EfficientRollout: System-Aware Self-Speculative Decoding for RL Rollouts

• URL: http://arxiv.org/abs/2606.18967v1
• Date: Wed, 17 Jun 2026 11:51:06 GMT
• Title: EfficientRollout: System-Aware Self-Speculative Decoding for RL Rollouts
• Abstract summary: We present EfficientRollout, a system-aware self-SD framework designed to address this gap for RL rollouts.<n> EfficientRollout reduces rollout and end-to-end latency by up to 19.6% and 12.7%, respectively, over an accelerated AR rollout baseline.
• Score: 20.460789503141466
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Reinforcement learning (RL) has become a representative post-training paradigm for LLMs, enabling strong reasoning and agentic capabilities. However, rollout generation remains a dominant latency bottleneck because autoregressive sampling decodes responses sequentially and a small number of long-tailed generations often determine completion time. Speculative decoding (SD) offers a natural way to address this bottleneck, as it is a well-established technique for serving fixed LLMs that reduces latency by rapidly drafting tokens and accepting them through parallel verification while preserving the target-model distribution. However, its practical speedups do not directly carry over to RL rollouts: (i) the evolving target policy makes any fixed drafter increasingly mismatched with the policy's output distribution; and (ii) active batch sizes shrink throughout rollout decoding, shifting decoding from compute-bound to memory-bound regimes where parallel verification can exploit underutilized compute. Therefore, accelerating RL rollouts requires both a drafter that remains effective under long, high-temperature generations from an evolving policy and system-aware use of SD that avoids compute-bound regimes. We present EfficientRollout, a system-aware self-SD framework designed to address this gap for RL rollouts. EfficientRollout induces a quantized drafter from the target model (i.e. self-speculative decoding), keeping it coupled to the evolving policy without separate drafter pretraining or online adaptation. It further coordinates a system-aware SD toggle policy with acceptance-aware draft-length adaptation, enabling speculation only in beneficial regimes while matching the drafting budget to evolving drafter quality. EfficientRollout reduces rollout and end-to-end latency by up to 19.6% and 12.7%, respectively, over an accelerated AR rollout baseline, while preserving final model quality.

