Accelerate CNNs from Three Dimensions: A Comprehensive Pruning Framework

Authors: Wenxiao Wang, Minghao Chen, Shuai Zhao, Long Chen, Jinming Hu, Haifeng Liu, Deng Cai, Xiaofei He, Wei Liu

ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments are conducted to show the superiority of our proposed pruning algorithm over the state-of-the-art pruning algorithms.
Researcher Affiliation Collaboration 1State Key Lab of CAD&CG, Zhejiang University, China 2Tencent Data Platform, China 3Columbia University, US 4Tencent, China.
Pseudocode Yes Algorithm 1 Iterative Pruning and Fine-tuning
Open Source Code No The paper does not provide any statement about making its source code publicly available or link to a code repository.
Open Datasets Yes Datasets: We take three popular datasets as testbeds of our algorithm: CIFAR-10 (Krizhevsky et al., 2009), Tiny Image Net (Wu et al., 2017), and Image Net (Russakovsky et al., 2015).
Dataset Splits No The paper mentions datasets like CIFAR-10, Tiny Image Net, and Image Net which have standard splits, but it does not explicitly provide details on how the training, validation, and test splits were performed for the models, such as percentages or sample counts.
Hardware Specification No The paper does not provide specific details about the hardware used for running experiments, such as GPU or CPU models.
Software Dependencies No The paper mentions 'Py Torch (Paszke et al., 2017)' but does not provide a specific version number or other software dependencies with version information.
Experiment Setup Yes For base models trained on CIFAR-10, we set batch size to 64 for Dense Net and 128 for Res Net, respectively. Weight decay is set to 10^-4. The models are trained for 160 epochs with the learning rate starting from 0.1 and divided by 10 at epochs 80 and 120. ... The MAP s hyperparameters are set to R = 1 and K = 3 in our pruning experiments. When collecting training data... the model is pruned along each dimension for four times (i.e., rds = 4 in Algorithm 1).