Trying out how different pruning methods affect performance of a model trained on MNIST. The methods used are post-training weight and unit pruning.
All the code, including dependancies as well as short description of what I am doing are inside the notebook network_pruning.ipynb which can be viewed by anyone. To run the notebook yourself using your own Google account and replicate my results just zip through the cells in the notebook, all the parameters I used for the experiments are in there.
