Add the extra .py files (from Nanda's work) to the workspace in google collab and it will easily run. 

This is literally Nanda's collab and we just added our methodology to the end of it to show the simple neuron assumption is learned by neurons in layer 1. Then we showed via CKA and RSM, that the representations are discs in layer 1 and clocks in layer 2.

Thanks again.