rl-transfer-learning

installation on Euler cluster

env2lmod

module load gcc/6.3.0

module load mesa/18.3.6

module load python_gpu/3.8.5

module load eth_proxy

create virtual environment

python3 -m venv RLenv

source RLenv/bin/activate

install MuJoCo

cd $HOME

nano .bashrc

go to the end of file and paste export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HOME/.mujoco/mujoco210/bin

press Ctrl+O then Enter then Ctrl+X

exec bash

curl -O https://mujoco.org/download/mujoco210-linux-x86_64.tar.gz

mkdir .mujoco

tar -xvzf mujoco210-linux-x86_64.tar.gz -C $HOME/.mujoco/

mkdir MuJoCo

cd MuJoCo

git clone https://github.com/openai/mujoco-py/

cd mujoco-py

(in the cloned directory) with RLenv active, run:

pip3 install -r requirements.txt

pip3 install -r requirements.dev.txt

python3 setup.py install --user

install Meta-World

git clone https://github.com/rlworkgroup/metaworld.git

cd metaworld

nano setup.py change in line 8 from mujoco-py<2.1 to mujoco-py<2.2

pip3 install -e .

test if Meta-World is installed correctly in a python console:

python3

import metaworld

print(metaworld.ML1.ENV_NAMES)

if the above command gives errors, install some additional libraries that may be required (ubuntu):

sudo apt-get install libosmesa6-dev

sudo add-apt-repository ppa:jamesh/snap-support 
sudo apt-get update
sudo apt install patchelf

install other requirements

in the project root directory run:

pip3 install -r requirements.txt