A Closer Look at Structured Pruning for Neural Network Compression

Last update: Dec 05, 2022

Related tags

Overview

A Closer Look at Structured Pruning for Neural Network Compression

Code used to reproduce experiments in https://arxiv.org/abs/1810.04622.

To prune, we fill our networks with custom MaskBlocks, which are manipulated using Pruner in funcs.py. There will certainly be a better way to do this, but we leave this as an exercise to someone who can code much better than we can.

Setup

This is best done in a clean conda environment:

conda create -n prunes python=3.6
conda activate prunes
conda install pytorch torchvision -c pytorch

Repository layout

-train.py: contains all of the code for training large models from scratch and for training pruned models from scratch
-prune.py: contains the code for pruning trained models
-funcs.py: contains useful pruning functions and any functions we used commonly

CIFAR Experiments

First, you will need some initial models.

To train a WRN-40-2:

python train.py --net='res' --depth=40 --width=2.0 --data_loc=
   
     --save_file='res'

The default arguments of train.py are suitable for training WRNs. The following trains a DenseNet-BC-100 (k=12) with its default hyperparameters:

python train.py --net='dense' --depth=100 --data_loc=
   
     --save_file='dense' --no_epochs 300 -b 64 --epoch_step '[150,225]' --weight_decay 0.0001 --lr_decay_ratio 0.1

These will automatically save checkpoints to the checkpoints folder.

Pruning

Once training is finished, we can prune our networks using prune.py (defaults are set to WRN pruning, so extra arguments are needed for DenseNets)

python prune.py --net='res'   --data_loc=
   
     --base_model='res' --save_file='res_fisher'
python prune.py --net='res'   --data_loc=
    
      --l1_prune=True --base_model='res' --save_file='res_l1'

python prune.py --net='dense' --depth 100 --data_loc=
     
       --base_model='dense' --save_file='dense_fisher' --learning_rate 1e-3 --weight_decay 1e-4 --batch_size 64 --no_epochs 2600
python prune.py --net='dense' --depth 100 --data_loc=
      
        --l1_prune=True --base_model='dense' --save_file='dense_l1'  --learning_rate 1e-3 --weight_decay 1e-4 --batch_size 64  --no_epochs 2600

Note that the default is to perform Fisher pruning, so you don't need to pass a flag to use it.
Once finished, we can train the pruned models from scratch, e.g.:

python train.py --data_loc=
   
     --net='res' --base_file='res_fisher_
    
     _prunes' --deploy --mask=1 --save_file='res_fisher_
     
      _prunes_scratch'

Each model can then be evaluated using:

python train.py --deploy --eval --data_loc=
   
     --net='res' --mask=1 --base_file='res_fisher_
    
     _prunes'

Training Reduced models

This can be done by varying the input arguments to train.py. To reduce depth or width of a WRN, change the corresponding option:

python train.py --net='res' --depth=
   
     --width=
    
      --data_loc=
     
       --save_file='res_reduced'

To add bottlenecks, use the following:

python train.py --net='res' --depth=40 --width=2.0 --data_loc=
   
     --save_file='res_bottle' --bottle --bottle_mult

With DenseNets you can modify the depth or growth, or use --bottle --bottle_mult as above.

Acknowledgements

Jack Turner wrote the L1 stuff, and some other stuff for that matter.

Code has been liberally borrowed from many a repo, including, but not limited to:

https://github.com/xternalz/WideResNet-pytorch
https://github.com/bamos/densenet.pytorch
https://github.com/kuangliu/pytorch-cifar
https://github.com/ShichenLiu/CondenseNet

Citing this work

If you would like to cite this work, please use the following bibtex entry:

@article{crowley2018pruning,
  title={A Closer Look at Structured Pruning for Neural Network Compression},
  author={Crowley, Elliot J and Turner, Jack and Storkey, Amos and O'Boyle, Michael},
  journal={arXiv preprint arXiv:1810.04622},
  year={2018},
  }

A Closer Look at Structured Pruning for Neural Network Compression

Related tags

Overview

A Closer Look at Structured Pruning for Neural Network Compression

Setup

Repository layout

CIFAR Experiments

Pruning

Training Reduced models

Acknowledgements

Citing this work

Owner

Bayesian and Neural Systems Group

PyTorch Extension Library of Optimized Scatter Operations

TorchShard is a lightweight engine for slicing a PyTorch tensor into parallel shards

A code copied from google-research which named motion-imitation was rewrited with PyTorch

GPU-accelerated PyTorch implementation of Zero-shot User Intent Detection via Capsule Neural Networks

Model summary in PyTorch similar to `model.summary()` in Keras

PyTorch wrappers for using your model in audacity!

Fast Discounted Cumulative Sums in PyTorch

A tutorial on "Bayesian Compression for Deep Learning" published at NIPS (2017).

On the Variance of the Adaptive Learning Rate and Beyond

Pretrained ConvNets for pytorch: NASNet, ResNeXt, ResNet, InceptionV4, InceptionResnetV2, Xception, DPN, etc.

A simplified framework and utilities for PyTorch

A Pytorch Implementation for Compact Bilinear Pooling.

OptNet: Differentiable Optimization as a Layer in Neural Networks

Code for paper "Energy-Constrained Compression for Deep Neural Networks via Weighted Sparse Projection and Layer Input Masking"

Reformer, the efficient Transformer, in Pytorch

Kaldi-compatible feature extraction with PyTorch, supporting CUDA, batch processing, chunk processing, and autograd

PyTorch implementations of normalizing flow and its variants.

Official implementations of EigenDamage: Structured Pruning in the Kronecker-Factored Eigenbasis.

An optimizer that trains as fast as Adam and as good as SGD.

A pure Python implementation of Compact Bilinear Pooling and Count Sketch for PyTorch.