README.md

This is a repository for PROL : Rehearsal Free Continual Learning in Streaming Data via Prompt Online Learning (ICCV 2025)

Paper link: https://openaccess.thecvf.com/content/ICCV2025/papers/Masum_PROL__Rehearsal_Free_Continual_Learning_in_Streaming_Data_via_ICCV_2025_paper.pdf

Arxiv paper : https://arxiv.org/abs/2507.12305

Example of running script

CIFAR100 dataset

python main_prol.py cifar100_prol --batch-size 10 --num_tasks 10 --data-path ./local_datasets/ --seed 1993 --lr 0.05 --epochs 1 --length 5

python main_prol.py cifar100_prol --batch-size 10 --num_tasks 10 --data-path ./local_datasets/ --seed 1993 --lr 0.01 --epochs 1 --length 5

ImageNet-R dataset

python main_prol.py imr_prol --batch-size 10 --num_tasks 10 --data-path ./local_datasets/ --seed 1993 --lr 0.05 --epochs 1 --length 5

python main_prol.py imr_prol --batch-size 10 --num_tasks 10 --data-path ./local_datasets/ --seed 1993 --lr 0.01 --epochs 1 --length 5

ImageNet-A dataset

python main_prol.py ima_prol --batch-size 10 --num_tasks 10 --data-path ./local_datasets/ --seed 1993 --lr 0.05 --epochs 1 --length 5

python main_prol.py ima_prol --batch-size 10 --num_tasks 10 --data-path ./local_datasets/ --seed 1993 --lr 0.01 --epochs 1 --length 5

CUB dataset

python main_prol.py cub_prol --batch-size 10 --num_tasks 10 --data-path ./local_datasets/ --seed 1993 --lr 0.05 --epochs 1 --length 5

python main_prol.py cub_prol --batch-size 10 --num_tasks 10 --data-path ./local_datasets/ --seed 1993 --lr 0.01 --epochs 1 --length 5

Citation

If you use the codes and models from this repo, please cite our work. Thanks!

@inproceedings{ma2025prol,
  title={PROL: Rehearsal Free Continual Learning in Streaming Data via Prompt Online Learning},
  author={Ma'sum, M Anwar and Pratama, Mahardhika and Ramasamy, Savitha and Liu, Lin and Habibullah, Habibullah and Kowalczyk, Ryszard},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={2471--2481},
  year={2025}
}