pytorch-a2c-ppo-acktr-gail
pytorch-a2c-ppo-acktr-gail copied to clipboard
Why Openai-Baselines is needed?
One of the attractions of this repository is that we can have a pytorch version without messing with tensorflow. It seems like openai-baselines actually requires tensorflow. Could you explain why the openai-baselines is needed? Or is it possible to get rid of it?
It's not using the tensorflow baselines that's in the openai-baselines, but it's using the helpful environment wrappers (atatri_wrappers
, vec_env
, etc).
The libraries imported from openai-baselines don't use tensorflow, so if you don't want to use openai-baselines, you can probably just use the files that is imported in this project like,
https://github.com/openai/baselines/blob/master/baselines/common/vec_env/vec_env.py
I will accept a PR that import these files as a submodule.