All Tools
P
AgentsFreeOpen Source
PFRL
PyTorch deep reinforcement learning library from Preferred Networks
MIT
ABOUT
Paper RL code is often one-off and hard to reuse across DQN, PPO, and SAC experiments. PFRL ships documented PyTorch implementations with a shared agent API and examples so teams can train published algorithms instead of reimplementing each paper's trainer.
INSTALL
pip install pfrlINTEGRATION GUIDE
1. Train Rainbow DQN or PPO on Atari using PFRL's documented examples
2. Swap SAC and TD3 on a continuous-control task with the same agent interface
3. Reproduce a Preferred Networks baseline before adding a custom network
4. Teach deep RL by starting from PFRL's algorithm implementations rather than scratch code
TAGS
pythonpytorchreinforcement-learningdqnpposacpreferred-networks