Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Q[option] should also have a detach() when calculating ac_loss #2

Closed
backpropper opened this issue Aug 20, 2020 · 2 comments
Closed

Q[option] should also have a detach() when calculating ac_loss #2

backpropper opened this issue Aug 20, 2020 · 2 comments

Comments

@backpropper
Copy link

option-critic-pytorch/option_critic.py

Line 241 in 0c57da7

policy_loss = -logp * (gt.detach() - Q[option]) - args.entropy_reg * entropy

@lweitkamp
Copy link
Owner

Q is detached in line 230 because it shouldn't be updated in the actor loss in general. Although it is sloppy. I will rewrite the code a bit to make this more clear when I have some time.

@backpropper
Copy link
Author

Ah I see. In that case, the termination loss has extra detaches.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@backpropper @lweitkamp