DistRL: An Asynchronous Distributed Reinforcement Learning Framework for On-Device Control Agents
Taiyi Wang, Zhihao Wu, Jianheng Liu, Jianye Hao, Jun Wang, Kun Shao
- 🏛 Institutions
- University of Cambridge, Powersense Technology Limited, Huawei Noah's Ark Lab, UCL, Tianjin University
- 📅 Date
- October 18, 2024
- 📑 Publisher
- ICLR 2025 (Poster)
- 💻 Env
- Mobile
- 🔑 Keywords
TLDR
DistRL is a distributed RL fine-tuning framework for mobile control agents that separates centralized training from decentralized data collection across worker devices. It is paired with the A-RIDE off-policy RL algorithm, and the paper reports 3x higher training efficiency, 2.4x faster data collection, and a 20% relative success-rate gain on open Android control tasks.
Related papers
- ClawGUI: A Unified Framework for Training, Evaluating, and Deploying GUI AgentsApril 13, 2026 · arXiv
- Proposer-Agent-Evaluator (PAE): Autonomous Skill Discovery For Foundation Model Internet AgentsDecember 17, 2024 · ICML 2025 (Poster)
- Reinforcement Learning on Web Interfaces Using Workflow-Guided ExplorationFebruary 24, 2018 · ICLR 2018 (Poster)
- Android Coach: Improve Online Agentic Training Efficiency with Single State Multiple ActionsApril 8, 2026 · arXiv
- Don't Act Blindly: Robust GUI Automation via Action-Effect Verification and Self-CorrectionApril 7, 2026 · ACL 2026
- UI-Voyager: A Self-Evolving GUI Agent Learning via Failed ExperienceMarch 25, 2026 · arXiv