This repository is an official PyTorch implementation of the ICLR 2025 paper 'Exploring Learning Complexity for Efficient Downstream Dataset Pruning'. The illustration of our algorithm core is shown as below:
pip install -r requirements.txt$ ./scripts/pruning.sh$ ./scripts/tuning.sh
If you find our repository useful for your research, please consider citing our paper:
@inproceedings{
jiang2025exploring,
title={Exploring Learning Complexity for Efficient Downstream Dataset Pruning},
author={Wenyu Jiang and Zhenlong Liu and Zejian Xie and Songxin Zhang and Bingyi Jing and Hongxin Wei},
booktitle={The Thirteenth International Conference on Learning Representations},
year={2025},
url={https://openreview.net/forum?id=FN7n7JRjsk}
}