Biography
I am a Ph.D. student (2023-2028) jointly at the College of Computer Science and Artificial Intelligence, Fudan University, and the Shanghai Innovation Institute, advised by Prof. Xipeng Qiu and Prof. Xuanjing Huang. I received my bachelor’s degree with honors from Northeastern University (2019-2023).
My current research focuses on AI Scientist, reinforcement learning, and multimodal reasoning.
I am always open to academic collaboration and discussion. Please feel free to contact me at jqtong25@m.fudan.edu.cn.
Selected Publications
AI Can Learn Scientific Taste
Jingqi Tong, Mingzhe Li, Hangcheng Li, Yongzhuo Yang, Yurong Mou, Weijie Ma, Zhiheng Xi, Hongji Chen, Xiaoran Liu, Qinyuan Cheng, et al.
▸ 100k+ total views on X. Discussed by 166 X posts from 153 X users, including Ethan Mollick (LinkedIn), alphaXiv, HuggingPapers, @ihtesham2005, @anishmoonka.
▸ Ranked in the top ~0.1% of comparable-age arXiv outputs and the top ~0.32% overall for impact across news and social media, according to Altmetric Attention Score rankings.
▸ The work was followed up on and cited by Anthropic's TASTE and Stanford's GIANTS.
Thinking with Video: Video Generation as a Promising Multimodal Reasoning Paradigm
Jingqi Tong, Yurong Mou, Hangcheng Li, Mingzhe Li, Yongzhuo Yang, Ming Zhang, Qiguang Chen, Tianyi Liang, Xiaomeng Hu, Yining Zheng, et al.
Game-RL: Synthesizing Multimodal Verifiable Game Data to Boost VLMs' General Reasoning
Jingqi Tong, Jixin Tang, Hangcheng Li, Yurong Mou, Ming Zhang, Jun Zhao, Yanbo Wen, Fan Song, Jiahao Zhan, Yuyang Lu, et al.
▸ Cumulative downloads of the GameQA original and derived datasets have exceeded 550k.
▸ During my internship at ByteDance, synthesized 200K interleaved image-text samples for Seedream using this data synthesis method.
▸ Adopted by
Vero (Princeton University),
VisWorld-Eval (THUML & ByteDance Seed),
ReaForest (Peking University & WeChat AI, ICML 2026),
UniREditBench (Shanghai Innovation Institute)
+6 more
,
Beacon (Peking University & Kuaishou Kling),
MoTiF (Shanghai AI Lab),
MMFineReason (Shanghai AI Lab),
DeepVision-103K (Alibaba Group & SJTU),
VR-Bench (Tsinghua University & DeepWisdom),
Gym-V (National University of Singapore).
Exploring the Compositional Deficiency of Large Language Models in Mathematical Reasoning Through Trap Problems
Jun Zhao, Jingqi Tong, Yurong Mou, Ming Zhang, Qi Zhang, Xuanjing Huang
LLaMA-MoE: Building Mixture-of-Experts from LLaMA with Continual Pre-Training
Tong Zhu, Xiaoye Qu, Daize Dong, Jiacheng Ruan, Jingqi Tong, Conghui He, Yu Cheng
All Publications
21 papers · Google Scholar
Auto-research and RSI Agent
Multimodal Reasoning and Agent
arXiv 2026MOVA: Towards Scalable and Synchronized Video-Audio Generation,arXiv 2026VBVR-Pro: A Scalable and Verifiable Suite for Native Visual Reasoning,ACL 2026VideoPro: Adaptive Program Reasoning for Long Video Understanding,ACL 2026OMIBench: Benchmarking Olympiad-Level Multi-Image Reasoning in Large Vision-Language Models,EMNLP 2026MotionInsight: Diagnosing Object Motion Deficiencies in Generated Videos,
LLM Reasoning and Agent
ICLR 2026 SubmittedWorld Potential Model: Pretrained World Knowledge as Progress Potentials,ICLR 2026 SubmittedLiveMACE: Process-Aware Evaluation of LLM Agent Capabilities in Evolving Markets,ICLR 2026 SubmittedCAFE: Self-Improving Search Agents Need Co-Evolving Feedback,
LLM Foundations & Evaluation
NeurIPS · Vol. 38 2026Understanding Parametric and Contextual Knowledge Reconciliation within Large Language Models,AAAI 2025LLMEval-3: A Large-Scale Longitudinal Study on Robust and Fair Evaluation of Large Language Models,ACL 2026Beyond Scaling: Measuring and Predicting the Upper Bound of Knowledge Retention in Language Model Pre-Training,EMNLP 2025LLMEval-Med: A Real-world Clinical Benchmark for Medical LLMs with Physician Validation,arXiv 2025SpeechRole: A Large-Scale Dataset and Benchmark for Evaluating Speech Role-Playing Agents,
Invited Talks
Paradigm Shift in Intelligence Evolution
AI Can Learn Scientific Taste
Thinking with Video: Video Generation as a Promising Multimodal Reasoning Paradigm
Qwen Team Invited Talk. Carnegie Mellon University (CMU) - Video Model Journal Club Invited. NICE NLP Invited. AI TIME Invited.
Game-RL: Synthesizing Multimodal Verifiable Game Data to Boost VLMs' General Reasoning
Qwen Team and MiMo Team Invited Talk