Skip to content
No description, website, or topics provided.
Python
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
analysis dummy commit to mask authorship Sep 24, 2019
datasets
experiment Slight refactoring of experiment class Sep 7, 2019
metrics Metrics for FLOPs Sep 19, 2019
models Add cifar-vgg for cifar100 Sep 19, 2019
plot Fix plot imports Sep 19, 2019
pruning Add cifar-vgg for cifar100 Sep 19, 2019
scripts Add resume options to cli script Sep 7, 2019
strategies Add imports for new strategies Sep 19, 2019
util CSVLogger for logging training/finetuning metrics Aug 26, 2019
.gitignore Add resume options to cli script Sep 7, 2019
LICENSE Initial commit Aug 14, 2019
README.md Update README.md Oct 27, 2019

README.md

ShrinkBench

Open source PyTorch library to facilitate development and standardized evaluation of neural network pruning methods.

The modules are organized as follows:

submodule Description
analysis/ Aggregated survey results over 80 pruning papers
datasets/ Standardized dataloaders for supported datasets
experiment/ Main experiment class with the data loading, pruning, finetuning & evaluation
metrics/ Utils for measuring accuracy, model size, flops & memory footprint
models/ Custom architectures not included in torchvision
plot/ Utils for plotting across the logged dimensions
pruning/ General pruning and masking API.
scripts/ Executable scripts for running experiments (see experiment/)
strategies/ Baselines pruning methods, mainly magnitude pruning based

Requirements:

  • PyTorch
  • Torchvision
  • NumPy
  • Pandas
  • Matplotlib
You can’t perform that action at this time.