README.md

---
license: apache-2.0
base_model: stabilityai/stable-diffusion-3.5-medium
library_name: diffusers
pipeline_tag: text-to-image
tags:
- text-to-image
- diffusion
- flow-matching
- meanflow
- reinforcement-learning
- lora
---

<!-- markdownlint-disable MD028 MD033 MD041 -->
<div align="center">

<img src="assets/meanflownft_logo.jpg" width="48%" alt="MeanFlowNFT logo">

<h2>Bringing Forward-Process RL to Average-Velocity Generators</h2>

[Yushi Huang](https://harahan.github.io/)<sup>1,2</sup>\* ·
[Xiangxin Zhou](https://zhouxiangxin1998.github.io/)<sup>1</sup>\*<sup>✉</sup> ·
[Jun Zhang](https://eejzhang.people.ust.hk/)<sup>2</sup> ·
[Liefeng Bo](https://research.cs.washington.edu/istc/lfb/)<sup>1</sup> ·
[Tianyu Pang](https://p2333.github.io/)<sup>1,✉</sup>

<sup>1</sup>Tencent Hunyuan &nbsp;&nbsp; <sup>2</sup>The Hong Kong University of Science and Technology

\* Equal contribution &nbsp;&nbsp; <sup>✉</sup>Corresponding authors

[![arXiv](https://img.shields.io/badge/arXiv-2607.15273-A42C25?style=for-the-badge&logo=arxiv&logoColor=white)](https://arxiv.org/abs/2607.15273)
[![Project Page](https://img.shields.io/badge/Project-Page-0F9D8A?style=for-the-badge&logo=googlechrome&logoColor=white)](https://harahan.github.io/meanflownft-project-page/)
[![Code](https://img.shields.io/badge/GitHub-MeanFlowNFT-181717?style=for-the-badge&logo=github)](https://github.com/Harahan/MeanFlowNFT)
[![Demo](https://img.shields.io/badge/🤗-Interactive_Demo-FF4B4B?style=for-the-badge)](https://huggingface.co/spaces/Harahan/meanflownft-fewstep-generation)

</div>

MeanFlowNFT optimizes reward in induced instantaneous-velocity space while
retaining the average-velocity parameterization and native any-step MeanFlow
sampler. Training uses only forward-noised clean samples, without
reverse-trajectory likelihood estimation.

<p align="center">
  <a href="https://harahan.github.io/meanflownft-project-page/"><img src="assets/teaser.webp" width="100%" alt="Selected MeanFlowNFT generations"></a>
</p>

## 🍭 Overview

<table align="center">
  <tr>
    <td align="center" width="58%" valign="middle">
      <img src="assets/overview.webp" width="100%" alt="MeanFlowNFT method overview">
      <br><sub>Optimize in instantaneous-velocity space while preserving MeanFlow sampling.</sub>
    </td>
    <td align="center" width="42%" valign="middle">
      <img src="assets/algorithm.jpg" width="100%" alt="MeanFlowNFT update algorithm">
      <br><sub>One practical MeanFlowNFT update.</sub>
    </td>
  </tr>
</table>

## 📦 Contents

This repository contains the three ordered adapter components required to
reconstruct the final policy:

```text
.
├── anyflow_pretrain/
│   └── generator_ema.pt
├── anyflow_onpolicy/
│   └── generator_ema.pt
└── meanflownft/
    └── generator_ema.pt
```

The loading contract is strict:

1. Merge `anyflow_pretrain/generator_ema.pt` into the base transformer.
2. Merge `anyflow_onpolicy/generator_ema.pt` on top of that result.
3. Keep `meanflownft/generator_ema.pt` as the active final adapter.
4. Load the complete `delta_embedder` state from the final checkpoint.

The adapter checkpoints use LoRA rank **32**, alpha **64**, and attention
projections matching the released code. They also contain the trainable
flow-map `delta_embedder`, so they are not conventional LoRA-only files.
`SHA256SUMS` records the checksum of every checkpoint.

## 📊 Results

<p align="center">
  <a href="assets/results_sd35.png"><img src="assets/results_sd35.png" width="100%" alt="MeanFlowNFT image-generation results"></a>
</p>

<p align="center">
  <a href="assets/results_wan.png"><img src="assets/results_wan.png" width="100%" alt="MeanFlowNFT video-generation results"></a>
</p>

MeanFlowNFT is best on **6 of 8** image metrics among the evaluated few-step
models, while 4-step Wan2.1 reaches **84.33 VBench**, surpassing 50-step
LongCat-Video RL. See the
[project page](https://harahan.github.io/meanflownft-project-page/) for full
comparisons, scaling curves, and qualitative results.

## 🚀 Usage

Use the [MeanFlowNFT codebase](https://github.com/Harahan/MeanFlowNFT), whose
inference loader preserves the ordered adapter composition and full
`delta_embedder` state.

```bash
git clone https://github.com/Harahan/MeanFlowNFT.git
cd MeanFlowNFT
pip install -r requirements.txt
pip install -e .

export HF_REPO_ID=Harahan/MeanFlowNFT
hf download "${HF_REPO_ID}" --local-dir ./checkpoints/meanflownft

hf auth login
hf download stabilityai/stable-diffusion-3.5-medium \
    --local-dir ./models/stable-diffusion-3.5-medium
```

Run the final 4-step policy:

```bash
python inference.py configs/inference/sd35m_meanflow_nft.yaml \
    --override \
    pretrained_path=./models/stable-diffusion-3.5-medium \
    num_stages=3 \
    stage1_lora_path=./checkpoints/meanflownft/anyflow_pretrain/generator_ema.pt \
    stage2_lora_path=./checkpoints/meanflownft/anyflow_onpolicy/generator_ema.pt \
    stage3_lora_path=./checkpoints/meanflownft/meanflownft/generator_ema.pt \
    num_steps=4 \
    eval_reward=false \
    --prompt "a cinematic photo of a red panda"
```

The first two adapters are merged in order. The final MeanFlowNFT adapter stays
active and supplies `delta_embedder`, matching training-time evaluation. The
released inference config generates 1024×1024 images with CFG-free flow-map
sampling.

> [!IMPORTANT]
> Do not load only the final checkpoint and do not use a generic LoRA loader
> that discards non-LoRA tensors. Either mistake drops part of the trained
> policy.

<!-- -->

> [!WARNING]
> These `.pt` files use PyTorch serialization. Load checkpoints only from a
> trusted source.

## 📄 Citation

```bibtex
@misc{huang2026meanflownftbringingforwardprocessrl,
  title  = {MeanFlowNFT: Bringing Forward-Process RL to Average-Velocity Generators},
  author = {Huang, Yushi and Zhou, Xiangxin and Zhang, Jun and Bo, Liefeng and Pang, Tianyu},
  year={2026},
  eprint={2607.15273},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2607.15273}, 
}
```

## 🙌 Acknowledgements

This work builds on
[MeanFlow](https://github.com/Gsunshine/meanflow),
[AnyFlow](https://github.com/NVlabs/AnyFlow),
[DiffusionNFT](https://github.com/NVlabs/DiffusionNFT),
[Diffusers](https://github.com/huggingface/diffusers),
[Transformers](https://github.com/huggingface/transformers), and
[PEFT](https://github.com/huggingface/peft).

## ⚖️ License

The repository metadata and accompanying code are released under Apache 2.0.
The base model and all derivative weights remain subject to the base model's
license and usage terms; review them before use or redistribution.