RT-Lynx: Putting the GEMM Sparsity In a Right Way for Diffusion Models

• URL: http://arxiv.org/abs/2605.26632v2
• Date: Mon, 01 Jun 2026 03:03:29 GMT
• Title: RT-Lynx: Putting the GEMM Sparsity In a Right Way for Diffusion Models
• Abstract summary: We show that DiT activations are intrinsically sparse and significantly more robust to N:M semi-structured sparsification than weights.<n>We propose RT-Lynx, which applies N:M sparsification to activations and incorporates error-compensation techniques to mitigate loss of accuracy.
• Score: 5.786869511584382
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Diffusion Transformers (DiT) achieve strong performance in image generation but incur substantial inference costs. While prior work has reduced this cost via quantization and distillation, semi-structured sparsity, which can nearly halve FLOPs, remains underexplored. A key reason is that most existing approaches focus on weight sparsification, and pruning 50% of the weights can remove critical model capacity and degrade generation quality. Our study, however, shows that DiT activations are intrinsically sparse and significantly more robust to N:M semi-structured sparsification than weights. Motivated by this observation, we advocate a paradigm shift from weight sparsification to activation sparsification. We propose RT-Lynx, which applies N:M sparsification to activations and incorporates error-compensation techniques to mitigate accuracy loss. We further implement highly optimized CUDA kernels tailored to this setting, achieving up to a 1.55x speedup on average in linear layers. Extensive experiments across multiple diffusion models demonstrate that our method preserves the generation quality of the original models while substantially accelerating inference.

Related papers

• Post-Training Pruning for Diffusion Transformers [23.732968732627825]
  Diffusion Transformers (DiTs) have demonstrated impressive performance in image generation but suffer from substantial computational overhead and resource consumption.<n>Traditional pruning methods amplify the relative contribution of weights in the saliency metric.<n>We propose DiT-Pruning, which improves pruning performance by introducing customized saliency criteria and pruning granularity.
  arXiv  Detail & Related papers  (2026-07-01T13:30:49Z)
• Stabilizing, Scaling & Enhancing MeanFlow for Large-scale Diffusion Distillation [37.28053263183048]
  Diffusion models exhibit remarkable generative capability, but their high latency limits practical deployment.<n>MeanFlow has attracted considerable attention due to its concise formulation and remarkable performance.<n>We introduce a warm-up technique, in which the original differential solution of MeanFlow is replaced by a discrete solution.
  arXiv  Detail & Related papers  (2026-05-18T04:16:15Z)
• $R_\text{dm}$: Re-conceptualizing Distribution Matching as a Reward for Diffusion Distillation [9.105357939499683]
  Diffusion models achieve state-of-the-art generative performance but are bottlenecked by their slow, iterative sampling process.<n>Recent approaches have attempted to break this ceiling by integrating Reinforcement Learning (RL)<n>We propose a novel paradigm by re-conceptualizing distribution matching as a reward, denoted as $R_textdm$.
  arXiv  Detail & Related papers  (2026-03-30T14:01:31Z)
• Revisiting Diffusion Models: From Generative Pre-training to One-Step Generation [2.3359837623080613]
  We show that diffusion training may be viewed as a form of generative pre-training.<n>We create a one-step generation model by fine-tuning a pre-trained model with 85% of parameters frozen.
  arXiv  Detail & Related papers  (2025-06-11T03:55:26Z)
• Denoising Score Distillation: From Noisy Diffusion Pretraining to One-Step High-Quality Generation [82.39763984380625]
  We introduce denoising score distillation (DSD), a surprisingly effective and novel approach for training high-quality generative models from low-quality data.<n>DSD pretrains a diffusion model exclusively on noisy, corrupted samples and then distills it into a one-step generator capable of producing refined, clean outputs.
  arXiv  Detail & Related papers  (2025-03-10T17:44:46Z)
• Determining Layer-wise Sparsity for Large Language Models Through a Theoretical Perspective [55.90119819642064]
  We address the challenge of determining the layer-wise sparsity rates of large language models (LLMs) through a theoretical perspective.<n>This refers to the cumulative effect of reconstruction errors throughout the sparsification process.<n>We derive a simple yet effective approach to layer-wise sparsity allocation that mitigates this issue.
  arXiv  Detail & Related papers  (2025-02-20T17:51:10Z)
• Efficient Diffusion as Low Light Enhancer [63.789138528062225]
  Reflectance-Aware Trajectory Refinement (RATR) is a simple yet effective module to refine the teacher trajectory using the reflectance component of images. textbfReflectance-aware textbfDiffusion with textbfDistilled textbfTrajectory (textbfReDDiT) is an efficient and flexible distillation framework tailored for Low-Light Image Enhancement (LLIE)
  arXiv  Detail & Related papers  (2024-10-16T08:07:18Z)
• Distilling Diffusion Models into Conditional GANs [90.76040478677609]
  We distill a complex multistep diffusion model into a single-step conditional GAN student model. For efficient regression loss, we propose E-LatentLPIPS, a perceptual loss operating directly in diffusion model's latent space. We demonstrate that our one-step generator outperforms cutting-edge one-step diffusion distillation models.
  arXiv  Detail & Related papers  (2024-05-09T17:59:40Z)
• A-SDM: Accelerating Stable Diffusion through Redundancy Removal and Performance Optimization [54.113083217869516]
  In this work, we first explore the computational redundancy part of the network. We then prune the redundancy blocks of the model and maintain the network performance. Thirdly, we propose a global-regional interactive (GRI) attention to speed up the computationally intensive attention part.
  arXiv  Detail & Related papers  (2023-12-24T15:37:47Z)
• Post-training Quantization for Text-to-Image Diffusion Models with Progressive Calibration and Activation Relaxing [49.800746112114375]
  We propose a novel post-training quantization method (Progressive and Relaxing) for text-to-image diffusion models. We are the first to achieve quantization for Stable Diffusion XL while maintaining the performance.
  arXiv  Detail & Related papers  (2023-11-10T09:10:09Z)
• A Theoretical Explanation of Activation Sparsity through Flat Minima and Adversarial Robustness [29.87592869483743]
  A recent empirical observation of activation sparsity in blocks offers an opportunity to drastically reduce computation costs for free. We propose the notion of sparsity as one source of activation sparsity and a theoretical explanation based on it.
  arXiv  Detail & Related papers  (2023-09-06T13:48:40Z)
• Powerpropagation: A sparsity inducing weight reparameterisation [65.85142037667065]
  We introduce Powerpropagation, a new weight- parameterisation for neural networks that leads to inherently sparse models. Models trained in this manner exhibit similar performance, but have a distribution with markedly higher density at zero, allowing more parameters to be pruned safely. Here, we combine Powerpropagation with a traditional weight-pruning technique as well as recent state-of-the-art sparse-to-sparse algorithms, showing superior performance on the ImageNet benchmark.
  arXiv  Detail & Related papers  (2021-10-01T10:03:57Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.