The source code of Machine Unlearning via Null Space Calibration

Content

• main.py : The main function to train original and retrained model
• **.ipynb : The jupyter notebook to visualize the result with its name indicates their experiments purpose.
• agent : The floder contains the agent class
• models : The floder contains the model class
• utils.py : Some help functions
• save : Put the checkpoints to this floder, you can use the provided checkpoints to reproduce the result or train your own model by running main.py or train_base_model.sh