paper-with-me

Papers

LongRLVR: Long-Context Reinforcement Learning Requires Verifiable Context Rewards

2026-03-02 · Guanzheng Chen, Michael Qizhe Shieh, Lidong Bing arxiv

Reinforcement Learning with Verifiable Rewards (RLVR) has significantly advanced the reasoning capabilities of Large Language Models (LLMs) by optimizing them against factual outcomes. However, this paradigm falters in long-context scenarios, as its reliance on internal parametric knowledge is ill-suited for tasks requiring contextual grounding--the ability to find and reason over externally provided information. We identify a key reason for this failure: a reward based solely on the final answer is too sparse to effectively guide the model for identifying relevant evidence. We formally prove that the outcome-only reward leads to significant vanishing gradients for the context grounding process, rendering learning intractable. To overcome this bottleneck, we introduce LongRLVR to augment the sparse answer reward with a dense and verifiable context reward. This auxiliary signal directly incentivizes the model for selecting the correct grounding information, providing a robust learning gradient that solves the underlying optimization challenge. We validate our method on challenging long-context benchmarks using Qwen and LLaMA models. LongRLVR consistently and significantly outperforms the standard RLVR across all models and benchmarks, e.g., boosting a 14B model's scores on RULER-QA from 73.17 to 88.90 and on LongBench v2 from 39.8 to 46.5. Our work demonstrates that explicitly rewarding the grounding process is a critical and effective strategy for unlocking the full reasoning potential of LLMs in long-context applications. Our code is available at https://github.com/real-absolute-AI/LongRLVR.

📄 PDF Abstract BibTeX arXiv:2603.02146

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

RePolicy: Reinforcement Learning for Safety-Policy Invocation in Agent Safeguards

2026-08-25 · Houcheng Jiang, Boxuan Zhang, Qiyong Zhong, Junfeng Fang 외 arxiv

Safeguarding language model agents requires assessing complete execution trajectories under context-dependent safety policies. Existing policy-aware safeguards mainly rely on prompting or supervised fine-tuning, limiting…

Reinforcement Learning

Policy-Conditioned Counterfactual Credit for Verifiable Reinforcement Learning of Long-Horizon Language Agents

2026-06-03 · Renwei Meng arxiv

Reinforcement learning with verifiable rewards improves reasoning and tool use, yet long-horizon language agents still learn unsupported evidence chains, belief drift, and shortcut actions that satisfy terminal checks. E…

Reinforcement Learning

Likelihood-Based Reward Designs for General LLM Reasoning

2026-02-03 · Ariel Kwiatkowski, Natasha Butt, Ismail Labiad, Julia Kempe 외 arxiv

Fine-tuning large language models (LLMs) on reasoning benchmarks via reinforcement learning requires a specific reward function, often binary, for each benchmark. This comes with two potential limitations: the need to de…

Reinforcement LearningMathematical Reasoning

Scaling Agentic Capabilities, Not Context: Efficient Reinforcement Finetuning for Large Toolspaces

2026-03-05 · Karan Gupta, Pranav Vajreshwari, Yash Pandya, Raghav Magazine 외 arxiv

Agentic systems operating over large tool ecosystems must plan and execute long-horizon workflows under weak or non-verifiable supervision. While frontier models mitigate these challenges through scale and large context …

Document Reconstruction Unlocks Scalable Long-Context RLVR

2026-02-09 · Yao Xiao, Lei Wang, Yue Deng, Guanzheng Chen 외 arxiv

Reinforcement Learning with Verifiable Rewards~(RLVR) has become a prominent paradigm to enhance the capabilities (i.e.\ long-context) of Large Language Models~(LLMs). However, it often relies on gold-standard answers or…

Reinforcement Learning