- clone the repository
git clone ... - Install the environment with conda
conda env create -f environment.yml - Activate environment
conda activate projectMSGAI - You should be set up.
Factors: "batch_size", "num_inf_steps", "image_size"
Levels: levels = [[8, 2], [25, 15], [(512, 512), (320, 320)]]
The experiments are stored in the following path with the following format:
experiments/inference_data/repetition_{repetition}__batch_size_{batch_size}__num_inf_steps_{num_inf_steps}__image_size_{image_size}/
In the experiments folder you find an image grid of samples generated and a result.json with some relevant data from the experiment like clip score and runtime.