paper-with-me

Papers

Revisiting Complete Reasoning Traces for Post-Training

2026-09-07 · Jaehui Hwang, Sangdoo Yun, Byeongho Heo, Dongyoon Han hf

Large language models (LLMs) are often post-trained on pre-collected reasoning trajectories to improve their reasoning capability. Such trajectories tend to be long due to complex, interwoven paths, which often include detours on the path toward the answer. However, it has been underexplored whether LLMs indeed benefit from learning complete trajectories in post-training, such as supervised fine-tuning (SFT). Starting from our pilot study, we find that full trajectories provide only limited benefit, while partial trajectories are effective even under heavy truncation. We analyze redundancy in reasoning trajectories through attention-based analyses and controlled token-removal studies, both of which show that intermediate tokens contribute minimally to final reasoning quality. This suggests that avoiding redundant information may allow LLMs to internally infer coherent alternatives by inferring missing steps from their internal knowledge, given known trajectory endpoints. Furthermore, we show that training LLMs using endpoints leads to consistent changes in reasoning behavior, and that it also benefits post-training methods based on reinforcement learning or on-policy distillation, highlighting the need to revisit complete reasoning traces. Code is available at https://github.com/naver-ai/revisiting-trace.

📄 PDF Abstract BibTeX arXiv:2609.07103

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

Is Next-Chunk Reasoning RL Really Better than SFT? Revisiting Training Strategies under no-CoT Data

2026-08-24 · Yinhao Tang, Youqing Fang, Yanan Sun, Jiangning Liu 외 arxiv

Recent work proposes next-chunk reasoning RL for leveraging no-CoT data---corpora such as worked solutions and textbook derivations that contain reasoning-rich content but lack explicit chain-of-thought annotations. The …

Mathematical Reasoning

Learning from Partial Chain-of-Thought via Truncated-Reasoning Self-Distillation

2026-02-27 · Gianluigi Silvestri, Edoardo Cetin arxiv

Reasoning-oriented language models achieve strong performance by generating long chain-of-thought traces at inference time. However, this capability comes with substantial and often excessive computational cost, which ca…

Learning to Reason: Training LLMs with GPT-OSS or DeepSeek R1 Reasoning Traces

2025-11-24 · Shaltiel Shmidman, Asher Fredman, Oleg Sudakov, Meriem Bendris arxiv

Test-time scaling, which leverages additional computation during inference to improve model accuracy, has enabled a new class of Large Language Models (LLMs) that are able to reason through complex problems by understand…

Revisiting the UID Hypothesis in LLM Reasoning Traces

2025-10-11 · Minju Gwak, Guijin Son, Jaehyung Kim arxiv

Large language models (LLMs) often solve problems using step-by-step Chain-of-Thought (CoT) reasoning, yet these intermediate steps are frequently unfaithful or hard to interpret. Inspired by the Uniform Information Dens…

Learn to Think: Improving Multimodal Reasoning through Vision-Aware Self-Improvement Training

2026-05-12 · Qihuang Zhong, Liang Ding, Wenjie Xuan, Juhua Liu 외 arxiv

Post-training with explicit reasoning traces is common to improve the reasoning capabilities of Multimodal Large Language Models (MLLMs). However, acquiring high-quality reasoning traces is often costly and time-consumin…

Multimodal Reasoning