Tonic: A Deep Reinforcement Learning Library for Fast Prototyping and Benchmarking

Deep reinforcement learning has been one of the fastest growing fields of\nmachine learning over the past years and numerous libraries have been open\nsourced to support research. However, most codebases have a steep learning\ncurve or limited flexibility that do not satisfy a need for fast prototyping in\nfundamental research. This paper introduces Tonic, a Python library allowing\nresearchers to quickly implement new ideas and measure their importance by\nproviding: 1) general-purpose configurable modules 2) several baseline agents:\nA2C, TRPO, PPO, MPO, DDPG, D4PG, TD3 and SAC built with these modules 3)\nsupport for TensorFlow 2 and PyTorch 4) support for continuous-control\nenvironments from OpenAI Gym, DeepMind Control Suite and PyBullet 5) scripts to\nexperiment in a reproducible way, plot results, and play with trained agents 6)\na benchmark of the provided agents on 70 continuous-control tasks. Evaluation\nis performed in fair conditions with identical seeds, training and testing\nloops, while sharing general improvements such as non-terminal timeouts and\nobservation normalization. Finally, to demonstrate how Tonic simplifies\nexperimentation, a novel agent called TD4 is implemented and evaluated.\n

Paper

Similar papers

© 2026 NYSGPT2525 LLC