search

araffin / rl-baselines-zoo

A collection of 100+ pre-trained RL agents using Stable Baselines, training and hyperparameter optimization included.
https://stable-baselines.readthedocs.io/
MIT License
1.12k stars 208 forks source link

Add support for CartPoleContinuousBulletEnvs-v0 #62

Closed araffin closed 4 years ago

araffin commented 4 years ago

closes #61

@erwincoumans this is the same PR (your commit is still here) but with trained model (i also updated the params for the discrete case). I need to update the docker image too before merging.

Colab: https://colab.sandbox.google.com/drive/15JSROMJbeiqxcUwifPR2NYeeFBKmyIlX#scrollTo=E2eWDjPZsQc5

erwincoumans commented 4 years ago

Thanks. Do you have some commands to train and enjoy, assuming this CL would be part of pypi (pip install pybullet)? For example, for the deep_mimic example (we still need to turn this into a proper OpenAI gym environment) pip3 install pybullet python -m pybullet_envs.deep_mimic.testrl --arg_file run_humanoid3d_walk_args.txt

araffin commented 4 years ago

Do you have some commands to train and enjoy

you mean 

python train.py --algo ppo2 --env CartPoleBulletEnv-v1

and

python enjoy.py --algo ppo2 --env CartPoleBulletEnv-v1

or

python enjoy.py --algo ppo2 --env CartPoleBulletEnv-v1 -f logs/ --exp-id 0

as in the readme?

erwincoumans commented 4 years ago

Actually my remark was for the new train/enjoy in pybullet_envs, ignore it here: https://github.com/bulletphysics/bullet3/pull/2565