Asymmetric Flow Models

TL;DR

AsymFlow introduces a rank-asymmetric velocity parameterization for flow models, enabling high-quality pixel-space image generation and seamless finetuning from latent models, achieving state-of-the-art results on ImageNet and text-to-image tasks.

Contribution

The paper proposes AsymFlow, a novel asymmetric velocity parameterization that improves flow-based image generation and allows effective finetuning from latent models without architectural changes.

Findings

Achieves 1.57 FID on ImageNet 256x256, outperforming prior models.

Enables finetuning latent flow models into pixel-space models effectively.

Sets new state-of-the-art in pixel-space text-to-image generation.

Abstract

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 256$\times$256, 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.