You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In the paper, we could see the loss during the supervising learning should be the KL divergence of between the trajectories and the neural network output:
However, in the code, why it becomes the || label - output ||2?
Is it because of the convenience of implementation?
The text was updated successfully, but these errors were encountered:
In the paper, we could see the loss during the supervising learning should be the KL divergence of between the trajectories and the neural network output:
However, in the code, why it becomes the || label - output ||2?
Is it because of the convenience of implementation?
The text was updated successfully, but these errors were encountered: