search

eric-mitchell / macaw

Code for "Offline Meta-Reinforcement Learning with Advantage Weighting" [ICML 2021]
45 stars 11 forks source link

Add entropy maximizing term #5

Closed danielhavir closed 1 year ago

danielhavir commented 1 year ago

The entropy maximizing term is inspired by this torch implementation of SAC: https://github.com/pranz24/pytorch-soft-actor-critic/blob/master/sac.py