Hongcheng Gao(高鸿成) is an incoming PhD student at College of AI at Tsinghua University. He may receive his M.S. degree in July 2026. He received his Bachelor's degree of Computer Science and Technology at Chongqing University in June 2023.

His research interests include large reasoning models, AI agents and unified models, spanning both textual and multimodal domains (especially spatial tasks).

His team is looking for collaborators and interns. Please feel free to reach out.

🔥 News

• 2025.11: One papers accepted to AAAI’26.
• 2025.09: Two papers accepted to Neurips’25 with one Oral.
• 2025.06: One paper accepted to ICCV 2025.

• 2025.06: Invited to serve as Senior PC Member for CIKM’25.

• 2025.01: Two papers accepted to ICLR 2025 with one Oral.
• 2024.10: Selected as NeurIPS 2024 Top Reviewer.
• 2024.09: Three papers accepted to NeurIPS 2024 with two Spotlights.
• 2024.09: Two papers accepted to EMNLP 2024.
• 2024.05: One paper accepted to ACL 2024.
• 2024.04: One paper accepted to NAACL 2024.
• 2024.01: 🎉 One paper accepted to ICLR 2024.
• 2023.09: 🎉 One paper accepted to NeurIPS 2023 (D&B Track).
• 2023.05: 🎉 One paper accepted to ACL 2023!
• 2022.10: 🎉 Two papers accepted by EMNLP 2022!
• 2022.08: 🎉 One paper accepted by NAACL 2022!

📝 Preprints

* indicates equal contribution.

• GuardReasoner: Towards Reasoning-based LLM Safeguards [Paper]
  Yue Liu, Hongcheng Gao, Shengfang Zhai, Jun Xia, Tianyi Wu, Zhiwei Xue, Yulin Chen, Kenji Kawaguchi, Jiaheng Zhang, Bryan Hooi

• StruEdit: Structured Outputs Enable the Fast and Accurate Knowledge Editing for Large Language Models [Paper]
  Baolong Bi, Shenghua Liu, Yiwei Wang, Lingrui Mei, Hongcheng Gao, Junfeng Fang, Xueqi Cheng

• Evaluating the Robustness of Text-to-image Diffusion Models against Real-world Attacks [Paper]
  Hongcheng Gao, Hao Zhang, Yinpeng Dong, Zhijie Deng.
  

• Kimi k1. 5: Scaling Reinforcement Learning with LLMs [Paper]
  Kimi Team

• Kimi-vl technical report [Paper]
  Kimi Team

• Kimi k2: Open agentic intelligence [Paper]
  Kimi Team

📝 Publications

* indicates equal contribution.

2025

• Representation Entanglement for Generation: Training Diffusion Transformers Is Much Easier Than You Think [Paper]
  Ge Wu, Shen Zhang, Ruijing Shi, Shanghua Gao, Zhenyuan Chen, Lei Wang, Zhaowei Chen, Hongcheng Gao, Yao Tang, Jian Yang, Ming-Ming Cheng, Xiang Li.
  NeurIPS 2025

• Guardreasoner-vl: Safeguarding vlms via reinforced reasoning [Paper]
  Yue Liu, Shengfang Zhai, Mingzhe Du, Yulin Chen, Tri Cao, Hongcheng Gao, Cheng Wang, Xinfeng Li, Kun Wang, Junfeng Fang, Jiaheng Zhang, Bryan Hooi
  NeurIPS 2025

• Meta-Unlearning on Diffusion Models: Preventing Relearning Unlearned Concepts [Paper]
  Hongcheng Gao*, Tianyu Pang*, Chao Du, Taihang Hu, Zhijie Deng, Min Lin.
  ICCV 2025

• Spider 2.0: Evaluating language models on real-world enterprise text-to-sql workflows [Paper]
  Fangyu Lei*, Jixuan Chen*, Yuxiao Ye, Ruisheng Cao, Dongchan Shin, Hongjin Su, Zhaoqing Suo, Hongcheng Gao, Wenjing Hu, Pengcheng Yin, Victor Zhong, Caiming Xiong, Ruoxi Sun, Qian Liu, Sida Wang, Tao Yu.
  ICLR 2025

• Is factuality decoding a free lunch for llms? evaluation on knowledge editing benchmark [Paper]
  Baolong Bi, Shenghua Liu, Yiwei Wang, Lingrui Mei, Junfeng Fang, Hongcheng Gao, Shiyu Ni, Xueqi Cheng.
  ICLR 2025

2024

• Spider2-V: How Far Are Multimodal Agents From Automating Data Science and Engineering Workflows? [Paper]
  Ruisheng Cao, Fangyu Lei, Haoyuan Wu, Jixuan Chen, Yeqiao Fu, Hongcheng Gao et al.
  NeurIPS 2024(Spotlight）

