Unofficial PyTorch Implementation of Data-Distortion Guided Self-Distillation for Deep Neural Networks (AAAI 2019)

src/
    └─ model/
        └─ net.py
    ├─ config.py
    ├─ dataset.py
    ├─ main.py
    ├─ trainer.py
    └─ utils.py
scripts/
    ├─ run_baseline.sh
    └─ run_ddgsd.sh
├─ .gitignore
├─ Dockerfile
├─ LICENSE
├─ README.md
└─ requirements.txt

• torch==1.6.0
• torchvision==0.7.0

All dependencies are written in requirements.txt, and you can also access through Dockerfile.

$ sh scripts/run_baseline.sh

$ sh scripts/run_ddgsd.sh

* Hyperparameters of this implementation follow paper settings.
* MMD Loss is replaced with MSE Loss in this implementation.
* For another differences, check this issue.