paper-with-me

Papers

DistRL: An Asynchronous Distributed Reinforcement Learning Framework for On-Device Control Agents

2024-10-18 · Taiyi Wang, Zhihao Wu, Jianheng Liu, Jianye Hao, Jun Wang, Kun Shao

On-device control agents, especially on mobile devices, are responsible for operating mobile devices to fulfill users' requests, enabling seamless and intuitive interactions. Integrating Multimodal Large Language Models (MLLMs) into these agents enhances their ability to understand and execute complex commands, thereby improving user experience. However, fine-tuning MLLMs for on-device control presents significant challenges due to limited data availability and inefficient online training processes. This paper introduces DistRL, a novel framework designed to enhance the efficiency of online RL fine-tuning for mobile device control agents. DistRL employs centralized training and decentralized data acquisition to ensure efficient fine-tuning in the context of dynamic online interactions. Additionally, the framework is backed by our tailor-made RL algorithm, which effectively balances exploration with the prioritized utilization of collected data to ensure stable and robust training. Our experiments show that, on average, DistRL delivers a 3X improvement in training efficiency and enables training data collection 2.4X faster than the leading synchronous multi-machine methods. Notably, after training, DistRL achieves a 20% relative improvement in success rate compared to state-of-the-art methods on general Android tasks from an open benchmark, significantly outperforming existing approaches while maintaining the same training time. These results validate DistRL as a scalable and efficient solution, offering substantial improvements in both training efficiency and agent performance for real-world, in-the-wild device control tasks.

📄 PDF Abstract BibTeX arXiv:2410.14803

Code (1)

distrl-lab/distrl-open pytorch

Similar Papers 제목 키워드 기반

More Benefits of Being Distributional: Second-Order Bounds for Reinforcement Learning

2024-02-11 · Kaiwen Wang, Owen Oertell, Alekh Agarwal, Nathan Kallus 외

In this paper, we prove that Distributional Reinforcement Learning (DistRL), which learns the return distribution, can obtain second-order bounds in both online and offline RL in general settings with function approximat…

Distributional Reinforcement LearningMulti-Armed BanditsOffline RL

LlamaRL: A Distributed Asynchronous Reinforcement Learning Framework for Efficient Large-scale LLM Trainin

2025-05-29 · Bo Wu, Sid Wang, Yunhao Tang, Jia Ding 외

Reinforcement Learning (RL) has become the most effective post-training approach for improving the capabilities of Large Language Models (LLMs). In practice, because of the high demands on latency and memory, it is parti…

GPUReinforcement Learning (RL)

The Benefits of Being Distributional: Small-Loss Bounds for Reinforcement Learning

2023-05-25 · NeurIPS 2023 11 · Kaiwen Wang, Kevin Zhou, Runzhe Wu, Nathan Kallus 외

While distributional reinforcement learning (DistRL) has been empirically effective, the question of when and why it is better than vanilla, non-distributional RL has remained unanswered. This paper explains the benefits…

Distributional Reinforcement LearningOffline RLreinforcement-learning

Efficient and Light-Weight Federated Learning via Asynchronous Distributed Dropout

2022-10-28 · Chen Dun, Mirian Hipolito, Chris Jermaine, Dimitrios Dimitriadis 외

Asynchronous learning protocols have regained attention lately, especially in the Federated Learning (FL) setup, where slower clients can severely impede the learning process. Herein, we propose \texttt{AsyncDrop}, a nov…

Federated Learning

Device Scheduling and Update Aggregation Policies for Asynchronous Federated Learning

2021-07-23 · Chung-Hsuan Hu, Zheng Chen, Erik G. Larsson

Federated Learning (FL) is a newly emerged decentralized machine learning (ML) framework that combines on-device local training with server-based model synchronization to train a centralized ML model over distributed nod…

Federated LearningScheduling