Related papers

• ResiSpec: Enhancing Multi-Candidate Speculative Sampling via Residual Distribution Shaping [79.1905186547363]
  ResiSpec is a framework that strategically reforms the proposal distribution during verification.<n>It prevents candidate obsolescence and achieves up to 1.92$times$ speedup over state-of-the-art multi-candidate methods.
  arXiv  Detail & Related papers  (2026-08-25T11:25:55Z)
• WAR: Workload-Aware Rollouts for Synchronous Agentic Reinforcement Learning [0.0]
  WAR is a workload-aware rollout system that substantially accelerates synchronous agentic RL training.<n>Under low load, WAR enables model-free speculative decoding with SuffixDecoding.<n>Under high load, WAR shifts the optimization focus to cache-aware scheduling.
  arXiv  Detail & Related papers  (2026-07-19T15:31:13Z)
• DC-Leap: Training-Free Acceleration of dLLMs via Draft-Guided Contiguous Leaping Decoding [52.13809682603516]
  DC-Leap is a training-free framework that enables reliable acceleration of dLLMs in the moderate-confidence regime.<n>It incorporates strictly-ordered causal constraints into the parallel decoding process.<n>It achieves substantial speedups, up to 53.19x on MBPP for long-sequence generation, and up to 105.02x when combined with KV-Cache.
  arXiv  Detail & Related papers  (2026-05-19T06:27:58Z)
• Accelerating RL Post-Training Rollouts via System-Integrated Speculative Decoding [8.43502622452414]
  We study speculative decoding as an acceleration primitive for RL rollouts.<n>We implement speculative decoding in NeMo-RL with a vLLM backend.<n>We project that combining speculative decoding with asynchronous RL yields up to 2.5x end-to-end training speedup at 235B scale.
  arXiv  Detail & Related papers  (2026-04-29T15:11:48Z)
• Generate, Filter, Control, Replay: A Comprehensive Survey of Rollout Strategies for LLM Reinforcement Learning [79.88942231770629]
  Reinforcement learning (RL) has become a central post-training tool for improving the reasoning abilities of large language models (LLMs)<n>We formalize rollout pipelines with unified notation and introduce Generate-Filter-Control-Replay (GFCR)<n>We synthesize methods spanning RL with verifiable rewards, process supervision, judge-based gating, guided and tree/segment rollouts, adaptive compute allocation, early-exit and partial rollouts, throughput optimization, and replay/recomposition for self-improvement.
  arXiv  Detail & Related papers  (2026-04-08T00:53:29Z)
• SortedRL: Accelerating RL Training for LLMs through Online Length-Aware Scheduling [54.276306194000405]
  We propose SortedRL, an online length-aware scheduling strategy for scaling reinforcement learning.<n>SortedRL reorders rollout samples based on output lengths, prioritizing short samples forming groups for early updates.<n>Experiments using LLaMA-3.1-8B and Qwen-2.5-32B on diverse tasks, including logical puzzles, show that SortedRL reduces RL training bubble ratios by over 50%.
  arXiv  Detail & Related papers  (2026-03-24T16:48:31Z)
• Contextual Rollout Bandits for Reinforcement Learning with Verifiable Rewards [69.74686029941881]
  Reinforcement Learning with Verifiable Rewards (RLVR) is an effective paradigm for improving the reasoning capabilities of large language models.<n>We propose a unified neural scheduling framework that adaptively selects high-value rollouts throughout training.<n>Experiments on six mathematical reasoning benchmarks demonstrate consistent gains in performance and training efficiency across multiple RLVR optimization methods.
  arXiv  Detail & Related papers  (2026-02-09T10:51:58Z)
• Scaling LLM Speculative Decoding: Non-Autoregressive Forecasting in Large-Batch Scenarios [76.85739138203014]
  We present SpecFormer, a novel architecture that accelerates unidirectional and attention mechanisms.<n>We demonstrate that SpecFormer achieves lower training demands and reduced computational costs.
  arXiv  Detail & Related papers  (2025-11-25T14:20:08Z)
• Taming the Long-Tail: Efficient Reasoning RL Training with Adaptive Drafter [52.111923076688505]
  Training Large Language Models (LLMs) with strong reasoning capabilities marks a significant milestone, unlocking new frontiers in complex problem-solving.<n>We propose TLT, a system that accelerates reasoning RL training losslessly by integrating adaptive speculative decoding.
  arXiv  Detail & Related papers  (2025-11-20T18:59:25Z)
• Beat the long tail: Distribution-Aware Speculative Decoding for RL Training [75.75462952580796]
  We propose a Distribution Aware Speculative decoding framework that accelerates RL rollouts without altering model outputs.<n>Experiments on math and code reasoning tasks show that DAS reduces rollout time up to 50% while preserving identical training curves.
  arXiv  Detail & Related papers  (2025-11-17T19:02:12Z)
• Pluto and Charon: A Time and Memory Efficient Collaborative Edge AI Framework for Personal LLMs Fine-Tuning [13.26886445965894]
  Pluto and Charon (PAC) is a time and memory efficient collaborative edge AI framework for personal LLMs fine-tuning. PAC implements a personal LLMs fine-tuning technique that is efficient in terms of parameters, time, and memory. Extensive evaluation based on prototype implementation demonstrates that PAC remarkably outperforms state-of-the-art approaches.
  arXiv  Detail & Related papers  (2024-08-20T11:30:12Z)
• When to Sense and Control? A Time-adaptive Approach for Continuous-Time RL [37.58940726230092]
  Reinforcement learning (RL) excels in optimizing policies for discrete-time Markov decision processes (MDP) We formalize an RL framework, Time-adaptive Control & Sensing (TaCoS), that tackles this challenge. We demonstrate that state-of-the-art RL algorithms trained on TaCoS drastically reduce the interaction amount over their discrete-time counterpart.
  arXiv  Detail & Related papers  (2024-06-03T09:57:18Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.