Back to Home
arXiv AI··Papers & Tech

Resource-Efficient Reinforcement Learning Post-Training through Elastic Expert Parallelism

中文摘要

该研究提出弹性专家并行技术,通过优化专家模型在强化学习训练中的分配效率,显著减少了计算资源浪费并提升了大型语言模型的训练速度。

English Summary

Elastic Expert Parallelism optimizes expert allocation in reinforcement learning post-training, significantly reducing computational bottlenecks and improving efficiency by minimizing idle worker time during large language model training.

Original Excerpt

arXiv:2611.06984v1 Announce Type: new Abstract: Reinforcement learning post-training improves the capabilities of large language models, but generating training responses is a major computational bottleneck. In mixture-of-experts models, workers that finish their assigned responses may remain occupied because their experts are still needed by workers processing longer responses. Consequently, approximately one-third of occupied worker time in the baseline occurs after local response generation finishes. Here we present FlexFloor, a system that enables earlier resource release during synchronous response generation while preserving the expert computation required by ongoing responses. FlexFloor jointly adapts response scheduling and the active group of workers. It uses historical response lengths and memory constraints to guide batching and prompt placement, then progressively releases workers while maintaining expert availability and correct communication. Compared with the verl baseline, FlexFloor generates approximately 65.40% more response tokens per unit of occupied computing time and increases response generation throughput by 52.74%, with little observed variation in rollout …