Diffusion Model Inference Acceleration

Overview

Diffusion models repeatedly denoise a sample, so inference acceleration studies how to reduce the cost of those steps for image, video, language, and action generation. Methods reduce the number of evaluations with faster solvers, timestep schedules, and few-step distillation, or reduce work within each step using caching, sparse attention, token pruning, compression, and lower-precision computation. Systems work also targets memory movement, GPU parallelism, and serving, while adaptive methods allocate computation according to uncertainty, content, or task difficulty. The central challenge is preserving fidelity, diversity, and task accuracy as skipped or approximated computations can accumulate errors; gains therefore depend on the model, workload, hardware, and end-to-end evaluation.

Latest papers

CardsList

545 papers36 papers in the last four weeks, down 54% on the four weeks before. 0.3% of all new papers.

  1. AudioTurbo: Fast Text-to-Audio Generation with Rectified Diffusion

    May 28, 2025Junqi Zhao, Jinheng Zhao, Haohe Liu +6Text-to-Audio GenerationDiffusion Model Inference Acceleration

  2. Disentangling Total-Variance and Signal-to-Noise-Ratio Improves Diffusion Models

    Feb 12, 2025Khaled Kahouli, Winfried Ripken, Stefan Gugler +3Diffusion Model SamplingDiffusion Timestep Scheduling

  3. DICE: Distilling Classifier-Free Guidance into Text Embeddings

    Feb 6, 2025Zhenyu Zhou, Defang Chen, Can Wang +2Diffusion Model AlignmentDiffusion Model Guidance

  4. RDM: Recurrent Diffusion Model for Human Motion Generation

    Jun 11, 2024Mirgahney Mohamed, Harry Jake Cunningham, Riccardo Barbano +2Diffusion Models for Motion GenerationDiffusion Models

  5. Decoupled Data Consistency with Diffusion Purification for Image Restoration

    Mar 10, 2024Xiang Li, Soo Min Kwon, Shijun Liang +3Diffusion Models for Image RestorationDiffusion-Based Inverse Problems