---
license: other
license_name: snap-non-commercial
license_link: LICENSE
pipeline_tag: unconditional-image-generation
---

This repository contains the checkpoints for the official implementation of **AlphaFlow**, as described in the paper [AlphaFlow: Understanding and Improving MeanFlow Models](https://huggingface.co/papers/2510.20771).

The code is available at [https://github.com/snap-research/alphaflow](https://github.com/snap-research/alphaflow).

![Comparison image from AlphaFlow GitHub README](https://github.com/snap-research/alphaflow/raw/main/figures/teaser.jpg)
*Uncurated* samples (seeds 1-8) from the DiT-XL model for MeanFlow [1] and $\alpha$-Flow (our proposed method) produced with 1 (upper) and 2 (lower) sampling steps for ImageNet-1K $256^2$.