SLM Lab

Modular Deep Reinforcement Learning framework in PyTorch.

Documentation:
https://slm-lab.gitbook.io/slm-lab/

Steps to install

• Install docker
• Start an Ubuntu 16.04 container

docker run -it --name ubuntu_16_04_slm_lab -e DISPLAY=${DISPLAY} -v /tmp/.X11-unix:/tmp/.X11-unix ubuntu:16.04

• Install conda from https://docs.continuum.io/anaconda/. Put conda executable in the PATH
• conda update conda
• conda init
• Exit shell, reenter shell
• git clone https://github.com/andrei-radulescu-banu/SLM-Lab.git
• cd SLM-Lab/
• ./bin/setup
• conda install tensorflow
• conda activate lab
• pip install future

Examples:

• python andrei/reinforce.py
• python run_lab.py slm_lab/spec/demo.json dqn_cartpole dev
• python run_lab.py slm_lab/spec/experimental/ppo/ppo_eps_search.json ppo_breakout dev
• In general, python run_lab.py {spec_file} {spec_name} {lab_mode}, where lab_mode can be dev|train|search|enjoy@saved_agent. See Sec. 11.3.1 in the book.

Note: You need Ubuntu 16.04. This will not work on Ubuntu 14.04, or 18.04. Nor on Centos 7.

BeamRider	Breakout	KungFuMaster	MsPacman
Pong	Qbert	Seaquest	Sp.Invaders
Ant	HalfCheetah	Hopper	Humanoid
Inv.DoublePendulum	InvertedPendulum	Reacher	Walker