DDPG-Unity-PyTorch

Overview

An implementation of Deep Deterministic Policy Gradients using PyTorch running running on Unity ML-Agents environments

Dependencies

• PyTorch
• Unity ML-Agents
• Pillow
• Numpy

Visualizations of some trained agents

• Walker environment
  • State space: 212
  • Action space: 39
  • Number of agents: 11

• 3DBall environment
  • State space: 8
  • Action space: 2
  • Number of agents: 12