The paper introduces T2T (Thickening-to-Thinning), a dynamic reward framework for large language models that mimics human learning by separating exploration and consolidation phases. During incorrect attempts, T2T encourages exploration to broaden the search space, while after correct solutions it applies length penalties to promote concise reasoning. Experiments on mathematical benchmarks across five mainstream LLMs show that T2T outperforms standard GRPO and recent baselines, improving overall reasoning performance.
By Wenze Lin, Zhen Yang, Xitai Jiang, Xiaoteng Ma, Gao Huang
arXiv:2410.23912v3 Announce Type: replace-cross
Abstract: The reasoning abilities of large language models (LLMs) have improved with chain-of-thought (CoT) prompting, allowing models to solve complex...
By Fu-Chieh Chang, Yu-Ting Lee, Hui-Ying Shih, Yi Hsuan Tseng, Pei-Yuan Wu
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. 02113v1 Announce Type: cross Abstract: Post-training has become a primary driver of recent progress in large reasoning models, and reasoning data are often the key variable determining whether this stage succeeds.
By Yaoming Li, Guangxiang Zhao, Qilong Shi, Lin Sun, Xiangzheng Zhang, Tong Yang
arXiv:2604. 06628v2 Announce Type: replace Abstract: A prevailing narrative in LLM post-training holds that supervised finetuning (SFT) memorizes while reinforcement learning (RL) generalizes.
By Qihan Ren, Peng Wang, Ruikun Cai, Shuai Shao, Dadi Guo, Yuejin Xie, Yafu Li, Quanshi Zhang, Xia Hu, Jing Shao, Dongrui Liu
The paper demonstrates that fine‑tuning reasoning models to predict their own confidence at intermediate steps—using only 600 self‑supervised examples—substantially improves inference efficiency. Without adding any explicit stopping or length penalties, the models generate up to 25 % fewer tokens while maintaining accuracy on mathematical, scientific, and coding benchmarks across several architectures. The study finds that confidence supervision preserves the models’ high‑level reasoning structure rather than merely suppressing specific behaviors.
By Parsa Hosseini, Akasha Tigalappanavara, Sumit Nawathe, Chenrui Fan, Sourya Basu, Genta Indra Winata, Anirban Das, Soheil Feizi, Nima Chitsazan
Rationale-Guided Policy Optimization (RGPO) is a reinforcement‑learning framework that adaptively uses ground‑truth rationale information to scaffold a language model’s reasoning process. Instead of treating reference solutions as fixed imitation targets, RGPO temporarily incorporates rationales to help the model generate better responses, then reverts to unguided learning with higher‑reward, model‑generated solutions. Experiments in both language‑only and vision‑language tasks show that RGPO consistently outperforms RLVR baselines, with ablation studies confirming that adaptive rationale guidance is a key factor in its success.
By Hoang Phan, Minh Pham, Chau Pham, Chinmay Hegde, Trung Le, Qi Lei
arXiv:2607. 16097v1 Announce Type: cross Abstract: Reinforcement learning (RL) has become central to improving large language models (LLMs) on complex reasoning tasks, yet RL post-training is largely studied in isolation from the pretraining that precedes it.
By Jingyan Shen, Ang Li, Salman Rahman, Yifan Sun, Micah Goldblum, Matus Telgarsky, Pavel Izmailov
arXiv:2604.08454v2 Announce Type: replace
Abstract: Large language models are increasingly deployed in high-stakes domains, where confident yet incorrect inferences may cause severe real-world harm,...
By Haokai Ma, Lee Yan Zhen, Gang Yang, Yunxiang Chen, Yunshan Ma, Tat-Seng Chua, Ee-Chien Chang
arXiv:2603. 09803v2 Announce Type: replace Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) improves reasoning in large language models but treats all correct solutions equally, potentially reinforcing flawed traces that arrive at correct answers by chance.
By Tiehua Mei, Minxuan Lv, Leiyu Pan, Zhenpeng Su, Hongru Hou, Hengrui Chen, Ao Xu, Deqing Yang
The paper introduces SIF, a reinforcement learning technique that uses selective internal feedback to improve language models’ reasoning without external verifiers. SIF constructs comparative rewards by comparing token‑level self‑certainty across multiple windows and peers, rewarding only when both voting levels agree, and applies sample and group‑level filtering to decide which responses influence policy updates. Experiments on three backbone models across 12 mathematical and scientific benchmarks show that SIF’s final checkpoints outperform the best‑validation checkpoints of Intuitor and Vigor, achieving a 4.71‑point accuracy gain on Qwen2.5‑Math‑1.5B and maintaining accuracy during continued optimization with minimal test‑time rollouts.
By Xun Lu, Shumeng Yang, Xinrui Wang, Songjun Tu, Linjing Li, Daniel Dajun Zeng
arXiv:2607. 26119v1 Announce Type: new Abstract: Large reasoning models trained via reinforcement learning (RL) have been increasingly shown to outperform their supervised fine-tuned (SFT) counterparts on mathematical reasoning tasks; Yet the mechanistic basis for this advantage remains unclear.
By Antyabha Rahman, Akshaj Gurugubelli, Omar Ankit, Kevin Zhu, Aishwarya Balwani