Simple multiprocess tool for gym environments
Project description
VectorGym
Multi-process any(most) gym environment. Automatically parallel the given gym environment using multiprocessing; VectorGym forwards all properties and function (not starting with __) of the underlying gym to you.
Quick Start
Check this demo for a skeleton for training using VectorSim. It deals with only running unfinished environments during trajectory collection.
Run gym environment in parallel.
from VectorGym import VectorGym
if __name__ == '__main__':
envs = VectorGym('CartPole-v1', 2)
print(envs.action_space)
print(envs.observation_space)
envs.reset()
for _ in range(500):
envs.render()
actions = envs.action_space.sample()
res = envs.step(actions)
dones = [r[-2] for r in res]
envs.reset(select=dones)
envs.close()
Install
git clone git@github.com:MRzNone/VectorGym.git
cd VectorGym
pip install -e .
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Close
Hashes for VectorGym-0.1.2.linux-x86_64.tar.gz
Algorithm | Hash digest | |
---|---|---|
SHA256 | f285972933b24519a37d7a5aa4057c33dbbb750d5d533f37b578453679650fae |
|
MD5 | 0f2f3df82e241872d0de5c162d61e486 |
|
BLAKE2b-256 | 89a7f1c59034ceb8fa9816f8e53c2b2aa954a4b36db6fd4724de0c90f53500e6 |