Preprint · 2025 · Open access
Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs
arXiv (Cornell University) · Published 17 June 2025 · DOI 10.48550/arxiv.2506.14245
In brief
“Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs” is a 2025 preprint by Xumeng Wen and 11 co-authors, published in arXiv (Cornell University). OpenAlex counts 301 citations as of 15 November 2026. It is open access under a CC BY licence: the full text is free to read at arXiv (Cornell University).
- Citations
- 301
- Most cited in
- 2026 (250)
- References
- 30
- Authors
- 12
Abstract
Recent advancements in long chain-of-thought (CoT) reasoning, particularly through the Group Relative Policy Optimization algorithm used by DeepSeek-R1, have led to significant interest in the potential of Reinforcement Learning with Verifiable Rewards (RLVR) for Large Language Models (LLMs). While RLVR promises to improve reasoning by allowing models to learn from free exploration, there remains debate over whether it truly enhances reasoning abilities or simply boosts sampling efficiency. This paper systematically investigates the impact of RLVR on LLM reasoning. We revisit Pass@K experiments and demonstrate that RLVR can extend the reasoning boundary for both mathematical and coding tasks. This is supported by our introduction of a novel evaluation metric, CoT-Pass@K, which captures reasoning success by accounting for both the final answer and intermediate reasoning steps. Furthermore, we present a theoretical framework explaining RLVR's incentive mechanism, demonstrating how it can encourage correct reasoning even when rewards are based solely on answer correctness. Our analysis of RLVR's training dynamics reveals that it incentivizes correct reasoning early in the process, with substantial improvements in reasoning quality confirmed through extensive evaluations. These findings provide strong evidence of RLVR's potential to enhance LLM reasoning, offering valuable insights into its mechanisms and performance improvements.
Cite this paper
Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.
APA 7
Wen, X., Liu, Z., Zheng, S., Ye, S., Wu, Z., Wang, Y., Xu, Z., Xiao, L., Li, J., Miao, Z., Bian, J., & Yang, M. (2025). Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2506.14245
MLA 9
Wen, Xumeng, et al. “Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs.” arXiv (Cornell University), 2025. https://doi.org/10.48550/arxiv.2506.14245.
Chicago
Wen, Xumeng, Zihan Liu, Shun Zheng, Shengyu Ye, Zhirong Wu, Yang Wang, Zhijian Xu, et al. 2025. “Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs.” arXiv (Cornell University). https://doi.org/10.48550/arxiv.2506.14245.
Harvard
Wen, X. et al. (2025) ‘Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs’, arXiv (Cornell University). doi:10.48550/arxiv.2506.14245.
IEEE
X. Wen et al., “Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs,” arXiv (Cornell University), 2025, doi: 10.48550/arxiv.2506.14245.
BibTeX
@misc{wen2025reinforcement,
title = {{Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs}},
author = {Wen, Xumeng and Liu, Zihan and Zheng, Shun and Ye, Shengyu and Wu, Zhirong and Wang, Yang and Xu, Zhijian and Xiao, Liang, and Li, Junjie and Miao, Ziming and Bian, Jiang and Yang, Mao},
howpublished = {arXiv (Cornell University)},
year = {2025},
doi = {10.48550/arxiv.2506.14245},
url = {https://doi.org/10.48550/arxiv.2506.14245}
}
Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)
Citations by year
| Year | Citations | Share of the best year |
|---|---|---|
| 2024 | 1 | |
| 2025 | 50 | |
| 2026 | 250 |
Topic and keywords
Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- reinforcement learning with verifiable rewards
- large language models
- Group Relative Policy Optimization
- DeepSeek-R1
- mathematical reasoning
- programming tasks
- incentive mechanisms
- training dynamics
Work with this paper
Most-cited papers in Computer Science
- ImageNet classification with deep convolutional neural networksAlex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton · Communications of the ACM · 2017 · 114,726 citations
- Long Short-Term MemorySepp Hochreiter, Jürgen Schmidhuber · Neural Computation · 1997 · 109,097 citations
- A Mathematical Theory of CommunicationClaude E. Shannon · Bell System Technical Journal · 1948 · 91,736 citations
- Deep learningYann LeCun, Yoshua Bengio, Geoffrey E. Hinton · Nature · 2015 · 90,226 citations
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal NetworksShaoqing Ren, Kaiming He, Ross Girshick and 1 more · IEEE Transactions on Pattern Analysis and Machine Intelligence · 2016 · 74,143 citations
Questions about this paper
How do I cite this paper in APA style?
Wen, X., Liu, Z., Zheng, S., Ye, S., Wu, Z., Wang, Y., Xu, Z., Xiao, L., Li, J., Miao, Z., Bian, J., & Yang, M. (2025). Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2506.14245
How many times has this paper been cited?
301 times, according to OpenAlex on 15 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.
Is this paper free to read?
Yes. It is open access: read the free full text at arXiv (Cornell University).
Who wrote this paper?
Xumeng Wen, Zihan Liu, Shun Zheng, Shengyu Ye, Zhirong Wu, Yang Wang, Zhijian Xu, Liang, Xiao, Junjie Li, Ziming Miao, Jiang Bian, Mao Yang.
What is the DOI of this paper?
10.48550/arxiv.2506.14245. The permanent link is https://doi.org/10.48550/arxiv.2506.14245.
Bibliographic data from OpenAlex (CC0), last checked 15 November 2026. ResearchGuru does not host this paper; the links go to the publisher or the repository that does.