaiwen Zhang +3

Nanyang Technological University · Tsinghua University · Princeton University

KV-Cache Management Autoregressive Video Generation Video Token Compression Diffusion Model Inference Acceleration KV-Cache Pruning

Sep 30, 2026cs.CV

PARK: Accurate Block Retrieval for Sparse Attention in Video Diffusion Transformers

Diffusion Transformers (DiTs) have become a dominant architecture for video generation, but their efficiency is limited by the quadratic complexity of full attention. Sparse attention reduces this cost by retrieving important blocks and computing attention only within them, but inaccurate retrieval can either degrade generation quality or yield unnecessary computation. We identify two retrieval mismatches in methods that retrieve blocks using the averaged representations of query and key blocks: (i) query-side aggregation mismatch, where averaging queries before Softmax fails to preserve their individual attention preferences, and (ii) key-side clustering metric mismatch, where standard Euclidean clustering in the original key space can group keys with dissimilar QK scores under the current query, so their average representation may not accurately represent how the current query scores individual keys. These mismatches can lead to inaccurate block retrieval. To address these mismatches, we propose PARK, a training-free sparse attention method for accurate block retrieval. PARK retains every original query, independently normalizes its attention over key blocks, and then averages these distributions within each query block. It also uses information from the current queries to transform keys before clustering, so that keys receiving similar QK scores are grouped together. A fused GPU kernel further reduces the overhead of block retrieval. Experiments on HunyuanVideo and Wan demonstrate that PARK improves block retrieval accuracy and preserves generation quality while accelerating inference, achieving the best quality-efficiency trade-off among the compared sparse attention methods.

Yun Dai, Jiarui Wen, Huiping Zhuang +2

South China University of Technology

Video Diffusion Models Diffusion Transformer Diffusion Model Inference Acceleration Block-Sparse Attention

Sep 29, 2026stat.ML

Acceleration of Diffusion Language Model through Discrete Average Generator

Discrete diffusion models and flow matching have emerged as powerful frameworks for generative modeling over discrete state spaces, yet efficient few-step generation remains a fundamental challenge. In this work, we introduce the Discrete Average Generator, a principled extension of MeanFlow to Continuous-Time Markov Chains (CTMCs). Analogously to how MeanFlow defines an average velocity field over a time interval in continuous spaces, we define an average generator as the normalized increment of the transition kernel over a time interval. We show that this average generator satisfies a self-consistency identity, which provides the foundation for our training objective. We further develop training strategies that align with the standard training paradigm of diffusion language models while keeping the resulting objective tractable. When projected onto per-coordinate marginals, the self-consistency identity admits a closed-form expression, enabling efficient training and inference. In Potts model simulations, our objective reduces the total variation distance of the

K

-step sampler by up to 67%. On OpenWebText, our method achieves the lowest generative perplexity among the evaluated methods for 8 to 64 sampling steps while enabling a

16\times

acceleration, and achieves comparable performance to existing methods on ImageNet.

Yidong Ouyang, Zhengyan Wan, Themis Haris +8

University of California, Los Angeles · Google · Work done as a research intern at Google

Few-Step Diffusion Sampling Diffusion Language Model Inference Flow-Based Generative Modeling Diffusion Language Models Diffusion Model Inference Acceleration

Sep 29, 2026cs.CV

LongLive-Plug: Once-for-All Distillation for Video Generation

Video diffusion models are increasingly developed into specialized models for diverse downstream tasks, and this development often includes a distillation stage, for example to accelerate sampling or to improve long-video generation. This stage is typically repeated for every specialized model. We introduce LongLive-Plug, a once-for-all distillation framework that learns reusable capabilities as LoRAs on a base model for training-free, plug-and-play deployment to compatible downstream models. These capabilities include single-pass classifier-free guidance, few-step sampling, and long-context error correction for autoregressive generation. The adapters remain reusable even when downstream models add conditioning branches, expand output channels. Despite training at a fixed guidance scale, our dedicated CFG LoRA provides text guidance control through its inference weight. Combining it with a few-step LoRA simultaneously preserves few-step generation and CFG controllability on downstream tasks. We verify training-free deployment on 54 downstream models across three backbone families and eight task categories, including world modeling, robotics, editing, and multimodal generation. The approach may support additional compatible models. Each capability can thus be distilled once per backbone family and reused without per-target retraining.

Shuai Yang, Luozhou Wang, Wei Huang +9

NVIDIA

Video Diffusion Models Diffusion Model Distillation Diffusion Model Inference Acceleration

Sep 29, 2026cs.LG

Time-Anchored Diffusion Language Models: Latent-Space Caching for Fast Generation

Recent work on anchored diffusion language models improves denoising by shaping an intermediate latent space with supervised important-token targets. In this work, we introduce time-based (self-supervised) anchoring, which learns and reuses latent anchors without requiring such targets. Our key observation is that anchors encode persistent properties of the clean sequence, such as its semantic intent, global structure, or intermediate plan. Although their hidden representations become stale as the token canvas evolves, their semantic content remains useful across nearby diffusion times. This is implemented through a two-stage architecture consisting of a relatively expensive anchor network that generates the latent cache state and a lightweight denoising network that intelligently combines the cached latent state with the current state at each reverse step using a fusion module. This gives anchoring a latent-space caching interpretation: the anchor network is evaluated periodically, while its cached representation is reused across multiple reverse steps. We instantiate this framework as TADM:Post-train, which time-anchorizes pretrained DLMs, and TADM:Pretraining, which learns time-based anchors during pretraining. Applied to DiffusionGemma-26B, TADM:Post-train improves throughput by approximately 49% to 79% on several math, code, and STEM benchmarks (GSM8K, AIME26, GPQA-Diamond, LiveCodeBench-v6, HumanEval, MMLU-Pro). TADM:Pretraining reduces Transformer-layer computation by up to 38% relative to a standard single-stage DLM, achieves up to 73% higher measured throughput than ADLM.

Joel Anto Paul, Litu Rout, Aditya Akella +1

UT Austin

Diffusion Language Model Inference Diffusion Model Caching Diffusion Language Models Diffusion Model Inference Acceleration