Reproduce “RL Unplugged: A Suite of Benchmarks for Offline Reinforcement Learning” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Caglar Gulcehre, Ziyu Wang, Alexander Novikov, Tom Le Paine, Sergio Gomez Colmenarejo, Konrad Zolna, Rishabh Agarwal, Josh Merel and 10 more
RL Unplugged provides shared datasets, tasks, an API, evaluation protocols, and baselines for offline reinforcement learning. It covers Atari, control, and locomotion tasks, with online and offline policy selection settings. For offline selection, its baseline transfers hyperparameters from online tasks. BC and RABM perform well on DM Locomotion, while BC performs poorly on Atari, in these tested tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.