This repository provides code for training Diffusion Transformer (DiT) models.
- Improved DiT training script with
accelerate - Supports multiple models.
- Easy to modify the code for your own research.
| Model | Training | Inference |
|---|---|---|
| DiT | ✅ | ✅ |
| SiT | ✅ | ✅ |
| REPA | ✅ | ✅ |
| Large-DiT | ✅ | ✅ |
conda create -n edit python=3.11 -y
conda activate edit
pip install -r requirements.txt- DiT: Scalable Diffusion Models with Transformers
- fast-DiT: Scalable Diffusion Models with Transformers
- SiT: SiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformers
- REPA: Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think
- Large-DiT: Large Diffusion Transformer
- PixArt-α: PixArt-α: Fast Training of Diffusion Transformer for Photorealistic Text-to-Image Synthesis
- LlamaGen: Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation