I did a training session that ended with network-480000.meta.
The deep_q_pong_networks folder has all the files you would assume it has after training that many iterations.
However, when resuming the training, the AI seems to be starting over from scratch.
Is this a bug, user error, or intended?
Loving this tutorial!
Thanks!