Channel-based pruning methods

[annahambi]

Dear shrinkbench authors

The channel based pruning strategies seem unsupported. Even though the code is there, it is commented out in the strategies.utils.py and not imported in the strategies/__init__.py. Can you comment on the reason or if there is a plan to activate this strategies?

for strategy in ['WeightNormChannelPruning', 
                'ActivationNormChannelPruning']:

    for  c in [1,2,4,8,16,32,64]:

        exp = PruningExperiment(dataset='MyCIFAR10', 
                                model='my_vgg16',
                                strategy=strategy,
                                compression=c,
                                train_kwargs={'epochs':10}, 
                                save_freq=9, 
                                pretrained=False)
        exp.run()
        clear_output()

==> AttributeError: module 'shrinkbench.strategies' has no attribute 'WeightNormChannelPruning'```