To start the training loop, run: cd floyd_net python dqn.py

Code directory: common: backend API functions to translate graphs between formats pytorch_structure2vec: pytorch implementation of S2V floyd_net: dqn.py: code for main training loop (i.e. get actions, update states, calculate loss, etc.) q_net.py: definition of DQN and full network pipeline message.py: random graph generation and efficiency calculation