Abstract: Gym and the Future of Reinforcement Learning
This talk will overview the past, present, and future of Gym, the most installed open source reinforcement learning library in the world which serves a role that's analogous to "HTTP for RL", and how Gym has and hopefully will continue to shape the field of reinforcement learning. We will specifically discuss the role of hardware accelerated environments to achieve 100x the performance, the role of standardization of reinforcement learning benchmarks, improved vector environment creation, the new API design and the countless quality of life improvements for researchers around the world. For more information, please see Gym's website at https://www.gymlibrary.ml/.
Intermediate knowledge of python and basics of reinforcement learning
Bio: Jordan is a PhD student at the University of Maryland College Park, and the founder and CEO of the farama foundation. Jordan is the maintainer of Gym, the largest reinforcement learning library in the world, and the maintainer and creator of PettingZoo, which serves an analogous role to Gym in multi-agent reinforcement learning.