👋 About me
Hi! I’m Junlin Liu, a graduate student majoring in Artificial Intelligence at 
Institute of Automation Chinese Academy of Sciences (CASIA) and 
University of Chinese Academy of Sciences (UCAS).
I’m currently a research intern focusing on Search Agents and Recursive Self-Improvement at 
Ant Ling Team(Ant-Star Plan-A Program). Previously, I held research intern positions at 
Alibaba Qwen MOS Lab, 
Meituan LongCat Team and 
Baidu Ernie Team.
Research Interests：
- 🧬 Self-Evolving Agent — Building adaptive autonomous agents that continuously improve through RL and agent–environment co-evolution.
- 🤖 Agentic Reinforcement Learning — Training general agent intelligence via fundamental RL-based optimization.
- 🔬 LLM Evaluation and Benchmarking — Evaluating and improving the reasoning capabilities of foundation models.
💬 If you are interested in collaborating with me, please feel free to add me on WeChat: AaronLiu0702.
蚂蚁星学术开放日 (北京·2026.11)
B站"B-UP"人才计划 (北京·2026.09)
中国科学院大学 人工智能学院 (北京·2026.06)
阿里星热爱之旅 (北京·2026.05)
VALSE2026 (武汉·2026.05)
腾讯青云计划 (北京·2026.03)
优秀学生标兵 (成都·2024.12)
🔥 News
- 2026.11: 🚀 Ling-3.2-Flash (560B-A25B, a native multimodal model) is out! So proud to have contributed to it!
Blog 
- 2026.10: 🚀 Ling-3.1-Flash (560B-A25B) is released! So proud to have contributed to it! Blog
HF 
MS - 2026.09: 🥂 Invited to Bilibili “B-UP” Talent Program Networking Event!
- 2026.09: 🌟 Joined the Ling Foundation LLM Team at 🐜 AntGroup as a Research Intern through the Ant-Star Plan-A Program!
- 2026.08: 🎉 One paper has been accepted by ICONIP 2026! DRG-MAPPO！See you in Sydney 🇦🇺!
- 2026.07: ✈️ I was honored to be invited to participate in the Qingyun|UBIQuent|Apex Intelligence Talent Dinner at ICML 2026!
- 2026.05: ✈️ I was honored to be invited to participate in the “AI Talent Corner” at VALSE 2026!
- 2026.04: 🧩 We released General365, advancing LLM reasoning beyond domain-specific tasks toward general real-world scenarios.
- 2026.04: 🎉 One paper has been accepted by ACL 2026! AMO-Bench！See you in San Diego 🇺🇸!
- 2026.03: 🎉 One paper has been accepted by IJCNN 2026! ACE-MAPPO！See you in Maastricht 🇳🇱!
- 2026.03: 🥂 I was honored to be invited to participate in the Tencent QingYun Talent Program! “Stars of the Future”
- 2025.06: 🎉 I received my B.E. degree from Sichuan Agricultural University (SICAU), awarded the Outstanding Graduate and Outstanding Thesis Award, ranking 1st/198 in overall GPA for three years (2022-2025) ! 🌟🌟 Student Spotlight🌟🌟
💻 Internship Experience
- 2026.09 – Present｜
AntGroup, 
Ling Foundation LLM Team. (Ant-Star Plan-A Program)
- I work on enhancing the agentic search capability of Ling series models, improving performance on benchmarks such as Draco and WideSearch through data synthesis and distillation, supervised fine-tuning, and agentic RL.
- 🚀 Model Releases Ling-3.1-Flash, Ling-3.1-Flash-1019, Ling-3.2-Flash.
- Mentor: YangYang Hou, XingYu Lu.
- 2026.04 – 2026.08｜
Alibaba, Qwen Business Unit, MOS Lab.
- I worked on search agents and LLM post-training, developing MAPD for on-policy multi-agent distillation, DecoEvo for co-evolution of solver and rubric-generator skills, and HarnessOpt for long-horizon harness self-improvement.
- Mentor: Bowen Zhang, Yuan Shen.
- 2025.09 – 2026.04｜
Meituan, LongCat Foundation LLM Team.
- I worked on mathematical and general reasoning of LLMs, building AMO-Bench and General365 for benchmarking. I also developed CRPO and PCSD for agentic RL via contrastive optimization and self-distillation, and ClawTrack for trace-level agent evaluation.
- 🚀 Model Releases LongCat-Flash-Thinking-2601, LongCat-Flash-Prover.
- Mentor: Shengnan An, Xuezhi Cao, Xunliang Cai.
- 2025.07 - 2025.09｜
Baidu, ERNIE Foundation LLM Team.
- I worked on post-training data engineering and multimodal evaluation for foundation models.
- Mentor: Dr.Xin Wang.
📝 Publications
AMO-Bench: Large Language Models Still Struggle in High School Math Competitions.
🧑💻 Junlin Liu, Shengnan An, Shuang Zhou, Dan Ma, Yehao Lin, Xinxuan Lv, Xuanlin Wang, Xiaoyu Li, Ziwen Wang, Xuezhi Cao, Xunliang Cai.
🏛️ The 64th Annual Meeting of the Association for Computational Linguistics, 2026.
Paper
Code
ProjectPage
BIB
General365: Benchmarking General Reasoning in LLMs Across Diverse and Challenging Tasks.
🧑💻 Junlin Liu, Shengnan An, Shuang Zhou, Dan Ma, Shixiong Luo, Ying Xie, Yuan Zhang, Wenling Yuan, Yifan Zhou, Xiaoyu Li, Ziwen Wang, Xuezhi Cao, Xunliang Cai.
