Accelerating Long-Tail Generation in Synchronous RLHF Training via Adaptive Tensor Parallelism

Computer Science > Artificial Intelligence [Submitted on 3 May 2026] Accelerating Long-Tail Generation in Synchronous RLHF Training via Adaptive Tensor Parallelism Long Zhao, Qinghe Wang, Jiaan Zhu, Youhui Bai, Zewen Jin, Chaoyi Ruan, Shengnan Wang, Cheng Li Reinforcement Learning from Human Feedback (RLHF) has become a key post-training paradigm for improving model quality. However, the synchronous three-stage RLHF pipeline is often bottlenecked by the generation stage, where response-length skew causes the effective batch size to shrink rapidly during decoding, leaving GPUs underutilized while a few long responses remain unfinished. Mainstream frameworks employ a static tensor parallelism (TP) configuration that cannot adapt to changing batch characteristics, leaving substantial performance headroom unexplored. We propose PAT, an adaptive TP method that dynamically reconfigures TP during the generation stage of each RLHF iteration. PAT introduces two key techniques. First, a predictor-guided online reconfiguration method decides both the reconfiguration point and the target TP configuration based on offline profiling, triggering reconfiguration only when the predicted latency benefit outweighs the reconfiguration overhead. Second, a lightweight online reconfiguration mechanism updates only the states and layouts affected by TP changes: it adapts unfinished decoding states through a cost-model-based choice between KV-cache migration and recomputation, performs in-place weight resharding, and reuses cached communication groups. We implement PAT on top of SGLang and integrate it with the VeRL framework. Evaluations on LLaMA3.1-8B and Qwen3-14B using DeepScaleR show that PAT reduces generation latency by up to 34.6% and end-to-end RLHF training iteration latency by up to 27.2% compared to the original VeRL setup. Comments: 11page, 14 figures Subjects: Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC) Cite as: arXiv:2605.23945 [cs.AI]   (or arXiv:2605.23945v1 [cs.AI] for this version)   https://doi.org/10.48550/arXiv.2605.23945 Focus to learn more Submission history From: Long Zhao [view email] [v1] Sun, 3 May 2026 05:53:32 UTC (753 KB) Access Paper: HTML (experimental) view license Current browse context: cs.AI < prev   |   next > new | recent | 2026-05 Change to browse by: cs cs.DC References & Citations NASA ADS Google Scholar Semantic Scholar Export BibTeX Citation Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Demos Related Papers About arXivLabs Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)