| Docs: https://torchrl.sanyamkapoor.com | Github: https://github.com/activatedgeek/torchrl |
TorchRL provides highly modular and extensible approach to experimenting with Reinforcement Learning. All this is accessible via a simple programmatic interface with experiment specifications.
- Modularity in the RL pipeline
- Clean implementations of fundamental ideas
- Fast Experimentation
- Scalability
- Low bar and High ceiling
pip install torchrlInstall from source for the latest changes that have not been published to PyPI.
pip install https://github.com/activatedgeek/torchrl/tarball/master