From Proprietary to Open-Source: Bridging the Distribution Gap via Multi-Agent Protocol Distillation in Agentic Search.
🧑💻 Junlin Liu, Chunji Lv, Jiangwang Chen, Zixin Song, Shuaiyu Zhou, Xingjian Wu, Kailin Jiang,
Jinyang Wu, Bohan yu, Chenxi Zhou, Xiao Yang, Da Zhu, Guanjun Jiang.
DecoEvo: Score-Decoupled Co-Evolution of Solver and Rubric-Generator Skills in Text Space.
🧑💻 Jiangwang Chen*, Zixin Song*, Junlin Liu*, Shuaiyu Zhou, Haiyan Wu, Haihan Shi, Chenxi Zhou, Hanqing Li, Xiao Yang, Da Zhu, Guanjun Jiang.
HarnessOpt: Long-Horizon Harness Self-Improvement via Accumulated Optimizer State.
🧑💻 Shuaiyu Zhou*, Jiaying Zhang*, Junlin Liu*, Jiangwang Chen, Zixin Song, Haiyan Wu, Hanqing Li, Jinzhou Song, Xiao Yang, Da Zhu, Guanjun Jiang.
ClawTrack: Towards Trace-Level Evaluation and Improvement of Real-World Autonomous Agents.
🧑💻 Xingjian Wu, Xuhang Zhu, Xingchen Liu, Junlin Liu, Jianing Wang, Linsen Guo, Xiaoyu Li,
Xuezhi Cao, Xunliang Cai.
CreativeBench2.0: Benchmarking Creative Method Discovery and Foresight Toward Recursive Self-Improvement.
🧑💻 Zihan Wang, Chenxu Zhang, Linji Hao, Zhengzhou Cai, Jinyang Wu, Zhiyuan Yao, Zhengxi Lu, Chengcheng Han, Yueqing Sun, Huichi Zhou, Junlin Liu, Xierui SONG, Yujiu Yang.
Beyond Relevance-Centric Retrieval: Rubric-Oriented Document Set Selection and Ranking.
🧑💻 Kailin Jiang, Lei Liu, Jian Xi, Hui Xu, Junlin Liu, Baochen Fu, Shaoqing Ren, Bin Li, Vichwang,
Yu Lu, Haibo Shi.
HalluAgent: Type-Conditioned Visual Evidence for Hallucination Mitigation in Large Vision-Language Models.
🧑💻 Ruipeng Zhang, Zhangtianyi Chen, Zixuan Huang, Junlin Liu, Tong Ji, Ruibo Duan.
Latent Rule Reasoning.
🧑💻 Yuxuan Liang, Xu Li, Haotian Chen, Xiaolei Chen, Zhe Liu, Yuchuan Wu, Rui Zhu, Bohan Zhang, Wenjuan Meng, Junlin Liu, Haiyang Yu, Fan Shi, Xiangyang Xue.
SAW: Sink-Aware Weighting for Effective Video Reasoning.
🧑💻 Gengyuan Liu, Shaoxiang Chen, Qinwen Wu, Jiajun Dong, Zhenhao Wang, Hong Wang, Haiyang Xin, Junxuan Du, Junlin Liu, jixv, Jiacong Wang.
DRG-MAPPO: Hierarchical Dynamic Role-Graph Multi-Agent Reinforcement Learning for Cooperative Air Combat.
🧑💻 Junlin Liu, Yang Gao, Chengwei Li, Hui Chang, Xinchen Zhang, Zhijun Zhao, Hao Zhao.
🏛️ International Conference on Neural Information Processing, 2026. (Oral)
Evolutionary Enhanced Multi-Agent Reinforcement Learning for Cooperative Air Combat.
🧑💻 Chengwei Li*, Junlin Liu*, Yang Gao, Hui Chang, Xinchen Zhang, Hao Zhao.
🏛️ International Joint Conference on Neural Networks, 2026.
📖 Educations
- 2025.09 – present｜ Institute of Automation Chinese Academy of Sciences (CASIA).
- Master’s Degree in Artificial Intelligence.
- 2025.09 – present｜ University of Chinese Academy of Sciences (UCAS).
- Master’s Degree in Artificial Intelligence.
- 2021.09 – 2025.06｜
Sichuan Agricultural University (SiCAU).
- Bachelor’s Degree in Computer Science and Technology.
- 2015.09 – 2021.06｜
Chengdu Foreign Languages School (CFLS).
🎫 Academic Activities
- 2026.09, Beijing, “B-UP” Top Talent Program of Bilibili.
- 2026.07, Seoul, “Qingyun, UBIQuent, Apex Intelligence Talent Dinner” at ICML 2026.
- 2026.05, Wuhan, “AI Talent Corner” at VALSE 2026.
- 2026.03, Beijing, Tencent QingYun Program, “Stars of the Future: Technology Exchange Exhibition”.
- 2025.09, Beijing, Wave Summit 2025, Deep Learning Developers Conference.
🏆 Honors and Awards
- 2025.06, Outstanding Graduates of Sichuan Province, China (Top 2%).
- 2025.06, Outstanding Graduation Thesis (Top 1%).
- 2024.12, Outstanding Student Model (The highest honor of the university, awarded to only 10 students).
- 2024.12, China National Scholarship (Top 0.2%).
- 2023.12, Outstanding Student Scholarship (Top 1%).
📌 Academic Service
- Reviewer for: ACL Rolling (ACL, EMNLP, NAACL, EACL, AACL), NeurIPS, AAAI, IJCNN
- Reviewer for: Applied Sciences