search

Denys88 / rl_games

RL implementations
MIT License
941 stars 155 forks source link

Are visual agents supported? #214

Closed davidoort closed 1 year ago

davidoort commented 1 year ago

Hi, as far as I can tell, the network architectures for the actor and critic in RL games are all MLPs, but I was wondering if there's a branch which contain encoders that are learnt by the agent and critic during training.

Thanks!

ViktorM commented 1 year ago

Hi @davidoort,

We have a number of examples available, that are not based only on MLPs. For example: https://github.com/Denys88/rl_games/blob/master/rl_games/configs/atari/ppo_pong_envpool.yaml or https://github.com/Denys88/rl_games/blob/master/rl_games/configs/atari/ppo_breakout_envpool_resnet.yaml

davidoort commented 1 year ago

Super cool! Thanks a lot ❤️