# SubFlow: Sub-mode Conditioned Flow Matching for Diverse One-Step Generation

Source: [SubFlow: Sub-mode Conditioned Flow Matching for Diverse One-Step Generation](<https://www.sota2.com/research/paper/69df1b6d7d0172140ec7b794>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-16T05:00:04.684Z

Authors: Yexiong Lin, Jia Shi, Shanshan Ye, Wanyu Wang, Yu Yao, Tongliang Liu

Published: 2026-04-14T04:36:02.000Z

## Abstract

Flow matching has emerged as a powerful generative framework, with recent few-step methods achieving remarkable inference acceleration. However, we identify a critical yet overlooked limitation: these models suffer from severe diversity degradation, concentrating samples on dominant modes while neglecting rare but valid variations of the target distribution. We trace this degradation to averaging distortion: when trained with MSE objectives, class-conditional flows learn a frequency-weighted mean over intra-class sub-modes, causing the model to over-represent high-density modes while systematically neglecting low-density ones. To address this, we propose SubFlow, Sub-mode Conditioned Flow Matching, which eliminates averaging distortion by decomposing each class into fine-grained sub-modes via semantic clustering and conditioning the flow on sub-mode indices. Each conditioned sub-distribution is approximately unimodal, so the learned flow accurately targets individual modes with no averaging distortion, restoring full mode coverage in a single inference step. Crucially, SubFlow is entirely plug-and-play: it integrates seamlessly into existing one-step models such as MeanFlow and Shortcut Models without any architectural modifications. Extensive experiments on ImageNet-256 demonstrate that SubFlow yields substantial gains in generation diversity (Recall) while maintaining competitive image quality (FID), confirming its broad applicability across different one-step generation frameworks. Project page: https://yexionglin.github.io/subflow.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Class-conditional image generation on ImageNet 256x256 (val)](<https://www.sota2.com/research/sota/class-conditional-image-generation-on-imagenet-256x256-val.md>): Class-conditional Image Generation; ImageNet 256x256 (val); Inception Score (IS): not recorded.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.12273>)

[PDF](<https://arxiv.org/pdf/2604.12273.pdf>)
