search

CherryPieSexy / imitation_learning

PyTorch implementation of some reinforcement learning algorithms: A2C, PPO, Behavioral Cloning from Observation (BCO), GAIL.
139 stars 15 forks source link

Image and dict observation spaces support by BCO and GAIL implementations #2

Closed akmandor closed 2 years ago

akmandor commented 2 years ago

You mentioned in the README

as: Behavioral Cloning from Observation (BCO) - technique to clone expert behavior into agent using only expert states, BCO (works bad for me, not supported now)

  1. What do you mean by "works bad for me"? Does that specific to your experience OR Does the current BCO library (functions, data structures, etc) not fully functional?
  2. Do the current BCO and GAIL support image and dict observation spaces?
CherryPieSexy commented 2 years ago

Hello!

  1. It means both - during my initial experiments I was able to solve only CartPole environment using BCO, and after that I decided to focus on GAIL. I believe BCO code requires some modifications to run and now it is not fully functional.
  2. GAIL supports image and dict observation spaces.
CherryPieSexy commented 2 years ago

@akmandor hello! Did I answer your question? Can I close the issue?

akmandor commented 2 years ago

Yeah, thanks for the answer!