Closed ayushtues closed 1 year ago
This is a great idea! We've provided a PPO example using torchrl, so maybe TRPO would be a good idea. Other possibilities are IQL or decision transformers, for which I see much interest for in the community.
/assigntome
Closing this issue, since we have published the PPO tutorial already.
@jlin27 @brianjo
Currently, there exists only one Reinforcement Learning tutorial on the official PyTorch tutorials page, i.e. the DQN tutorial. Considering the popularity of using PyTorch for RL, me and @hemilpanchiwala wanted to contribute to adding some more RL tutorials to the page. We are planning to start with adding some policy-gradient based approach like TRPO/DDPG and are willing to add more "standard" RL algorithms as well while keeping in mind to be as diverse as possible.
cc @vmoens @nairbv