Open ghost opened 4 years ago
Is there an error in the training loop code for playing Atari-Freeway: specifically generating the predictions?
_pred2_batch = dist_dqn(state2_batch.detach(),theta2,aspace=aspace)
Should the state2_batch be state_batch?
I believe the original is correct. Does it not work for you?
Is there an error in the training loop code for playing Atari-Freeway: specifically generating the predictions?
_pred2_batch = dist_dqn(state2_batch.detach(),theta2,aspace=aspace)
Should the state2_batch be state_batch?