README.md

RS-Diffusion

https://arxiv.org/abs/2407.02906

Dataset

Download the RS-Real dataset from HuggingFace

Training

You can try training your own diffusion model using the following command:

python train_RS_real.py --config config/RS_real_config.yaml

Multi-GPU Training

The trainer has been equipped with 🤗Accelerator, please refer to the link if you are not yet familiar with it.

Multi-GPU training can be simply set up by using the following commands:

# First, configure the Accelerator on your own machine
accelerate config

# Then, launch your training
accelerate launch train_RS_real.py --config config/RS_real_config.yaml

Testing

Download the Pre-trained Weights for RS-Real dataset from HuggingFace and put it to RS-Diffusion/checkpoint and run the following command for testing.

python sample_RS_real.py --config config/RS_real_config_teat.yaml