Diffusion Timestep Scheduling

Follow this topic Ask about this topic

Diffusion timestep scheduling studies when noise is added during training and where a sampler evaluates the denoising process at inference. Approaches include hand-designed cosine or log-SNR schedules, learned monotone splines, and optimization using numerical-error estimates, optimal control, reinforcement learning, or Bayesian search. Adaptive methods also allocate steps by input difficulty or task, and schedule sparse guidance or cached computation; they are evaluated across image, video, language, scientific, and discrete-generation settings. A central tension is whether schedules should be fixed and transferable or adapted to each model and sample, since benefits depend on the solver, denoiser, step budget, and quality measure.

Latest papers

64 papers10 papers in the last four weeks, up 67% on the four weeks before. 0.1% of all new papers.

LLaCO: Continuous Diffusion for Large Language Models with Competition-Aware Semantic Scheduling
Oct 23, 2026Zijian Jia, Peixuan Zhang, Genying Li +5Diffusion Language Models Continuous Diffusion Language Models
School of Artificial Intelligence, Beijing University of Posts and Telecommunications · Yuanpei College, Peking University · Beijing Key Laboratory of Multimodal Data Intelligent Perception and Governance +3
SoftCache: Adaptive Step Size Caching for Diffusion Transformers
Oct 15, 2026Shimon Malnick, Nadav Orenstein, Ohad Fried +1Diffusion Model Caching Diffusion Model Inference Acceleration
Tel Aviv University · Reichman University
Value-Informed Schedule Trimming for Accelerated Sequential Monte Carlo Guidance in Discrete Diffusion
Oct 14, 2026Jaehyeon Kim, Ziseok Lee, Siyoung Choi +1Inference-Time Guidance Diffusion Models
Seoul National University
A Step in the Right Place: Understanding and Optimizing Timestep Placement in Continuous Diffusion Language Models
Oct 11, 2026Shutian Zheng, Zhihong Yu, Linye Wei +2Continuous Diffusion Language Models Diffusion Timestep Scheduling
College of Engineering, Peking University · Weiyang College, Tsinghua University · Institute for Artificial Intelligence, Peking University +3
Streaming-Aware Diffusion for Real-Time Video Super-Resolution via Cross-Step Attention
Oct 8, 2026Harris Partaourides, Sotirios ChatzisVideo Super-Resolution Diffusion Model Inference Acceleration
Ethical AI Novelties Limassol, Cyprus · Cyprus University of Technology Limassol, Cyprus
CacheScope: Exact-Compute Allocation for Training-Free Diffusion Acceleration
Oct 8, 2026Qinhan Lyu, Jusheng Zhang, Zhixing Zhang +4Adaptive Computation for Diffusion Models Diffusion Model Caching
Sun Yat-sen University
Energy-Conditioned Noise Schedule and Whitening for Spectral Diffusion
Oct 5, 2026Bata Vasic, Bane VasicImage Generation Diffusion Timestep Scheduling
University of Nis, Faculty of Electronic Engineering, Aleksandra Medvedeva 4, Nis, 18104,, Serbia
Adaptive Second-Order Solvers for Fast Stochastic Diffusion Sampling
Oct 2, 2026Ella Kemperman, Luca AmbrogioniDiffusion Sampling Diffusion Timestep Scheduling
Radboud University
The Golden Path Hypothesis: Reusable Schedules in Diffusion Caching
Sep 30, 2026Dong Wang, Wenwu Tang, Francesco Corti +3Diffusion Model Caching Diffusion Timestep Scheduling
Graz University of Technology, Austria · Swiss Data Science Center, Switzerland · ETH Zurich, Switzerland +1
Sol-H3: Recursive Self-Improvement for MiniMax-H3 Inference Acceleration on Sol-Engine across Cloud and Edge
Sep 28, 2026Yitong Li, Jincheng Yu, Junsong Chen +6Video Diffusion Models GPU Kernel Optimization
NVIDIA Research, Efficient AI Team & Singapore Lab.
GeoShrink: Accelerating Diffusion Transformers with Two Lines of Code
Sep 27, 2026Haosen Li, Wenshuo Chen, Shaofeng Liang +3Video Diffusion Models Diffusion Transformer
The Hong Kong University of Science and Technology (Guangzhou) · Griffith University · Data61, CSIRO +1
Schedule optimization for tau-leaping in masked discrete diffusion
Sep 18, 2026Cecilia Secchi, Giacomo ZanellaMasked Diffusion Models Diffusion Sampling
Bocconi University, Department of Decision Sciences, Milan, Italy. · Bocconi University, Department of Decision Sciences and BIDSA, Milan, Italy.
Efficient Text-to-Image Generation: An Adaptive Step Schedule Controller for Diffusion Models
Sep 15, 2026Kuluhan Binici, Cihan Acar, Shivam Aggarwal +2Diffusion Model Sampling Diffusion Timestep Scheduling
1SAP · 2National University of Singapore · Institute for Infocomm Research (I2R), A*STAR, Singapore
CrossDistill: Balancing Quality and Diversity via Trajectory-Level Hybrid Few-Step Distillation
Sep 13, 2026Yuxi Liu, Haoyu Li, Yixiang Cai +8Video Diffusion Models Diffusion Model Distillation
Peking University, Melon Group · Alibaba Group · Tsinghua University
Model-Aware Schedules Improve Generation via Fiberwise Optimal Transport
Sep 10, 2026Luyi Jia, Boyan Zhang, Yilun Liu +1Flow Matching Diffusion Timestep Scheduling
Arnold-Sommerfeld-Center for Theoretical Physics, Ludwig-Maximilians-Universität München, Munich, Germany · Institute of Informatics, Ludwig-Maximilians-Universität München, Munich, Germany · Munich Center for Machine Learning, Munich, Germany
Generative Diffusion Surrogates with Analytical Variance Schedule
Sep 1, 2026Patrick Reichherzer, Gianluca Gregori, David N. Hosking +1Diffusion Timestep Scheduling
Department of Physics, University of Oxford, Oxford OX1 3PU, UK. · Department of Astrophysical Sciences, Princeton University, Princeton NJ 08544, USA. · Max Planck Institute for Security and Privacy, Bochum 44799, Germany. +2
Discrete Diffusion Bridges for Spatiotemporally Aligned Image Translation and Generation
Aug 30, 2026Xing Xie, Jiawei Liu, Shijun Zhou +4Masked Diffusion Models Image-to-Image Translation
State Key Laboratory of Robotics and Intelligent Systems, Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang, China · University of Chinese Academy of Sciences, Beijing, China · School of Computing and Data Science, The University of Hong Kong, Hong Kong
The information geometry of product-reference discrete diffusion: Interaction growth complexity and optimal scheduling
Aug 28, 2026Martin J. WainwrightDiffusion Sampling Diffusion Timestep Scheduling
Lab for Information and Decision Systems Statistics and Data Science Center EECS and Mathematics, Massachusetts Institute of Technology
I-SplineFlow: Learning Monotone Spline Stochastic Interpolant Schedulers for Few-Step Generation
Aug 23, 2026Md Sakib Hossain Shovon, Md Rifat Ur Rahman, Md Abtahi Majeed Chowdhury +3Few-Step Diffusion Sampling Diffusion Timestep Scheduling
KAIST · Bangladesh University of Engineering and Technology
Optimize Your Sampling: Tuned Diffusion Sampling with Bayesian Optimization
Aug 18, 2026Travis Zhang, Christian Belardi, Justin Lovelace +4Diffusion Timestep Scheduling Few-Step Diffusion Sampling
Cornell University
When Denoising Hurts: Rethinking the Terminal Step of Diffusion Time Series Forecasters -- Extended Version
Aug 14, 2026Dat Nguyen-Cong, Luong Tran, Tung KieuTime Series Forecasting Diffusion Sampling
FPT Software AI Center, FPT Corporation · Department of Computer Science, Aalborg University, Denmark
Adversarial Learning of Classifier-Free Guidance Schedules
Aug 14, 2026Ashwini Pokle, Alexandre Galashov, Arnaud Doucet +2Classifier-Free Guidance T2I Diffusion Models
Google DeepMind
The data geometry of masking diffusion: Certified-optimal schedules via unmasking growth complexity
Aug 13, 2026Martin J. WainwrightMasked Diffusion Models Diffusion Sampling
Lab for Information and Decision Systems · Statistics and Data Science Center · EECS and Mathematics
Analytic Distribution of Classifier-Free Guidance for Schedule Design
Jul 22, 2026Enze Jiang, Zheng MaClassifier-Free Guidance Diffusion Sampling
School of Mathematical Sciences Shanghai Jiao Tong University Shanghai, China
FlashDiff: Efficient Regional Execution and Scheduling for Diffusion Model Serving
Jul 13, 2026Yaqi Qiao, Ping He, Songrun Xie +4Diffusion Model Serving Adaptive Computation for Diffusion Models
University of Illinois Urbana-Champaign · Vanderbilt University · HKUST +2
D2PO: Optimizing Diffusion Samplers via Dynamic Preference
Jul 7, 2026Jinkyu Kim, Jinyoung Choi, Bohyung HanDiffusion Model Alignment Direct Preference Optimization
1ECE & 2IPAI, Seoul National University, Korea · 3AIGS, Ulsan National Institute of Science and Technology, Korea
Mixture-of-Gaussians-Guided Schedule Design for Brownian Bridge Diffusion Models
Jul 3, 2026Ron Levi, Michael EladDiffusion Models for Image Restoration Diffusion Timestep Scheduling
Department of Computer Science Technion, Israel
Adaptive Reparametrized Time for Score-Based Diffusion Sampling
Jul 2, 2026Yilie Huang, Wenpin Tang, Xun Yu ZhouDiffusion Sampling Adaptive Sampling
Department of Applied Mathematics, The Hong Kong Polytechnic University, Hong Kong. · Department of Industrial Engineering and Operations Research, Columbia University, New York, NY 10027, USA. · Department of Industrial Engineering and Operations Research and Data Science Institute, Columbia University, New York, NY 10027, USA.
Information-Theoretic Classifier-Free Guidance with Adaptive Schedule Optimization
Jun 23, 2026Haobo Chen, Xiangxiang Xu, Yuheng BuDiffusion Model Guidance Classifier-Free Guidance
Department of Computer Science University of California, Santa Barbara · Department of Computer Science University of Rochester
Learning When to Denoise: Optimizing Asynchronous Schedules for Latent Diffusion
Jun 18, 2026Bingshuo Qian, Xiang ChengFlow Matching Latent Diffusion Models
Department of Electrical and Computer Engineering Duke University

Oct 23, 2026J/K move · Enter open · S save

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

Zijian Jia, Peixuan Zhang, Genying Li, Di Wu+4

School of Artificial Intelligence, Beijing University of Posts and Telecommunications · Yuanpei College, Peking University · Beijing Key Laboratory of Multimodal Data Intelligent Perception and Governance+3

Diffusion Language Models Continuous Diffusion Language Models Diffusion Models Diffusion Timestep Scheduling

Open paper AskPDF arXiv

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.