---
title: Rethinking Test Time Scaling for Flow-Matching Generative Models
license: apache-2.0
arxiv: 2511.22242
tags:
- flow-matching
- test-time-scaling
- generative-models
---

# Rethinking Test Time Scaling for Flow-Matching Generative Models
[![GitHub](https://img.shields.io/badge/GitHub-Repo-181717?logo=github)](https://github.com/TerrysLearning/DOGTrimTTS)
[![arXiv](https://img.shields.io/badge/arXiv-2511.22242-b31b1b.svg)](https://arxiv.org/abs/2511.22242)

## About
This repository contains the models and configuration for our paper **[Rethinking Test Time Scaling for Flow-Matching Generative Models](https://arxiv.org/abs/2511.22242)**.

After analyzing the limitations of existing methods on ODE flow-matching models, we propose: 

***DOG-Trim: Diversity enhanced Order aligned Global flow Trimming***

![Motivation](doc/motivation_plot.png)

## Qualitative examples using Flux1.dev:
![Example](doc/dog_knife.png)
![Example2](doc/someresults.png)

## Citation
If you find this work useful, please consider citing our arXiv preprint.
```bash
@article{yu2026RethinkTTS,
  title={Rethinking Test Time Scaling for Flow-Matching Generative Models},
  author={Yu, Qingtao and Song, Changlin and Sun, Minghao and Yu, Zhengyang and Verma, Vinay Kumar and Roy, Soumya and Negi, Sumit and Li, Hongdong and Campbell, Dylan},
  journal={arXiv preprint arXiv:2511.22242},
  year={2026}
}
```