Reproduce “MAI-Gym: An Open All-in-One Platform for GUI Agent Evaluation and RL Training across Mobile and Computer Environments” 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
Panrong Tong, Quyu Kong, Pengxiang Li, Tianyu Xia, Hanzhang Zhou, Lei Wang, Chenglin Cai, Long Li and 7 more
MAI-Gym provides a shared rollout path for evaluating agents, collecting data, and training GUI reinforcement learning across resettable mobile, computer, and web environments. It also supplies synthesized tasks with checked outcomes. Warm-started with supervised fine-tuning, the 9B agent reaches 78.4% on MobileWorld and 76.2% on OSWorld-Verified on held-out tasks in trained environments; this does not test unseen environments.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.