The paper introduces Thinking-Free Policy Initialization (TFPI), a simple adaptation to Reinforcement Learning with Verifiable Reward (RLVR) that removes explicit thinking content during training and inference. TFPI bridges long Chain-of-Thought distillation with standard RLVR, reducing token usage and improving performance. Experiments show that TFPI accelerates RL convergence, raises performance ceilings, and yields token‑efficient reasoning models, enabling a 4B model to achieve 89.0% accuracy on AIME24 and 65.5% on LiveCodeBench with under 4K H20 hours.
By Xin Xu, Clive Bai, Kai Yang, Tianhao Chen, Yang Wang, Saiyong Yang, Can Yang
When2Think introduces a post‑training framework that dynamically allocates reasoning depth in Large Reasoning Models based on instance difficulty. The method uses Instance‑level Difficulty‑Aware Control (IDAC) to shape rewards with pre‑computed accuracy and token usage statistics, enabling stable, critic‑free optimization without learned reward models. Experiments on mathematical benchmarks show that When2Think improves accuracy‑efficiency trade‑offs, achieving higher Pass@3 scores while reducing token usage compared to baseline models.
By Jaejun Shim, HyunJin Kim, Young Jin Kim, JinYeong Bak
The paper reports that in on‑policy distillation for large language models, reasoning performance can be improved by supervising only a tiny fraction of generated tokens—sometimes just one or two tokens per reasoning trajectory, about 0.05% of all tokens. This sparse supervision consistently matches or exceeds full‑token training across nine teacher‑student setups on mathematical reasoning, and is also validated on coding reasoning, Llama models, and PPO‑based reinforcement learning with verifiable reward. The findings suggest that effective post‑training does not require token‑intensive supervision and may align more closely with natural learning processes that focus on critical reasoning steps.
By Zhishuai Liu, Xingzi Xu, Mehmet Saygin Seyfioglu, Pan Xu, Karim Bouyarmane
arXiv:2606. 07950v1 Announce Type: new Abstract: RL with verifiable rewards can substantially improve LLM reasoning, yet standard GRPO-style training often treats easy, hard, and learnable questions alike through uniform sampling and weighting, leading to inefficient compute allocation.
By Zhanke Zhou, Xiangyu Lu, Chentao Cao, Brando Miranda, Tongliang Liu, Bo Han, Sanmi Koyejo
arXiv:2605.06165v2 Announce Type: replace
Abstract: As the widespread adoption of Large Language Models (LLMs) accelerates, token consumption from intermediate reasoning traces increasingly contribut...
By Richmond Sin Jing Xuan, Rishabh Bhardwaj, Soujanya Poria
arXiv:2609.37304v1 Announce Type: new
Abstract: Large reasoning models improve performance on challenging problems by allocating additional computation before answering, but longer reasoning does not...
By Zhibin Wen, Tao Han, Lei Bai, Can Li, Yang Xu
The paper investigates whether the high costs of training chain-of-thought reasoning models can be reduced through algorithmic design. It introduces an autocurriculum approach that lets the model select which problems to focus on during training, showing that this method provably improves both supervised fine‑tuning and reinforcement learning. For supervised fine‑tuning, autocurriculum requires exponentially fewer reasoning demonstrations by targeting prompts where the model struggles, while for reinforcement learning it decouples computational cost from the quality of the reference model, making the burn‑in cost nearly independent of target accuracy.
By Nived Rajaraman, Audrey Huang, Miro Dudik, Robert Schapire, Dylan J. Foster, Akshay Krishnamurthy
arXiv:2606. 17687v1 Announce Type: cross Abstract: Despite remarkable performance on complex tasks, Large Reasoning Models (LRMs) often generate excessively long Chain-of-Thoughts (CoT), inflating computational costs even for simple queries.
By Jiahao Wang, Bingyu Liang, Chenhao Hu, Longhui Zhang, Xuebo Liu, Min zhang, Jing Li, Xuelong Li
arXiv:2510.27042v3 Announce Type: replace
Abstract: Increasing the thinking budget of AI models can significantly improve accuracy, but not all questions warrant the same amount of reasoning. Users m...
By Michael Kleinman, Matthew Trager, Alessandro Achille, Wei Xia, Stefano Soatto
arXiv:2604.02007v3 Announce Type: replace
Abstract: Building general-purpose reasoning models using reinforcement learning with verifiable rewards (RLVR) across diverse domains has been widely adopte...
By Rafael Pardinas, Ehsan Kamalloo, David Vazquez, Alexandre Drouin
arXiv:2608. 20256v1 Announce Type: new Abstract: Reasoning language models trained with reinforcement learning typically operate under a fixed token budget rather than an explicitly adaptive one, which can lead to over-computation on easy problems and insufficient computation on difficult ones.
By Gijs Kassenaar, Zhao Yang, Vincent Fran\c{c}ois-Lavet
The paper introduces a method to enhance large language model (LLM) exploration in Reinforcement Learning with Verifiable Rewards (RLVR) by guiding the target model with partial reasoning trajectories from smaller, weaker language models. This weak-model guidance disrupts over‑confidence, preserves generative diversity, and mitigates entropy collapse without extra fine‑tuning or complex reward designs. Experiments on mathematical benchmarks show consistent improvements over vanilla RLVR, especially as the number of allowed attempts ($k$) increases, indicating broader reasoning coverage.
By Xingyu Shen, Huishuai Zhang, Peng Li, Yinchun Wang, Dongyan Zhao