paper-with-me

홈 › Papers

Prioritize the Process, Not Just the Outcome: Rewarding Latent Thought Trajectories Improves Reasoning in Looped Language Models

2026-02-11 · Jonathan Williams, Esin Tureci arxiv

Looped Language Models (LoopLMs) perform multi-step latent reasoning prior to token generation and outperform conventional LLMs on reasoning benchmarks at smaller parameter budgets. However, attempts to further improve LoopLM reasoning with reinforcement learning have failed - standard objectives such as Group Relative Policy Optimization (GRPO) only assign credit to the final latent state, creating a fundamental mismatch with the model's internal computation. To resolve this, we introduce RLTT (Reward Latent Thought Trajectories), a reinforcement learning framework which distributes reward across the full latent reasoning trajectory. RLTT provides dense, trajectory-level credit assignment without relying on external verifiers and can directly replace GRPO with negligible overhead. Across extensive experiments with Ouro-1.4B/2.6B-Thinking under identical training and inference conditions, RLTT yields statistically significant improvements over GRPO on challenging mathematical reasoning benchmarks, improving mean accuracy over MATH-500, AIME24/26, and BeyondAIME by +5.8% on the 1.4B scale, and +10.9% on the 2.6B scale. Despite being trained exclusively on mathematics, RLTT also transfers effectively to non-mathematical reasoning benchmarks, demonstrating the effectiveness of trajectory-level credit assignment for reinforcement learning in LoopLMs. Code is available at https://github.com/jonwill8/RLTT.git.

📄 PDF Abstract BibTeX arXiv:2602.10520

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement LearningMathematical Reasoning

Similar Papers 제목 키워드 기반

Reinforcing Chain-of-Thought Reasoning with Self-Evolving Rubrics

2026-02-11 · Leheng Sheng, Wenchang Ma, Ruixin Hong, Xiang Wang 외 arxiv

Despite chain-of-thought (CoT) playing crucial roles in LLM reasoning, directly rewarding it is difficult: training a reward model demands heavy human labeling efforts, and static RMs struggle with evolving CoT distribut…

Beyond Correctness: Rewarding Faithful Reasoning in Retrieval-Augmented Generation

2025-10-15 · Zhichao Xu, Zongyu Wu, Yun Zhou, Aosong Feng 외 arxiv

Inspired by the success of reinforcement learning (RL) in Large Language Model (LLM) training for domains like math and code, recent work has begun training LLMs to dynamically plan, query, and reason with search engines…

Reinforcement Learning

Calibrated Self-Rewarding Vision Language Models

2024-05-23 · Yiyang Zhou, Zhiyuan Fan, Dongjie Cheng, Sihan Yang 외

Large Vision-Language Models (LVLMs) have made substantial progress by integrating pre-trained large language models (LLMs) and vision models through instruction tuning. Despite these advancements, LVLMs often exhibit th…

HallucinationLanguage ModellingVisual Question Answering

Process-based Self-Rewarding Language Models

2025-03-05 · Shimao Zhang, Xiao Liu, Xin Zhang, Junxiao Liu 외

Large Language Models have demonstrated outstanding performance across various downstream tasks and have been widely applied in multiple scenarios. Human-annotated preference data is used for training to further improve …

Mathematical Reasoning

Self-Rewarding Vision-Language Model via Reasoning Decomposition

2025-08-27 · Zongxia Li, Wenhao Yu, Chengsong Huang, Zhenwen Liang 외 arxiv

Vision-Language Models (VLMs) often suffer from visual hallucinations: generating things that are not consistent with visual inputs and language shortcuts, where they skip the visual part and just rely on text priors. Th…

Reinforcement LearningVisual Reasoning