A small notebook to:
- Preprocess images
- Evaluate how difficult the model grasp the features of your images
- evaluated mainly by perceptual and GAN's discriminator loss.
- Just build
Dockerfile
.
docker build . -t eval_images
- Then, run.
docker run -it --runtime=nvidia --gpus all -d --restart=always eval_images:latest bash
- Create virtual environment from
environment.yaml
.
conda env create -f=environment.yaml
conda activate sd
- Run
setup.sh
.
chmod +x ./setup.sh
./setup.sh