The paper proposes a new method for training large language models to handle long-context reasoning by combining Group Relative Policy Optimization (GRPO) with on‑policy distillation (OPD). It introduces a synthetic multilingual dataset called LongBlocks that tests multi‑hop reasoning, contextual grounding, and long‑form generation. Experiments show that the combined approach outperforms either GRPO or OPD alone while maintaining short‑context performance.
By Miguel Moura Ramos, Duarte M. Alves, Andr\'e F. T. Martins
arXiv:2607. 19345v1 Announce Type: cross Abstract: Large language models that generate step-by-step reasoning traces have achieved strong performance on complex tasks, and extending them to long-context settings has emerged as an important frontier.
By Lizhe Fang, Weizhou Shen, Tianyi Tang, Yisen Wang
arXiv:2511. 19314v2 Announce Type: replace Abstract: Information-seeking is a core capability for AI agents, requiring them to gather and reason over tool-generated information across long trajectories.
By Jaewoo Lee, Archiki Prasad, Justin Chih-Yao Chen, Zaid Khan, Elias Stengel-Eskin, Mohit Bansal
The paper introduces POISE, a reinforcement learning algorithm that uses a model’s internal states as a value estimator to reduce variance in reinforcement learning with verifiable rewards (RLVR). By employing a lightweight probe that reads internal signals during the forward pass, POISE predicts baselines online and uses a cross‑rollout construction to keep gradients unbiased. Experiments on Qwen3‑4B and OLMo3‑7B‑Instruct‑DPO across six domains show POISE outperforms existing RLVR baselines, offering more stable training and a value model that generalizes across tasks and scales with the policy.
By Yunho Choi, Jongwon Lim, Woojin Ahn, Minjae Oh, Jeonghoon Shim, Yohan Jo
arXiv:2604. 08477v2 Announce Type: replace-cross Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has substantially improved reasoning in formal domains such as mathematics and code, but extending these gains beyond STEM remains challenging.
By Ashima Suvarna, Kendrick Phan, Mehrab Beikzadeh, Hritik Bansal, Saadia Gabriel
arXiv:2607. 02073v1 Announce Type: new Abstract: Long-context reasoning requires models to locate, revise, and synthesize evidence distributed across lengthy inputs.
By Ya Gao, Pekka Marttinen
arXiv:2505. 15062v5 Announce Type: replace-cross Abstract: Knowledge extrapolation is the process of inferring novel information by combining and extending existing knowledge that is explicitly available.
By Jiashu He, Jinxuan Fan, Bowen Jiang, Ignacio Houine, Dan Roth, Alejandro Ribeiro
arXiv:2607. 16205v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards has emerged as a standard approach for enhancing reasoning in large language models, which typically optimizes the policy by contrasting multiple self generated rollouts.
By Dayu Wang, Jiaye Yang, Weikang Li, Jiahui Liang, Liwei Qian, Xin Pei, Jizhou Huang
AgenticRag‑R1 is a reinforcement‑learning framework that integrates reasoning, retrieval, and memory through a stack and fine‑grained action space. It uses hierarchical action‑aware rewards and an information‑aware trajectory rejection strategy to support long‑horizon learning. Experiments on multi‑hop, open‑domain, and agentic reasoning benchmarks show that AgenticRag‑R1 outperforms strong baselines and produces robust, interpretable, memory‑aware reasoning behaviors.
By Xinke Jiang, Yue Fang, Zhibang Yang, Jiaran Gao, Zhixin Zhang, Tao Feng, Rihong Qiu, Wentao Zhang, Hongxin Ding, Ruizhe Zhang, Yongxin Xu, Yuheng Huang, Xu Chu, Junfeng Zhao, Yasha Wang
arXiv:2609.15851v1 Announce Type: new
Abstract: Language models can learn from experience, but raw solution trajectories are often too long and noisy to provide effective guidance. In this work, we p...
By Guanheng Chen, Tianzhu Ye, Li Dong, Xun Wu, Shaohan Huang, Furu Wei
arXiv:2605. 01248v3 Announce Type: replace Abstract: Reinforcement learning (RL) post-training has enabled newer capabilities in models, such as agentic tool-use for search.
By Harsh Goel, Akhil Udathu, Susmija Jabbireddy, Pradnesh Kalkar, Atharva Parulekar
The paper introduces T2T (Thickening-to-Thinning), a dynamic reward framework for large language models that mimics human learning by separating exploration and consolidation phases. During incorrect attempts, T2T encourages exploration to broaden the search space, while after correct solutions it applies length penalties to promote concise reasoning. Experiments on mathematical benchmarks across five mainstream LLMs show that T2T outperforms standard GRPO and recent baselines, improving overall reasoning performance.
By Wenze Lin, Zhen Yang, Xitai Jiang, Xiaoteng Ma, Gao Huang