PARL/examples/QuickStart

• Quick Start
• How to use
• • Dependencies:
  • Start Training:
  • Expected Result

Quick Start

Train an agent with PARL to solve the CartPole problem, a classical benchmark in RL.

How to use

Dependencies:

• paddlepaddle>=1.5.1
• parl
• gym

Start Training:

# Install dependencies
pip install paddlepaddle  
# Or use Cuda: pip install paddlepaddle-gpu

pip install gym
git clone https://github.com/PaddlePaddle/PARL.git
cd PARL
pip install .

# Train model
cd examples/QuickStart/
python train.py  

Expected Result

The agent can get around 200 points in a few minutes.