search

zhejz / carla-roach

Roach: End-to-End Urban Driving by Imitating a Reinforcement Learning Coach. ICCV 2021.
https://zhejz.github.io/roach
Other
274 stars 50 forks source link

Would like some tips and advice to train big Reinforcement Learning agents like these #34

Closed fcharity closed 7 months ago

fcharity commented 7 months ago

Thank you for your wonderful work! I am very much impressed with the consistency at which the RL training progresses and that must have required some phenomenal hyperparameter tuning and debugging skills. I would like to learn to tune hyperparamters for big and dynamic environments like these. I would appreciate it a lot if you could give me some suggestions and tips for hyperparameter tuning and debugging RL training process. Thank you once again! Cheers!

zhejz commented 7 months ago

Thanks! This is a really nice question. First of all, you can start with the famous "The Nuts and Bolts of Deep RL Research" by John Schulman (link). Then personally I'd like to emphasize two things that matter a lot for our specific application autonomous driving: (1) Parallel simulation environment for accelerated experience collection. (2) Logging and visualizing everything you can think of such that later you can debug and reason.