Awesome

Network Pruning via Performance Maximization

PyTorch Implementation of Network Pruning via Performance Maximization (CVPR 2021).

Requirements

pytorch==1.7.1
torchvision==0.8.2

Usage

To train a base model

CUDA_VISIBLE_DEVICES=0 python train_model.py

To train the pruning algorithm

CUDA_VISIBLE_DEVICES=0 python resnet_pm.py --epm_flag True --nf 15 --reg_w 2 --base 3.0

To prune the model

python pruning_resnet.py 

To finetune the model

CUDA_VISIBLE_DEVICES=0 python train_model.py --train_base False

Citation

If you found this repository is helpful, please consider to cite our paper:

@InProceedings{Gao_2021_CVPR,
    author    = {Gao, Shangqian and Huang, Feihu and Cai, Weidong and Huang, Heng},
    title     = {Network Pruning via Performance Maximization},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2021},
    pages     = {9270-9280}
}