First download the full_fisher_dict.pkl file.

wget https://huggingface.co/ajrheng/selective-amnesia/resolve/main/full_fisher_dict.pkl

Run train

Create a file, eg, my_trainer.py and use examples and modify your configs to run the file.

Using Unlearn Canvas dataset

from mu.algorithms.selective_amnesia.algorithm import SelectiveAmnesiaAlgorithm
from mu.algorithms.selective_amnesia.configs import (
    selective_amnesia_config_unlearn_canvas,
)

algorithm = SelectiveAmnesiaAlgorithm(
    selective_amnesia_config_unlearn_canvas,
    ckpt_path="models/compvis/style50/compvis.ckpt",
    raw_dataset_dir=(
        "data/quick-canvas-dataset/sample"
    ),
    dataset_type = "unlearncanvas",
    template = "style",
    template_name = "Abstractionism",
    use_sample = True # to run on sample dataset

)
algorithm.run()

Run the script

WANDB_MODE=offline python my_trainer.py

How It Works * Default Values: The script first loads default values from the train config file as in configs section.

• Parameter Overrides: Any parameters passed directly to the algorithm, overrides these configs.

• Final Configuration: The script merges the configs and convert them into dictionary to proceed with the training.

Notes

1. Ensure all dependencies are installed as per the environment file.
2. The training process generates logs in the logs/ directory for easy monitoring.
3. Use appropriate CUDA devices for optimal performance during training.
4. Regularly verify dataset and model configurations to avoid errors during execution.

---