TorchRL is a Deep Reinforcement Learning (RL), Offline RL, Imitation Learning and Multi Agent RL algorithms re-implementation using pytorch as simple and standalone python notebooks.
List of implemented algorithms
Algorithm | Notebook |
---|---|
Proximal Policy Gradient (PPO) | ppo_atari.ipynb |
Deep Q-Learning (DQN) | dqn.ipynb |
:TODO:
List of environments used.
:TODO:
The development docker uses torch-jupyter as base image so running docker-buid and docker-run will setup everything.