Elegantrl: Build deep reinforcement-learning agents

Understand the project

Practical overview

Use Elegantrl when your work involves developing deep RL agents in Python and you have a simulation or benchmark environment to train against.

Developers building reinforcement-learning agents for custom simulations.
Researchers and practitioners comparing model-free RL algorithms on shared tasks.
Teams exploring parallel simulation and training workflows.

Implement agents using supported continuous-action, discrete-action or multi-agent algorithms.
Train and compare agents on simulation or benchmark tasks.
Use parallel simulations to increase sampling for deep reinforcement-learning experiments.

Before you choose

The repository lists Python 3.6+ and PyTorch 1.6+ as necessary requirements.
Environment packages are task-specific: Gym, PyBullet, Box2D and Matplotlib are listed for particular purposes.
The documented simulator set includes Isaac Gym for massively parallel simulation and Gym, MuJoCo, PyBullet and FinRL for benchmarking.
The project is a developer framework that requires a user-provided task and environment.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install the documented Python and PyTorch prerequisites, plus optional packages used for environments and plots.

Install Python 3.6 or later and PyTorch 1.6 or later.
For the listed gym environment and plotting dependencies, run the command below, or install from requirements.txt.
Documented command
```
pip3 install gym==0.17.0 pybullet Box2D matplotlib
```
If using a StarCraft II environment, run the command below, then run the command below.
Documented command
```
bash ./elegantrl/envs/installsc2.sh
```
Documented command
```
pip install -r sc2_requirements.txt
```

The documented core prerequisites and selected optional environment dependencies are installed.

The README identifies NumPy as unnecessary for the core setup and notes it is installed with PyTorch; gym, PyBullet, Box2D, and matplotlib are optional.
This covers dependency setup only; it does not document launching or training an agent.
The README notes an env.render() bug with gym==0.18 and pyglet==1.6, and recommends gym==0.17.0 and pyglet==1.5.

Evidence and transparency

Source record details

Source availability: Active
Sources checked: 18 October 2026
Original website title: AI4Finance-Foundation/ElegantRL · Official homepage title · retained in its source language