🧑‍🔬 Lin Yan

🏛️ Affiliations

🎓Northwest UniversityXi'an · China2026 · 1 papercurrent🏢Merck & Co., Inc., Rahway, NJ, USARahway · United States2004–2025 · 26 papers🎓Beihang UniversityBeijing · China2005–2015 · 4 papers🎓Yan'an UniversityShanxi · China2021–2023 · 3 papers🎓Hunan University of Science and TechnologyXiangtan · China2009–2014 · 3 papers🎓Jiangxi Agricultural UniversityNanchang · China2011 · 3 papers🎓North University of ChinaTaiyuan · China2021–2022 · 2 papers🏛️Chinese Academy of ForestryBeijing · China2019–2021 · 2 papers🔬Institute of Chemical Industry of Forest ProductsNanjing · China2019–2021 · 2 papers🎓Shenyang University of Chemical TechnologyShenyang · China2016–2019 · 2 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 14 more institutions on single bylines are not listed.

🧭 Fields, topics and concepts

Of its works in our library, with the number of works on each.

📅 Works by year

2020s

📄 Papers

The most cited first (OpenAlex and our library).

1. Exploring Data Scaling Trends and Effects in Reinforcement Learning from Human Feedback (2025 · 12 citations)
   This ByteDance Seed study asks why adding more prompt data to PPO-based RLHF stops improving large language models.
2. Process Supervision-Guided Policy Optimization for Code Generation (2024 · 12 citations)
   RL from unit-test feedback helps LLMs write better code, but the reward is sparse. A single pass/fail signal arrives only after the whole program runs, and a program that fails every test gives the model nothing to learn from. The authors (ByteDance and Oregon State University) propose Process Supervision-Guided…
3. PAG: Multi-Turn Reinforced LLM Self-Correction with Policy as Generative Verifier (2025 · 8 citations)
   LLMs can write elaborate reasoning but are poor at judging whether their own answers are right. Existing fixes have drawbacks. A separate verifier model adds cost. Multi-stage RL pipelines such as SCoRe need a warm-up phase to avoid model collapse, where the second attempt barely differs from the first. PAG instead…

🤝 Co-authors here

12 of 13 co-authors with a page, the most papers together first.

• Chao Xin ×2
• Guanlin Liu ×2
• Yu Yue ×2
• Zheng Wu ×2
• Liang Huang
• Ning Dai
• Renjie Zheng
• Ruofei Zhu
• Wei Shen
• Wenlei Shi
• Wenyuan Xu
• Xing Jin

🔗 Cited by

Papers of our library cite their works 58 times (of the 149,439 papers whose bibliographies we read); 32 citing papers with a page here →.

Updated 3 November 2026

This page in Markdown → https://md.research.elamur.ai/free/en/authors/lin-yan/