• Leveraging Catastrophic Forgetting to Develop Safe Diffusion Models against Malicious Finetuning[Paper]
  Jiadong Pan*, Hongcheng Gao*, Zongyu Wu, Tanghang Hu, Li Su, Qingming Huang, Liang Li.
  NeurIPS 2024(Spotlight）

• Token Merging for Training-Free Semantic Binding in Text-to-Image Synthesis[Paper]
  Taihang Hu, Linxuan Li, Joost van de Weijer, Hongcheng Gao, Fahad Shahbaz Khan, Jian Yang, Ming-Ming Cheng, Kai Wang, Yaxing Wang.
  NeurIPS 2024

• AdaMoE: Token-Adaptive Routing with Null Experts for Mixture-of-Experts Language Models [Paper]
  Zihao Zeng*, Yibo Miao*, Hongcheng Gao, Hao Zhang, Zhijie Deng.
  Findings of EMNLP 2024.
  

• Adaptive Token Biaser: Knowledge Editing via Biasing Key Entities [Paper]
  Baolong Bi, Shenghua Liu, Yiwei Wang, Lingrui Mei, Hongcheng Gao, Yilong Xu, Xueqi Cheng.
  Findings of EMNLP 2024.
  

• Efficient Detection of LLM-generated Texts with a Bayesian Surrogate Model [Paper]
  Zhijie Deng*, Hongcheng Gao*, Yibo Miao, Hao Zhang.
  Findings of ACL 2024.
  

• Universal Prompt Optimizer for Safe Text-to-Image Generation [Paper]
  Zongyu Wu*, Hongcheng Gao*, Yueze Wang, Xiang Zhang, Suhang Wang.
  NAACL 2024

• Generative Pretraining in Multimodality [Paper]
  Quan Sun*, Qiying Yu*, Yufeng Cui*, Fan Zhang*, Xiaosong Zhang*, Yueze Wang, Hongcheng Gao, Jingjing Liu, Tiejun Huang, Xinlong Wang
  ICLR 2024

2023

• Revisiting Out-of-distribution Robustness in NLP: Benchmark, Analysis, and LLMs Evaluations [Paper]
  Lifan Yuan, Yangyi Chen, Ganqu Cui, Hongcheng Gao, Fangyuan Zou, Xingyi Cheng, Heng Ji, Zhiyuan Liu, Maosong Sun.
  NeurIPS 2023 (D&B Track)

• From Adversarial Arms Race to Model-centric Evaluation: Motivating a Unified Automatic Robustness Evaluation Framework [Paper]
  Yangyi Chen*, Hongcheng Gao*, Ganqu Cui*, Lifan Yuan, Dehan Kong, Hanlu Wu, Ning Shi, Bo Yuan, Longtao Huang, Hui Xue, Zhiyuan Liu, Maosong Sun, Heng Ji.
  Findings of ACL 2023.
  

2022

• Why Should Adversarial Perturbations be Imperceptible? Rethink the Research Paradigm in Adversarial NLP [Paper]
  Yangyi Chen*, Hongcheng Gao*, Ganqu Cui, Fanchao Qi, Longtao Huang, Zhiyuan Liu, Maosong Sun.
  EMNLP 2022.
  

• Textual Backdoor Attacks Can Be More Harmful via Two Simple Tricks [Paper]
  Yangyi Chen*, Fanchao Qi*, Hongcheng Gao, Zhiyuan Liu, Maosong Sun.
  EMNLP 2022.
  

• Exploring the Universal Vulnerability of Prompt-based Learning Paradigm [Paper]
  Lei Xu, Yangyi Chen, Ganqu Cui, Hongcheng Gao, Zhiyuan Liu.
  Findings of NAACL 2022.
  

💻 Projects

• RobTest: an open-source toolkit for textual model robustness evaluation .
  
• Spider2-V: Spider2-V is a multimodal agent benchmark spanning across the entire data science and engineering workflow
• OpenAttack: an open-source Python-based textual adversarial attack toolkit.
  
• Emu: an open multimodal generalist for both image-to-text and text-to-image tasks.
  

📄 Academic Services

• Area Chiar/Senior PC Member:
  • 2025: NeurIPS, CIKM
• Reviewer/PC Member:
  • 2025: ICML, AISTATS, ICCV, ACL, ARR, ECAI
    
  • 2024: NeurIPS(Top Reviewer), ICLR, CVPR, ACL, EMNLP, NAACL, ARR, Machine Learning, AISTATS.

📖 Experiences

• 2018.09 - 2023.06, Undergraduate Student at Chongqing University.
• 2021.06 - 2023.06, Intern at THUNLP, Tsinghua University. Mentor: Prof. Zhiyuan Liu