Understanding, Accelerating, and Improving MeanFlow Training

• URL: http://arxiv.org/abs/2511.19065v1
• Date: Mon, 24 Nov 2025 12:59:27 GMT
• Title: Understanding, Accelerating, and Improving MeanFlow Training
• Authors: Jin-Young Kim, Hyojun Go, Lea Bogensperger, Julius Erbach, Nikolai Kalischek, Federico Tombari, Konrad Schindler, Dominik Narnhofer,
• Abstract summary: MeanFlow promises high-quality generative modeling in few steps, by jointly learning instantaneous and average velocity fields.<n>We analyze the interaction between the two velocities and find: (i) well-established instantaneous velocity is a prerequisite for learning average velocity.<n>We design an effective training scheme that accelerates the formation of instantaneous velocity, then shifts emphasis from short to long-interval average velocity.
• Score: 64.84964628592418
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: MeanFlow promises high-quality generative modeling in few steps, by jointly learning instantaneous and average velocity fields. Yet, the underlying training dynamics remain unclear. We analyze the interaction between the two velocities and find: (i) well-established instantaneous velocity is a prerequisite for learning average velocity; (ii) learning of instantaneous velocity benefits from average velocity when the temporal gap is small, but degrades as the gap increases; and (iii) task-affinity analysis indicates that smooth learning of large-gap average velocities, essential for one-step generation, depends on the prior formation of accurate instantaneous and small-gap average velocities. Guided by these observations, we design an effective training scheme that accelerates the formation of instantaneous velocity, then shifts emphasis from short- to long-interval average velocity. Our enhanced MeanFlow training yields faster convergence and significantly better few-step generation: With the same DiT-XL backbone, our method reaches an impressive FID of 2.87 on 1-NFE ImageNet 256x256, compared to 3.43 for the conventional MeanFlow baseline. Alternatively, our method matches the performance of the MeanFlow baseline with 2.5x shorter training time, or with a smaller DiT-L backbone.

Related papers

• FastFlow: Accelerating The Generative Flow Matching Models with Bandit Inference [10.34801095627052]
  Flow-matching models deliver state-of-the-art fidelity in image and video generation, but the inherent sequential denoising process renders them slower.<n>We propose FastFlow, a plug-and-play adaptive inference framework that accelerates generation in flow matching models.<n> Experiments demonstrate a speedup of over 2.6x while maintaining high-quality outputs.
  arXiv  Detail & Related papers  (2026-02-11T18:21:11Z)
• FlowConsist: Make Your Flow Consistent with Real Trajectory [99.22869983378062]
  We argue that current fast-flow training paradigms suffer from two fundamental issues.<n> conditional velocities constructed from randomly paired noise-data samples introduce systematic trajectory drift.<n>We propose FlowConsist, a training framework designed to enforce trajectory consistency in fast flows.
  arXiv  Detail & Related papers  (2026-02-06T03:24:23Z)
• Flow Straighter and Faster: Efficient One-Step Generative Modeling via MeanFlow on Rectified Trajectories [14.36205662558203]
  Rectified MeanFlow is a framework that models the mean velocity field along the rectified trajectory using only a single reflow step.<n>Experiments on ImageNet at 64, 256, and 512 resolutions show that Re-MeanFlow consistently outperforms prior one-step flow distillation and Rectified Flow methods in both sample quality and training efficiency.
  arXiv  Detail & Related papers  (2025-11-28T16:50:08Z)
• MeanFlow Transformers with Representation Autoencoders [71.45823902973349]
  MeanFlow (MF) is a diffusion-motivated generative model that enables efficient few-step generation by learning long jumps directly from noise to data.<n>We develop an efficient training and sampling scheme for MF in the latent space of a Representation Autoencoder (RAE)<n>We achieve a 1-step FID of 2.03, outperforming vanilla MF's 3.43, while reducing sampling GFLOPS by 38% and total training cost by 83% on ImageNet 256.
  arXiv  Detail & Related papers  (2025-11-17T06:17:08Z)
• Decoupled MeanFlow: Turning Flow Models into Flow Maps for Accelerated Sampling [68.76215229126886]
  We introduce Decoupled MeanFlow, a simple decoding strategy that converts flow models into flow map models without architectural modifications.<n>Our method conditions the final blocks of diffusion transformers on the subsequent timestep, allowing pretrained flow models to be directly repurposed as flow maps.<n>On ImageNet 256x256 and 512x512, our models attain 1-step FID of 2.16 and 2.12, respectively, surpassing prior art by a large margin.
  arXiv  Detail & Related papers  (2025-10-28T14:43:48Z)
• MeanFlowSE: one-step generative speech enhancement via conditional mean flow [13.437825847370442]
  MeanFlowSE is a conditional generative model that learns the average velocity over finite intervals along a trajectory.<n>On VoiceBank-DEMAND, the single-step model achieves strong intelligibility, fidelity, and perceptual quality with substantially lower computational cost than multistep baselines.
  arXiv  Detail & Related papers  (2025-09-18T11:24:47Z)
• SplitMeanFlow: Interval Splitting Consistency in Few-Step Generative Modeling [23.539625950964876]
  Generative models like Flow Matching have achieved state-of-the-art performance but are often hindered by a computationally expensive iterative sampling process.<n>MeanFlow, a leading method in this area, learns this field by enforcing a differential identity that connects the average and instantaneous velocities.<n>In this work, we argue that this differential formulation is a limiting special case of a more fundamental principle.<n>We introduce SplitMeanFlow, a new training framework that enforces this algebraic consistency directly as a learning objective.
  arXiv  Detail & Related papers  (2025-07-22T16:26:58Z)
• A Stable Whitening Optimizer for Efficient Neural Network Training [99.7641280234926]
  Building on the Shampoo family of algorithms, we identify and alleviate three key issues, resulting in the proposed SPlus method.<n>First, we find that naive Shampoo is prone to divergence when matrix-inverses are cached for long periods.<n>Second, we adapt a shape-aware scaling to enable learning rate transfer across network width.<n>Third, we find that high learning rates result in large parameter noise, and propose a simple iterate-averaging scheme which unblocks faster learning.
  arXiv  Detail & Related papers  (2025-06-08T18:43:31Z)
• Mean Flows for One-step Generative Modeling [64.4997821467102]
  We propose a principled and effective framework for one-step generative modeling.<n>A well-defined identity between average and instantaneous velocities is derived and used to guide neural network training.<n>Our method, termed the MeanFlow model, is self-contained and requires no pre-training, distillation, or curriculum learning.
  arXiv  Detail & Related papers  (2025-05-19T17:59:42Z)

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.