[NeurIPS2024] SAFE: Slow and Fast Parameter-Efficient Tuning for Continual Learning with Pre-Trained Models

Requirements

• PyTorch >= version 1.4
• tqdm == 4.65.0
• timm == 0.6.5

Datasets

We follow RanPAC setting to use the same data index_list for training.

Running scripts

$ python main.py -d cifar224

• for -d choose from 'cifar224', 'imageneta'

Acknowledgment

Our project references the codes in the following repos.

• RanPAC
• VPT
• AdaptFormer
• SSF

Citation

If you find our work useful in your research, please consider citing:

@inproceedings{zhao2024safe,
  title={SAFE: Slow and Fast Parameter-Efficient Tuning for Continual Learning with Pre-Trained Models},
  author={Zhao, Linglan and Zhang, Xuerui and Yan, Ke and Ding, Shouhong and Huang, Weiran},
  booktitle={Advances in Neural Information Processing Systems (NeurIPS)},
  year={2024}
}