Asymmetric Flow Models

Hugging Face Daily Papers Papers
flow-based-generation high-dimensional low-rank text-to-image pixel-space latent-models fine-tuning

Summary

Asymmetric Flow Modeling (AsymFlow) restricts noise prediction to low-rank subspaces for efficient high-dimensional flow-based generation, achieving state-of-the-art results on ImageNet and text-to-image tasks by fine-tuning from latent flow models.

Flow-based generation in high-dimensional spaces is difficult because velocity prediction requires modeling high-dimensional noise, even when data has strong low-rank structure. We present Asymmetric Flow Modeling (AsymFlow), a rank-asymmetric velocity parameterization that restricts noise prediction to a low-rank subspace while keeping data prediction full-dimensional. From this asymmetric prediction, AsymFlow analytically recovers the full-dimensional velocity without changing the network architecture or training/sampling procedures. On ImageNet 256times256, AsymFlow achieves a leading 1.57 FID, outperforming prior DiT/JiT-like pixel diffusion models by a large margin. AsymFlow also provides the first-ever route for finetuning pretrained latent flow models into pixel-space models: aligning the low-rank pixel subspace to the latent space gives a seamless initialization that preserves the latent model's high-level semantics and structure, so finetuning mainly improves low-level mismatches rather than relearning pixel generation. We show that the pixel AsymFlow model finetuned from FLUX.2 klein 9B establishes a new state of the art for pixel-space text-to-image generation, beating its latent base on HPSv3, DPG-Bench, and GenEval while qualitatively showing substantially improved visual realism.
Original Article
View Cached Full Text

Cached at: 05/14/26, 04:17 AM

Paper page - Asymmetric Flow Models

Source: https://huggingface.co/papers/2605.12964

Abstract

Asymmetric Flow Modeling enables efficient high-dimensional flow-based generation by restricting noise prediction to low-rank subspaces while maintaining full-dimensional data prediction, achieving superior performance in pixel-space text-to-image generation through effective fine-tuning from latent models.

Flow-based generationin high-dimensional spaces is difficult becausevelocity predictionrequires modelinghigh-dimensional noise, even when data has stronglow-rank structure. We present Asymmetric Flow Modeling (AsymFlow), arank-asymmetric velocity parameterizationthat restricts noise prediction to a low-rank subspace while keeping data prediction full-dimensional. From this asymmetric prediction, AsymFlow analytically recovers the full-dimensional velocity without changing the network architecture or training/sampling procedures. On ImageNet 256times256, AsymFlow achieves a leading 1.57 FID, outperforming prior DiT/JiT-likepixel diffusion modelsby a large margin. AsymFlow also provides the first-ever route for finetuning pretrainedlatent flow modelsinto pixel-space models: aligning the low-rank pixel subspace to thelatent spacegives a seamless initialization that preserves the latent model’s high-level semantics and structure, so finetuning mainly improves low-level mismatches rather than relearning pixel generation. We show that the pixel AsymFlow model finetuned from FLUX.2 klein 9B establishes a new state of the art for pixel-spacetext-to-image generation, beating its latent base on HPSv3, DPG-Bench, and GenEval while qualitatively showing substantially improved visual realism.

View arXiv pageView PDFProject pageGitHub290Add to collection

Get this paper in your agent:

hf papers read 2605\.12964

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper2

#### Lakonik/AsymFlow-ImageNet Updatedabout 2 hours ago #### Lakonik/AsymFLUX.2-klein-9B Text-to-Image• Updatedabout 2 hours ago • 2

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2605.12964 in a dataset README.md to link it from this page.

Spaces citing this paper1

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Language Modeling with Hyperspherical Flows

arXiv cs.LG

This paper introduces S-FLM, a novel flow-based language model that operates in a hyperspherical latent space to address the computational costs and semantic limitations of existing discrete diffusion and continuous flow models.

SDFlow: Similarity-Driven Flow Matching for Time Series Generation

arXiv cs.AI

This paper introduces SDFlow, a similarity-driven flow matching framework for time series generation that addresses exposure bias in autoregressive models. It achieves state-of-the-art performance and inference speedups by operating in the frozen VQ latent space with low-rank manifold decomposition.

AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillation

Hugging Face Daily Papers

AnyFlow introduces a novel any-step video diffusion distillation framework that optimizes full ODE sampling trajectories through flow-map transition learning and backward simulation, achieving performance that matches or surpasses consistency-based counterparts while scaling with sampling step budgets.