用强化学习动态调整分布式训练批大小，提升效率与精度。
DYNAMIX: RL-based Adaptive Batch Size Optimization in Distributed Machine Learning Systems
- 基于PPO的强化学习框架，根据系统状态自适应调节批大小。
- 相比静态策略，最高提升6.3%模型精度，训练时间减少46%。
- 支持跨模型和集群规模泛化，适合大规模分布式训练场景。
现有分布式机器学习中的批大小选择方法依赖静态分配或简单启发式策略，难以适应异构、动态的计算环境。本文提出DYNAMIX，一个基于强化学习的框架，将批大小优化建模为序列决策问题，采用近端策略优化（PPO）算法。该方法利用包含网络级指标、系统资源利用率及训练统计效率在内的多维状态表示，实现对多样计算资源的智能决策，无需显式系统建模，可无缝集成至现有分布式训练框架中。在多种工作负载、硬件配置和网络条件下评估表明，DYNAMIX 最高可提升6.3%的最终模型精度，并减少46%的总训练时间。可扩展性实验显示，当集群规模增至32节点时，DYNAMIX仍保持最优性能；策略迁移实验也证明，所学策略可在相关模型架构间有效泛化。
原文摘要 · Abstract (English)
Existing batch size selection approaches in distributed machine learning rely on static allocation or simplistic heuristics that fail to adapt to heterogeneous, dynamic computing environments. We present DYNAMIX, a reinforcement learning framework that formulates batch size optimization as a sequential decision-making problem using Proximal Policy Optimization (PPO). Our approach employs a multi-dimensional state representation encompassing network-level metrics, system-level resource utilization, and training statistical efficiency indicators to enable informed decision-making across diverse computational resources. Our approach eliminates the need for explicit system modeling while integrating seamlessly with existing distributed training frameworks. Through evaluations across diverse workloads, hardware configurations, and network conditions, DYNAMIX achieves up to 6.3% improvement in the final model accuracy and 46% reduction in the total training time. Our scalability experiments demonstrate that DYNAMIX maintains the best performance as cluster size increases to 32 nodes, while policy transfer experiments show that learned policies generalize effectively across related model architectures.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。