Paper Detail
Hayate Iso, Tiyasa Mitra, Sudipta Mondal, Rasoul Shafipour, Venmugil Elango, Terry Kong, Yuki Huang, Seonjin Na, Izzy Putterman, Benjamin Chislett, Maor Ashkenazi, Joseph Guman, Gerald Shen, Tugrul Konuk, Ashwath Aithal, Ritika Borkar, Ran Zilberstein, Bita Rouhani
RL post-training of frontier language models is increasingly bottlenecked by autoregressive rollout generation, making rollout acceleration a central systems challenge. Many existing efficiency methods improve throughput by changing the rollout or optimization regime, for example, through off-policy execution, replay, or lower-precision generation. We study speculative decoding as a lossless acceleration primitive for RL rollouts that preserves the target model's output distribution. We implement speculative decoding in NeMo-RL with a vLLM backend, supporting both synchronous and asynchronous pipelines and enabling speculation during RL rollouts. This benefit is realizable across speculation mechanisms, such as pretrained MTP heads, small external draft models or even techniques such as Eagle3, which are traditionally applied after RL phase. This yields a deployment path for state-of-the-art speculative decoding inside RL training. In a reasoning post-training workload at 8B scale under synchronous RL, speculative decoding improves rollout throughput by 1.8x. Using a high-fidelity performance simulator, we project that combining speculative decoding with asynchronous RL yields up to 2.5x end-to-end training speedup at 235B scale.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{iso2026accelerating,
title = {Accelerating RL Post-Training Rollouts via System-Integrated Speculative Decoding},
author = {Hayate Iso and Tiyasa Mitra and Sudipta Mondal and Rasoul Shafipour and Venmugil Elango and Terry Kong and Yuki Huang and Seonjin Na and Izzy Putterman and Benjamin Chislett and Maor Ashkenazi and Joseph Guman and Gerald Shen and Tugrul Konuk and Ashwath Aithal and Ritika Borkar and Ran Zilberstein and Bita Rouhani},
year = {2026},
abstract = {RL post-training of frontier language models is increasingly bottlenecked by autoregressive rollout generation, making rollout acceleration a central systems challenge. Many existing efficiency methods improve throughput by changing the rollout or optimization regime, for example, through off-policy execution, replay, or lower-precision generation. We study speculative decoding as a lossless acceleration primitive for RL rollouts that preserves the target model's output distribution. We implemen},
url = {https://huggingface.co/papers/2604.26779},
keywords = {speculative decoding, RL post-training, autoregressive rollout generation, rollout acceleration, vLLM backend, synchronous pipeline, asynchronous pipeline, MTP heads, draft models, Eagle3, performance simulator, end-to-end training speedup, huggingface daily},
eprint = {2604.26779},
archiveprefix = {arXiv},
}
{}