✨ About me

Hi, everyone! I am currently a Third-year PhD student (10.2024-) at King’s College London, NLP group, School of Informatics. I am fortunate to be supervised by Dr. Lin Gui and Prof. Yulan He. I finished my MSC AI at the University of Edinburgh and my BEng EEE project jointly at the University of Edinburgh and North China Electric Power University(NCEPU). I am fortunate to be supervised by Prof. Frank Keller for my MSC and Dr. Jiabin Jia for my BEng.

I am currently a Qingyun intern at Tencent YuanBao for Agent Memory, welcome any chat with me!

⭐ I am actively seeking job opportunities. Please feel free to contact me about relevant openings.

Email / Twitter / RedNote / Google Scholar / GitHub

🔍 Research Summary

My research focuses on Agent Memory, Agent Harnesses, and Self-Improving AI Agents, aiming to build reliable and adaptive LLM agents for complex, long-horizon tasks. My work explores how agents can organise and leverage past experience, how their capabilities can be evaluated in realistic environments, and how their memory and supporting systems can evolve through feedback. My current research interests include:

• Agent Memory and Retrieval. I investigate how agents organise, retrieve, and utilise past experience for reliable reasoning and decision-making. Building on my earlier work in EEE-QA (LREC-COLING 2024), EmbQA (ACL 2025), and SPS (AAAI 2026 Oral), I developed xMemory (NeurIPS 2026), a hierarchical memory framework extending structured retrieval beyond conventional RAG. My ongoing work explores adaptive memory structures and utility-driven memory evolution.

• Agent Harnesses and Evaluation. I explore how memory, retrieval, and external agent systems support reliable behaviour in complex environments. My recent work includes [GroupAssistBench (ICLR 2027 Submission)], evaluating memory-informed group assistance, and [CoMateEval (ICLR 2027 Submission)], benchmarking LLMs in dynamic multi-party collaboration.

• Self-Improving Agents (RSI). My long-term interest is in enabling agents to iteratively improve their memory, harnesses, and behaviours through experience and feedback.

Beyond these directions, my broader research experience includes efficient LLM reasoning (CODI, EMNLP 2025), robustness and hallucination detection (OSCR-Attack, ACL Findings 2026; ICML 2026), and multimodal reasoning and generation (VQG, ACL ALVR 2024; Human Motion Generation, TPAMI 2025).

🔥 News

🚀 I am always open to new collaborations and engaging discussions. Feel free to reach out if you are interested in working together or just want to chat!

📚 Selected Publications

† Equal contribution. * Corresponding author. Selected and representative works are shown below. For a complete publication list, please visit my Google Scholar .

ICLR 2027 Submission

One Assistant, Many Memories: Benchmarking Real-World Group Assistance When Recall Is Not Enough

Zhanghao Hu†, Linhai Zhang†, Qian Zhao, Qi Zhu, Yuan Hua, Di Liang, Jiasheng Si, Xin Zhao, Yulan He, Zhumin Chen, Lin Gui

Paper

A real-world group-memory benchmark showing that successful recall does not necessarily translate into effective assistance, especially when memories are distributed or conflicting.

ICLR 2027 Submission

CoMateEval: Benchmarking LLMs Across Team Roles in Dynamic Multi-Party Collaboration

Xianjie Wu†, Zhanghao Hu†, Tianle Gu†, Wuzhenghong Wen, Xiaohang Xu, Yuhui Wang, Yujia Chen, Naifu Liang, et al.

Paper

A benchmark for evaluating how well LLMs fulfil diverse team responsibilities across dynamic multi-party collaborative environments.

NeurIPS 2026

Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation

Zhanghao Hu†, Qinglin Zhu†, Runcong Zhao, Di Liang, Hanqi Yan, Yulan He, Lin Gui

Project / Code / Paper

A hierarchical agent-memory framework that decouples correlated interactions into semantic components and aggregates them for structure-aware retrieval beyond conventional top-k RAG.

