How to add seperate reward terms to tensorboard

[P1terQ]

In environment.hpp, i define 4 reward terms and i want to log them in tensorboard. However, in runner.py, it seems that reward, dones = env.step(action) only returns rewards.sum() in each environment. It's inconvenient to tune the reward coefficients . How can i get the seperate values of reward terms instead of their sum? Looking forward to your reply. Much thanks.

[jhwangbo]

I already do this in my research code. I'll add this feature in the next push.

[P1terQ]

Thanks a lot! looking forward to it!

[jhwangbo]

I'll keep it open so that I don't forget to reflect this

[jhwangbo]

I added the feature. Can you pull and test?