Just like with dbns and autoencoders, we need a script to output a summary of the HF models we have trained.