This Deep RL course was taught at The African Master's in Machine Intelligence AMMI in Fall 2021. It was instructed by researchers at DeepMind: Bilal Piot, Corentin Tallec and Florian Strub. This project is the coursework of Deep RL where we Catalyst Agents team trying to re-implement RL algorithm(s) for continuous control tasks. We chose three types of environments: easy, medium, and hard to run the algorithm(s). The course project meant to submit only one algorithm, but we plan to continue working on this repo making it an open project by this team of student from AMMI. This is why we're trying to make a modular repo to ease the re-implementation of future algorithms.
Algorithm we re-implementing/plannning to re-implement:
-
Soft Actor-Critic (SAC) Paper (Now)
-
Model-Based Policy Optimization (MBPO) Paper (Next; Future work)
-
Model Predictive Control-Soft Actor Critic (MPC-SAC) Paper (Next; Future work)
-
Model Predictive Actor-Critic (MoPAC) Paper (Next; Future work)
Move into rl-ammi
directory, and then run the following:
conda create -n rl-ammi python=3.8
pip install -e .
pip install numpy
pip install torch
pip install wandb
pip install gym
If you want to run MuJoCo Locomotion tasks, and ShadowHand, you should install MuJoCo first (it's open sourced until 31th Oct), and then install mujoco-py:
sudo apt-get install ffmpeg
pip install -U 'mujoco-py<2.1,>=2.0'
If you are using A local GPU of Nvidia and want to record MuJoCo environments issue link, run:
unset LD_PRELOAD
Move into rl-ammi
directory, and then run the following:
conda create -n rl-ammi python=3.8
pip install -e .
pip install numpy
pip install torch
pip install wandb
pip install gym
If you want to run MuJoCo Locomotion tasks, and ShadowHand, you should install MuJoCo first (it's open sourced until 31th Oct), and then install mujoco-py:
brew install ffmpeg
pip install -U 'mujoco-py<2.1,>=2.0'
If you are using A local GPU of Nvidia and want to record MuJoCo environments issue link, run:
unset LD_PRELOAD
Move into rl-ammi/
directory, and then:
python experiment.py -cfg <cfg_file-.py> -seed <int>
for example:
python experiment.py -cfg sac_hopper -seed 1
To evaluate a saved policy model, run the following command:
python evaluate_agent.py -env <env_name> -alg <alg_name> -seed <int> -EE <int>
for example:
python evaluate_agent.py -env Walker2d-v2 -alg SAC -seed 1 -EE 5
(first name alphabetical order)
- MohammedElfatih Salah
- Rami Ahmed
- Ruba Mutasim
- Wafaa Mohammed
This repo was inspired by many great repos, mostly the following ones: