rlmesh 0.1.0b1

Python SDK for model-environment evaluation workflows

RLMesh

RLMesh is a Python SDK for model-environment evaluation workflows. It can serve Gymnasium-style environments, connect clients over local or remote transports, and adapt values for plain Python, NumPy, and Torch users.

Beta: APIs and package structure may change before the stable release.

Installation

Install the published beta from PyPI:

pip install --pre rlmesh

Install optional adapters as needed:

pip install --pre "rlmesh[numpy]"
pip install --pre "rlmesh[gymnasium]"
pip install --pre "rlmesh[torch]"

Quickstart

Install RLMesh with Gymnasium support:

pip install --pre "rlmesh[gymnasium]"

In one process, serve any Gymnasium-compatible environment:

import gymnasium as gym
import rlmesh

env = gym.make("CartPole-v1")
rlmesh.EnvServer(env, "127.0.0.1:5555").serve()

In another process, connect to it as a remote environment:

from rlmesh.numpy import RemoteEnv

env = RemoteEnv("127.0.0.1:5555")
observation, info = env.reset(seed=0)

terminated = truncated = False
while not (terminated or truncated):
    action = env.action_space.sample()
    observation, reward, terminated, truncated, info = env.step(action)

env.close()

Runnable examples and exact commands live in the repository under examples/python.

Links

• Homepage: https://rlmesh.dev
• Documentation: https://docs.rlmesh.dev
• Repository: https://github.com/ArenaX-Labs/rlmesh
• Examples: https://github.com/ArenaX-Labs/rlmesh/tree/main/examples/python
• Issues: https://github.com/ArenaX-Labs/rlmesh/issues
• Default contact: [e-mail]

License

Licensed under either MIT or Apache-2.0.