paper-with-me

Papers

Hi-TTRL: Regulating Consensus with Hints for Test-Time Reinforcement Learning

2026-08-04 · Kunbin Xu, Xingzuo Li, Xuefeng Bai, Kehai Chen arxiv

Test-time reinforcement learning (TTRL) improves the reasoning capabilities of large language models without labeled data by updating the policy with pseudo-labels constructed through majority voting. While effective, the reward signal assigned from majority voting is highly sensitive to consensus strength, defined as the frequency of the most common answer within a rollout group. In TTRL, consensus strength plays a dual role: it reflects both the reliability of the pseudo-label and the distribution of advantages. Low consensus can amplify updates from unreliable pseudo-labels through disproportionately large advantages, whereas high consensus reduces reward contrast and ultimately yields vanishing gradients. In this paper, we introduce Hi-TTRL, a test-time reinforcement learning framework that utilizes hints during sampling to regulate rollout consensus strength. Hi-TTRL first estimates consensus strength from a partial rollout group. When the consensus strength falls outside a target interval, it invokes a Markov chain Monte Carlo (MCMC) hint sampler. The sampler targets the power-transformed prefix distribution and uses finite-step approximate sampling to generate rollout prefixes as hints. By tuning the power exponent, Hi-TTRL generates hints with a sharpened or flattened power target, steering rollout consensus strength toward the target interval. Experiments on multiple datasets and backbones show that Hi-TTRL consistently improves over standard TTRL, with ablations and consensus-steering analyses validating the effectiveness of adaptive hint-guided consensus regulation.

📄 PDF Abstract BibTeX arXiv:2608.03545

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

CG-TTRL: Context-Guided Test-Time Reinforcement Learning for On-Device Large Language Models

2025-11-09 · Peyman Hosseini, Ondrej Bohdal, Taha Ceritli, Ignacio Castro 외 arxiv

Test-time Reinforcement Learning (TTRL) has shown promise in adapting foundation models for complex tasks at test-time, resulting in large performance improvements. TTRL leverages an elegant two-phase sampling strategy: …

Reinforcement Learning

What If Consensus Lies? Selective-Complementary Reinforcement Learning at Test Time

2026-03-20 · Dong Yan, Jian Liang, Yanbo Wang, Shuo Lu 외 arxiv

Test-Time Reinforcement Learning (TTRL) enables Large Language Models (LLMs) to enhance reasoning capabilities on unlabeled test streams by deriving pseudo-rewards from majority voting consensus. However, existing TTRL m…

Reinforcement Learning

MAPLE: Elevating Medical Reasoning from Statistical Consensus to Process-Led Alignment

2026-03-09 · Kailong Fan, Anqi Pu, Yichen Wu, Wanhua Li 외 arxiv

Recent advances in medical large language models have explored Test-Time Reinforcement Learning (TTRL) to enhance reasoning. However, standard TTRL often relies on majority voting (MV) as a heuristic supervision signal, …

Reinforcement Learning

Collaborative Multi-Agent Test-Time Reinforcement Learning for Reasoning

2026-01-14 · Zhiyuan Hu, Yunhai Hu, Juncheng Liu, Shuyue Stella Li 외 arxiv

Multi-agent systems have evolved into practical LLM-driven collaborators for many applications, gaining robustness from diversity and cross-checking. However, multi-agent RL (MARL) training is resource-intensive and unst…

Reinforcement Learning

Tool Verification for Test-Time Reinforcement Learning

2026-03-02 · Ruotong Liao, Nikolai Röhrich, Xiaohan Wang, Yuhui Zhang 외 arxiv

Test-time reinforcement learning (TTRL) has emerged as a promising paradigm for Recursive Self-Improving AI (RSI) by adapting Large Reasoning Models (LRMs) on unlabeled test inputs, using self-consensus rewards derived f…

Reinforcement Learning