---
title: DiT-XL-2-512
canonical_url: "https://www.modelscope.cn/models/facebook/DiT-XL-2-512"
md_url: "https://www.modelscope.cn/models/facebook/DiT-XL-2-512.md"
repository: facebook/DiT-XL-2-512
last_updated: 2025-05-20
license: cc-by-nc-4.0
library_name:
  - pytorch
frameworks:
  - pytorch
downloads: 381
stars: 1
---

# DiT-XL-2-512

> DiT-XL-2-512 - facebook 在 ModelScope 开源的模型。Scalable Diffusion Models with Transformers (DiT)

facebook/DiT-XL-2-512 是 ModelScope 魔搭社区上的机器学习模型，采用 cc-by-nc-4.0 许可。

- **Repository**: facebook/DiT-XL-2-512
- **License**: cc-by-nc-4.0
- **Downloads**: 381
- **Stars**: 1
- **Last updated**: 2025-05-20

Source: https://www.modelscope.cn/models/facebook/DiT-XL-2-512

---

# Scalable Diffusion Models with Transformers (DiT)

## Abstract

We train latent diffusion models, replacing the commonly-used U-Net backbone with a transformer that operates on latent patches. We analyze the scalability of our Diffusion Transformers (DiTs) through the lens of forward pass complexity as measured by Gflops. We find that DiTs with higher Gflops---through increased transformer depth/width or increased number of input tokens---consistently have lower FID. In addition to good scalability properties, our DiT-XL/2 models outperform all prior diffusion models on the class-conditional ImageNet 512×512 and 256×256 benchmarks, achieving a state-of-the-art FID of 2.27 on the latter.