RavnaBergsndot

The loss function

5

There's a difference between reinforcement and supervised learning in the AGZ paper. The paper mentioned that although for the reinforcement version, the loss function is like "loss = action_loss +...