Image de couverture de livre

SPEC-RL: Accelerating On-Policy Reinforcement Learning with Speculative Rollouts

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Liu, Bingshuai, Wang, Ante, Min, Zijun, Yao, Liang, Zhang, Haibo, Liu, Yang, Han, Xu, Li, Peng, Zeng, Anxiang, Su, Jinsong
Format:	Preprint
Publié:	2025
Sujets:	Machine Learning Artificial Intelligence Computation and Language
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Orchestrating Tokens and Sequences: Dynamic Hybrid Policy Optimization for RLVR
par: Min, Zijun, et autres
Publié: (2026)

Retrieval-augmented Multi-modal Chain-of-Thoughts Reasoning for Large Language Models
par: Liu, Bingshuai, et autres
Publié: (2023)

Exploring Optimal Transport-Based Multi-Grained Alignments for Text-Molecule Retrieval
par: Min, Zijun, et autres
Publié: (2024)

Accelerating RL Post-Training Rollouts via System-Integrated Speculative Decoding
par: Iso, Hayate, et autres
Publié: (2026)

Jet-RL: Enabling On-Policy FP8 Reinforcement Learning with Unified Training and Rollout Precision Flow
par: Xi, Haocheng, et autres
Publié: (2026)

Mitigating the Negative Impact of Over-association for Conversational Query Production
par: Wang, Ante, et autres
Publié: (2024)

Sparse-RL: Breaking the Memory Wall in LLM Reinforcement Learning via Stable Sparse Rollouts
par: Luo, Sijia, et autres
Publié: (2026)

Decomposing and Steering Functional Metacognition in Large Language Models
par: Li, Yanshi, et autres
Publié: (2026)

A Dual-Perspective Metaphor Detection Framework Using Large Language Models
par: Lin, Yujie, et autres
Publié: (2024)

Stable and Efficient Single-Rollout RL for Multimodal Reasoning
par: Liu, Rui, et autres
Publié: (2025)

Not All Rollouts are Useful: Down-Sampling Rollouts in LLM Reinforcement Learning
par: Xu, Yixuan Even, et autres
Publié: (2025)

SRT: Accelerating Reinforcement Learning via Speculative Rollout with Tree-Structured Cache
par: Chang, Chi-Chih, et autres
Publié: (2026)

Compass-Embedding v4: Robust Contrastive Learning for Multilingual E-commerce Embeddings
par: Ueareeworakul, Pakorn, et autres
Publié: (2025)

Beyond Passive Critical Thinking: Fostering Proactive Questioning to Enhance Human-AI Collaboration
par: Wang, Ante, et autres
Publié: (2025)

Investigating Inference-time Scaling for Chain of Multi-modal Thought: A Preliminary Study
par: Lin, Yujie, et autres
Publié: (2025)

HaS: Accelerating RAG through Homology-Aware Speculative Retrieval
par: Peng, Peng, et autres
Publié: (2026)

Towards Better RL Training Data Utilization via Second-Order Rollout
par: Yang, Zhe, et autres
Publié: (2026)

Learning to Draft: Adaptive Speculative Decoding with Reinforcement Learning
par: Zhang, Jiebin, et autres
Publié: (2026)

Mid-Training of Large Language Models: A Survey
par: Mo, Kaixiang, et autres
Publié: (2025)

EAGLE-Pangu: Accelerator-Safe Tree Speculative Decoding on Ascend NPUs
par: Han, Chang, et autres
Publié: (2026)

Are Full Rollouts Necessary for On-Policy Distillation?
par: Zhang, Yaocheng, et autres
Publié: (2026)

Response Enhanced Semi-supervised Dialogue Query Generation
par: Huang, Jianheng, et autres
Publié: (2023)

Enhancing Agentic RL with Progressive Reward Shaping and Value-based Sampling Policy Optimization
par: Su, Jianghao, et autres
Publié: (2025)

Not All Languages are Equal: Insights into Multilingual Retrieval-Augmented Generation
par: Wu, Suhang, et autres
Publié: (2024)

Optimal Transport-Based Token Weighting scheme for Enhanced Preference Optimization
par: Li, Meng, et autres
Publié: (2025)

Length-Unbiased Sequence Policy Optimization: Revealing and Controlling Response Length Variation in RLVR
par: Liu, Fanfan, et autres
Publié: (2026)

Towards Transparent RAG: Fostering Evidence Traceability in LLM Generation via Reinforcement Learning
par: Ren, Jingyi, et autres
Publié: (2025)

AdaSPEC: Selective Knowledge Distillation for Efficient Speculative Decoders
par: Hu, Yuezhou, et autres
Publié: (2025)

RoRecomp: Enhancing Reasoning Efficiency via Rollout Response Recomposition in Reinforcement Learning
par: Li, Gang, et autres
Publié: (2025)

EditGRPO: Reinforcement Learning with Post-Rollout Edits for Clinically Accurate Chest X-Ray Report Generation
par: Zhang, Kai, et autres
Publié: (2025)

Each Prompt Matters: Scaling Reinforcement Learning Without Wasting Rollouts on Hundred-Billion-Scale MoE
par: Zeng, Anxiang, et autres
Publié: (2025)

REA-RL: Reflection-Aware Online Reinforcement Learning for Efficient Reasoning
par: Deng, Hexuan, et autres
Publié: (2025)

Beyond Negative Rollouts: Positive-Only Policy Optimization with Implicit Negative Gradients
par: Xu, Mingwei, et autres
Publié: (2026)

Mitigating Catastrophic Forgetting in Large Language Models with Self-Synthesized Rehearsal
par: Huang, Jianheng, et autres
Publié: (2024)

DocTER: Evaluating Document-based Knowledge Editing
par: Wu, Suhang, et autres
Publié: (2023)

EMS-SD: Efficient Multi-sample Speculative Decoding for Accelerating Large Language Models
par: Ni, Yunsheng, et autres
Publié: (2024)

SpecExit: Accelerating Large Reasoning Model via Speculative Exit
par: Yang, Rubing, et autres
Publié: (2025)

Writing-RL: Advancing Long-form Writing via Adaptive Curriculum Reinforcement Learning
par: Lei, Xuanyu, et autres
Publié: (2025)

LogitSpec: Accelerating Retrieval-based Speculative Decoding via Next Next Token Speculation
par: Liu, Tianyu, et autres
Publié: (2025)

Accelerate Speculative Decoding with Sparse Computation in Verification
par: Wang, Jikai, et autres
Publié: (2025)