search

rail-berkeley / rlkit

Collection of reinforcement learning algorithms
MIT License
2.52k stars 553 forks source link

No value function in Soft Actor Critic? #91

Closed KK666-AI closed 4 years ago

KK666-AI commented 4 years ago

Dear author,

In your implementation of soft actor critic, there is no value function V(s)?

In the original paper of SAC, the authors said such value function can stabilize training and is convenient to train simultaneously with other networks.

KK666-AI commented 4 years ago

I find another paper entitled by "Soft Actor-Critic Algorithms and Applications", the author said it was unnecessary to use the additional function approximator for the value function. I think this issue could be closed.