LAMP

This is the official implementation of the paper: "Layerwise Sparsity for Magnitude-based Pruning", ICLR 2021.

• The key file is the tools/pruners.py, where we implement various layerwise sparsity determination methods for the magnitude-based pruning.
• Run iterate.py to run simulations.
• This codebased only contains CIFAR-10 experiments in the paper. To add more experiments, you may want to add models to tools/models/, datasets to tools/datasets/ and tools/dataloaders.py, hyperparameter setups to tools/modelloaders.py.

Enjoy!
Best,
Authors.