Instructions for generating the synthetic dataset

After fine-tuning the T2I model, modify run.sh for the appropriate yaml file and run the following

bash run.sh [0-49]

To parallelize generation across GPU, the example batch_submission.py script is provided