The actor loss in maddpg seems to be wrong.

starry-sky6688 / MADDPG

Pytorch implementation of the MARL algorithm, MADDPG, which correspondings to the paper "Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments".

516 stars 80 forks source link

The actor loss in maddpg seems to be wrong. #14

Closed verystrongjoe closed 2 years ago

verystrongjoe commented 2 years ago

Here there is no predefined actor_loss variable in advance.

https://github.com/starry-sky6688/MADDPG/blob/b50ca15c1a976eade3153f86852a457159efc686/maddpg/maddpg.py#L90

verystrongjoe commented 2 years ago

my mistake.