paper-with-me

홈 › Papers

Faster Synchronous On-Policy RL via Straggler-Aware Group Sizing

2026-06-01 · Azal Ahmad Khan, Ammar Ahmed, Zeshan Fayyaz, Sheng Di, Mingyi Hong, Ali Anwar arxiv

Synchronous reinforcement learning methods such as Group Relative Policy Optimization (GRPO) provide stable and reproducible on-policy training, but they are highly vulnerable to stragglers, a single unusually long rollout can delay reward computation and parameter updates for the entire group. This problem becomes more severe as group size increases, creating a tension between the benefits of larger groups and the wall-clock cost of synchronization stalls. We propose Straggler-Aware Group Control (SAGC), a dynamic group-size controller that adapts the training group online based on observed rollout behavior. SAGC formulates group-size selection as an online constrained optimization problem, seeking to retain the benefits of larger groups while controlling the long-term rate of straggler events. Across synchronous GRPO and DAPO training, and on top of both vanilla and strong engineered baselines, SAGC consistently reduces straggler incidence and improves wall-clock efficiency while achieving competitive or better training reward. We further show that these gains transfer to final model quality: SAGC is competitive with or better than the strongest static group-size baseline on downstream reasoning benchmarks, and often produces shorter outputs without any explicit length penalty. These results position dynamic group control as a practical way to make synchronous on-policy RL more efficient and robust.

📄 PDF Abstract BibTeX arXiv:2606.02218

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

Stragglers Can Contribute More: Uncertainty-Aware Distillation for Asynchronous Federated Learning

2025-11-25 · Yujia Wang, Fenglong Ma, Jinghui Chen arxiv

Asynchronous federated learning (FL) has recently gained attention for its enhanced efficiency and scalability, enabling local clients to send model updates to the server at their own pace without waiting for slower part…

Federated Learning

FedCompass: Efficient Cross-Silo Federated Learning on Heterogeneous Client Devices using a Computing Power Aware Scheduler

2023-09-26 · Zilinghan Li, Pranshu Chaturvedi, Shilan He, Han Chen 외

Cross-silo federated learning offers a promising solution to collaboratively train robust and generalized AI models without compromising the privacy of local datasets, e.g., healthcare, financial, as well as scientific p…

Federated Learning

FedAT: A High-Performance and Communication-Efficient Federated Learning System with Asynchronous Tiers

2020-10-12 · Zheng Chai, Yujing Chen, Ali Anwar, Liang Zhao 외

Federated learning (FL) involves training a model over massive distributed devices, while keeping the training data localized. This form of collaborative learning exposes new tradeoffs among model convergence speed, mode…

Federated Learning

Faster Distributed Synchronous SGD with Weak Synchronization

2018-01-01 · ICLR 2018 1 · Cong Xie, Oluwasanmi O. Koyejo, Indranil Gupta

Distributed training of deep learning is widely conducted with large neural networks and large datasets. Besides asynchronous stochastic gradient descent~(SGD), synchronous SGD is a reasonable alternative with better con…

Stragglers-Aware Low-Latency Synchronous Federated Learning via Layer-Wise Model Updates

2024-03-27 · Natalie Lang, Alejandro Cohen, Nir Shlezinger

Synchronous federated learning (FL) is a popular paradigm for collaborative edge learning. It typically involves a set of heterogeneous devices locally training neural network (NN) models in parallel with periodic centra…

Federated Learning