TMPDiff: Temporal Mixed-Precision for Diffusion Models

• URL: http://arxiv.org/abs/2603.14062v1
• Date: Sat, 14 Mar 2026 18:06:46 GMT
• Title: TMPDiff: Temporal Mixed-Precision for Diffusion Models
• Abstract summary: TMPDiff is a temporal mixed-precision framework for diffusion models.<n>It assigns different numeric precision to different denoising timesteps.<n>It consistently outperforms uniform-precision baselines at matched speedup.
• Score: 10.106598516648418
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Diffusion models are the go-to method for Text-to-Image generation, but their iterative denoising processes has high inference latency. Quantization reduces compute time by using lower bitwidths, but applies a fixed precision across all denoising timesteps, leaving an entire optimization axis unexplored. We propose TMPDiff, a temporal mixed-precision framework for diffusion models that assigns different numeric precision to different denoising timesteps. We hypothesize that quantization errors accumulate additively across timesteps, which we then validate experimentally. Based on our observations, we develop an adaptive bisectioning-based algorithm, which assigns per-step precisions with linear evaluation complexity, reducing an otherwise exponential search problem. Across four state-of-the-art diffusion models and three datasets, TMPDiff consistently outperforms uniform-precision baselines at matched speedup, achieving 10 to 20% improvement in perceptual quality. On FLUX.1-dev, TMPDiff achieves 90% SSIM relative to the full-precision model at a speedup of 2.5x over 16-bit inference.

Related papers

• Rethinking Pixel Mean Flows via Interval Denoiser [95.11499828543344]
  We propose the Interval Denoiser, a framework for latent-free generation.<n>Our model achieves an FID of 4.55 in one step (1-NFE) and 3.98 in two steps (2-NFE) without perceptual losses.
  arXiv  Detail & Related papers  (2026-08-05T13:22:43Z)
• From Atoms to Entropy: Optimal Noise Allocation for Diffusion Training in the Convex Regime [52.459513910242485]
  We develop a general statistical framework for studying optimal noise-level allocation in diffusion training.<n>We show that the optimized training schedule admits an atomic minimizer, concentrated on finitely many noise levels.<n>We then evaluate the entropic schedule in larger-scale experiments, where full schedule optimization is currently intractable.
  arXiv  Detail & Related papers  (2026-07-10T20:59:06Z)
• Efficient Coarse-to-Fine Diffusion Models with Time Step Sequence Redistribution [1.2744523252873352]
  We introduce Coarse-to-Fine Denoising (C2F) to reduce computation during coarse feature generation.<n>We also design Time Step Sequence Redistribution (TRD) for efficient sampling trajectory adjustment, requiring less than 10 minutes for search.<n> Experimental results demonstrate that the proposed methods achieve near-lossless performance with an 80% to 90% reduction in computation on CIFAR10 and LSUN-Church.
  arXiv  Detail & Related papers  (2026-03-22T18:16:14Z)
• Timestep-Aware Block Masking for Efficient Diffusion Model Inference [48.764748095994996]
  Diffusion Probabilistic Models (DPMs) have achieved great success in image generation but suffer from high inference latency.<n>We propose a novel framework to optimize the computational graph of pre-trained DPMs on a per-timestep basis.
  arXiv  Detail & Related papers  (2026-03-20T13:35:03Z)
• Noise Conditional Variational Score Distillation [60.38982038894823]
  Noise Conditional Variational Score Distillation (NCVSD) is a novel method for distilling pretrained diffusion models into generative denoisers.<n>By integrating this insight into the Variational Score Distillation framework, we enable scalable learning of generative denoisers.
  arXiv  Detail & Related papers  (2025-06-11T06:01:39Z)
• Geodesic Diffusion Models for Efficient Medical Image Enhancement [6.689992048758046]
  We propose a family of geodesic noise schedules corresponding to the shortest paths in probability space under the Fisher-Rao metric.<n>Based on these schedules, we propose Geodesic Diffusion Models (GDMs), which significantly improve training and sampling efficiency.<n>We evaluate GDM on two medical image enhancement tasks: CT image denoising and MRI image super-resolution.
  arXiv  Detail & Related papers  (2025-03-02T05:57:51Z)
• Towards More Accurate Diffusion Model Acceleration with A Timestep Tuner [112.99126045081046]
  A diffusion model, which is formulated to produce an image using thousands of denoising steps, usually suffers from a slow inference speed.<n>We propose a textbftimestep tuner that helps find a more accurate integral direction for a particular interval at the minimum cost.<n>Experiments show that our plug-in design can be trained efficiently and boost the inference performance of various state-of-the-art acceleration methods.
  arXiv  Detail & Related papers  (2023-10-14T02:19:07Z)
• AdaDiff: Accelerating Diffusion Models through Step-Wise Adaptive Computation [32.74923906921339]
  Diffusion models achieve great success in generating diverse and high-fidelity images, yet their widespread application is hampered by their inherently slow generation speed. We propose AdaDiff, an adaptive framework that dynamically allocates computation resources in each sampling step to improve the generation efficiency of diffusion models.
  arXiv  Detail & Related papers  (2023-09-29T09:10:04Z)
• Accelerating Diffusion-based Combinatorial Optimization Solvers by Progressive Distillation [43.60874566377421]
  This paper proposes to use progressive distillation to speed up the inference by taking fewer steps during the denoising process. Our experimental results show that the progressively distilled model can perform inference 16 times faster with only 0.019% degradation in performance.
  arXiv  Detail & Related papers  (2023-08-12T21:25:24Z)
• Parallel Sampling of Diffusion Models [76.3124029406809]
  Diffusion models are powerful generative models but suffer from slow sampling. We present ParaDiGMS, a novel method to accelerate the sampling of pretrained diffusion models by denoising multiple steps in parallel.
  arXiv  Detail & Related papers  (2023-05-25T17:59:42Z)
• PTQD: Accurate Post-Training Quantization for Diffusion Models [22.567863065523902]
  Post-training quantization of diffusion models can significantly reduce the model size and accelerate the sampling process without re-training. Applying existing PTQ methods directly to low-bit diffusion models can significantly impair the quality of generated samples. We propose a unified formulation for the quantization noise and diffusion perturbed noise in the quantized denoising process.
  arXiv  Detail & Related papers  (2023-05-18T02:28:42Z)
• Pseudo Numerical Methods for Diffusion Models on Manifolds [77.40343577960712]
  Denoising Diffusion Probabilistic Models (DDPMs) can generate high-quality samples such as image and audio samples. DDPMs require hundreds to thousands of iterations to produce final samples. We propose pseudo numerical methods for diffusion models (PNDMs) PNDMs can generate higher quality synthetic images with only 50 steps compared with 1000-step DDIMs (20x speedup)
  arXiv  Detail & Related papers  (2022-02-20T10:37:52Z)

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.