- Accepted to ICCV 2023
- https://arxiv.org/abs/2210.06788
- This repository is the official implementation by the authors.
- Watch our poster, slides, or the 5 minute video presentation of this paper for more details.
- Linux or macOS
- Python 3.10
- CPU compatible but NVIDIA GPU + CUDA CuDNN is highly recommended.
Install required Python packages
pip install requirements.txt
To train the model(s) and evaluate in the paper, run this command:
python main.py -d cifar10 -i 1 -m TiDAL -q Entropy