Parameterized data set optimization to evaluate limits of neural models and mechanisms Requires python 3
conda env create -f environment.yml adv source activate adv
- psql adv -h 127.0.0.1 -d adv
python db_tools.py --exp=experiments/psvrt.yaml --init
python db_tools.py --grads --exp=experiments/gradients.yaml --reset_grads
CUDA_VISIBLE_DEVICES=5 python generate_samples.py --param_path=results/biggan_resnet18_2020-01-10_18:28:36_final_params.npz
CUDA_VISIBLE_DEVICES=-1 python generate_interpolation.py --param_path=results/biggan_resnet50_2020-01-16-16_39_13_final_params.npz --cat=gorilla --steps=10 --n=5 CUDA_VISIBLE_DEVICES=-1 python generate_interpolation.py --param_path=results/biggan_resnet50_2020-01-16-16_39_13_final_params.npz --cat=fly --steps=10 --n=5