English
Related papers

Related papers: Inference-Time Diversity in RL-Trained Lean Theore…

200 papers

Large language models (LLMs) achieve high accuracy on many reasoning benchmarks but remain brittle under structural perturbations of rule-based systems. We introduce a diagnostic framework with four stress tests -- redundant vs. essential…

Artificial Intelligence · Computer Science 2026-05-26 Qiming Bao , Xiaoxuan Fu , Michael Witbrock

Large Language Models (LLMs) have shown significant promise in automated theorem proving, yet progress is often constrained by the scarcity of diverse and high-quality formal language data. To address this issue, we introduce…

Computation and Language · Computer Science 2025-12-02 Xinyuan Zhou , Yi Lei , Xiaoyu Zhou , Jingyi Sun , Yu Zhu , Zhongyi Ye , Weitai Zhang , Quan Liu , Si Wei , Cong Liu

Reinforcement learning (RL) promises to expand the capabilities of language models, but it is unclear if current RL techniques promote the discovery of novel behaviors, or simply sharpen those already present in the base model. In this…

Machine Learning · Computer Science 2025-10-14 Jens Tuyls , Dylan J. Foster , Akshay Krishnamurthy , Jordan T. Ash

Reasoning models improve their problem-solving ability through inference-time scaling, allocating more compute via longer token budgets. Identifying which reasoning traces are likely to succeed remains a key opportunity: reliably predicting…

Artificial Intelligence · Computer Science 2025-10-14 Martina G. Vilas , Safoora Yousefi , Besmira Nushi , Eric Horvitz , Vidhisha Balachandran

Reasoning models post-hoc rationalize answers they have already committed to internally, producing chains of *reasoning theater*: deliberative-looking steps that contribute nothing to correctness. This wastes inference tokens, pollutes…

Machine Learning · Computer Science 2026-05-13 Swapnil Parekh

Reinforcement learning with verifiable rewards (RLVR) enables large language models to acquire slow, multi-step reasoning from sparse final-answer signals. We provide a statistical-physics picture of this emergence. We show that an…

Artificial Intelligence · Computer Science 2026-05-08 Sihan Hu , Xiansheng Cai , Yuan Huang , Zhiyuan Yao , Linfeng Zhang , Pan Zhang , Youjin Deng , Kun Chen

While reinforcement learning with verifiable rewards (RLVR) is effective to improve the reasoning ability of large language models (LLMs), its reliance on human-annotated labels leads to the scaling up dilemma, especially for complex tasks.…

Machine Learning · Computer Science 2026-03-17 Zizhuo Zhang , Jianing Zhu , Xinmu Ge , Zihua Zhao , Zhanke Zhou , Xuan Li , Xiao Feng , Jiangchao Yao , Bo Han

Recent studies show that the reasoning capabilities of Large Language Models (LLMs) can be improved by applying Reinforcement Learning (RL) to question-answering (QA) tasks in areas such as math and coding. With a long context length, LLMs…

Computation and Language · Computer Science 2025-10-17 Stephen Chung , Wenyu Du , Jie Fu

Post-training improves instruction-following and helpfulness of large language models (LLMs) but often reduces generation diversity, which leads to repetitive outputs in open-ended settings, a phenomenon known as mode collapse. Motivated by…

Computation and Language · Computer Science 2026-02-09 Bowen Zhang , Meiyi Wang , Harold Soh

Recent advances in Automated Theorem Proving have shown the effectiveness of leveraging a (large) language model that generates tactics (i.e. proof steps) to search through proof states. The current model, while trained solely on successful…

Artificial Intelligence · Computer Science 2024-07-31 Chenyang An , Zhibo Chen , Qihao Ye , Emily First , Letian Peng , Jiayun Zhang , Zihan Wang , Sorin Lerner , Jingbo Shang

We study how to scale reasoning token budgets for competitive programming through two complementary approaches: training-time reinforcement learning (RL) and test-time parallel thinking. During RL training, we observe an approximately…

Computation and Language · Computer Science 2026-04-03 Qianfan Zhang , Tianyu Guo , Xuandi Ren , Jiale Chen , Ming Ding , Ran Xin , Xia Xiao

We introduce Goedel-Prover-V2, a series of open-source language models that set a new state-of-the-art in automated theorem proving. Built on the standard expert iteration and reinforcement learning pipeline, our approach incorporates three…

Recent neural theorem provers use reinforcement learning with verifiable rewards (RLVR), where proof assistants provide binary correctness signals. While verifiable rewards are cheap and scalable without reward hacking issues, they suffer…

Artificial Intelligence · Computer Science 2026-05-12 Zeynel A. Uluşan , Burak S. Akbudak , Can S. Erer , Gözde Gül Şahin

Medical RAG needs evidence-grounded claims, so plugging a claim-level NLI checker into retrieval-augmented RL is intuitive. \textbf{We find that the checker's \emph{output distribution} during training, not its held-out accuracy, decides…

Computation and Language · Computer Science 2026-05-26 Yuelyu Ji , Min Gu Kwak , Hang Zhang , Xizhi Wu , Chenyu Li , Yanshan Wan

Recent reinforcement learning (RL) techniques have yielded impressive reasoning improvements in language models, yet it remains unclear whether post-training truly extends a model's reasoning ability beyond what it acquires during…

Computation and Language · Computer Science 2025-12-09 Charlie Zhang , Graham Neubig , Xiang Yue

Reasoning LLMs (RLMs) such as OpenAI o1, DeepSeek-R1, and Qwen3 deliver strong multi-step reasoning through chain-of-thought generation, but their large model sizes and lengthy decode-time outputs make them costly to deploy and unsuitable…

Computation and Language · Computer Science 2025-12-03 Ziyan Wang , Enmao Diao , Qi Le , Pu Wang , Guanchu Wang , Minwoo Lee , Shu-ping Yeh , Li Yang

While Large Language Models (LLMs) demonstrate exceptional performance in surface-level text generation, their nature in handling complex multi-step reasoning tasks often remains one of ``statistical fitting'' rather than systematic logical…

Machine Learning · Computer Science 2026-01-27 Lianlei Shan , Han Chen , Yixuan Wang , Zhenjie Liu , Wei Li

Large language models (LLMs) have achieved strong performance on complex reasoning tasks using techniques such as chain-of-thought and self-consistency. However, ensemble-based approaches, especially self-consistency which relies on…

Artificial Intelligence · Computer Science 2025-12-23 Qinglin Zeng , Jing Yang , Keze Wang

Model collapse, the progressive degradation of LLMs trained on their own outputs, has been characterized statistically but lacks a linguistic explanation for which structures degrade, in what order, and why. We show that iterated learning…

Computation and Language · Computer Science 2026-05-25 Dongxin Guo , Jikun Wu , Siu Ming Yiu

Vision-language models (VLMs) frequently generate hallucinated content plausible but incorrect claims about image content. We propose a training-free self-correction framework enabling VLMs to iteratively refine responses through…

Computer Vision and Pattern Recognition · Computer Science 2025-12-11 Kassoum Sanogo , Renzo Ardiccioni