Recommendation: Alan Turing Institute · VentureBeat · DAIR.AI · Maxim AI · EmergentMind

AAAI 2026 Oral

Beyond Perplexity: Let the Reader Select Retrieval Summaries via Spectrum Projection Score

Zhanghao Hu, Qinglin Zhu, Siya Qi, Yulan He, Hanqi Yan, Lin Gui

Oral 🌟 around 4% (900 / 23,680) / Project / Paper / Code

A reader-aware metric and inference-time retrieval controller for selecting summaries according to their alignment with downstream LLM representations.

ACL 2025 Main

Beyond Prompting: An Efficient Embedding Framework for Open-Domain Question Answering

Zhanghao Hu, Hanqi Yan, Qinglin Zhu, Zhenyi Shen, Yulan He, Lin Gui

Project / Paper

An embedding-level framework for refining retrieval and diversifying answer generation without relying on additional prompting.

LREC-COLING 2024

EEE-QA: Exploring Effective and Efficient Question-Answer Representations

Zhanghao Hu†, Yijun Yang†, Junjie Xu†, Yifu Qiu, Pinzhen Chen

Paper

Studies effective and memory-efficient question-answer representations while maintaining competitive QA performance.

EMNLP 2025 Main

CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillation

Zhenyi Shen, Hanqi Yan, Linhai Zhang, Zhanghao Hu, Yali Du, Yulan He

Paper

Compresses explicit chain-of-thought reasoning into continuous latent representations through self-distillation.

ACL 2026 Findings

OSCR-Attack: One-Shot Character Level Attacks through Self-Optimizing Continuous Relaxation

Lingyi Kong†, Zhuo Liu†, Zhanghao Hu†, Qilong Qiu, Yutao Yang, Jingjing Xue, Zheng Wang, Lin Gui, Feiping Nie

Paper

An efficient character-level adversarial attack that transforms discrete perturbation choices into continuous optimisation for one-shot attacks on LLMs.

ICML 2026

Detecting Contextual Hallucinations in Large Language Models with Frequency-Aware Attention

Siya Qi, Yudong Chen, Runcong Zhao, Qinglin Zhu, Zhanghao Hu, Wei Liu, Yulan He, Zheng Yuan, Lin Gui

Paper

Detects contextual hallucinations through frequency-aware attention features that capture fragmented and unstable grounding during generation.

TPAMI 2025

Human Motion Video Generation: A Survey

Haiwei Xue, Xiangyang Luo, Zhanghao Hu, Xin Zhang, Xunzhi Xiang, Yuqin Dai, et al.

Paper

A comprehensive taxonomy and survey of human-motion video generation covering the full generation pipeline and major task settings.

ACL ALVR 2024

Causal and Temporal Inference in Visual Question Generation by Utilizing Pre-trained Models

Zhanghao Hu, Frank Keller

Paper

Uses pretrained vision-language representations to generate questions requiring causal and temporal inference over videos.

😆 Mentee

• LLM Safety
  • Lingyi Kong (LLM Security and Attack)
• Multi-modal Alignment
  • Zipeng Zhu (Image Edit)

💬 Invited Talks

• 02/2026. Queen Mary University of London, NLP Group

Professional Service

• Volunteer:
  • AAAI 2026
• Reviewers:
  • NLP: EMNLP 2025, ACL 2026，EMNLP 2026
  • AI/ML: AAAI 2026, ICML 2026，NIPS 2026，ICLR 2026

🎖 Honors and Awards

• 2023.01 IBM Shortlist for Best Project in Machine Learning Practical Course, Ranked 5/103.

• 2022.06 Outstanding Graduate Award, NCEPU

• 2022.05 £3000 Scholarship, University of Edinburgh, For Excellent 2+2 International Students

• 2021.05 £2500 Scholarship, University of Edinburgh, For Excellent 2+2 International Students

• 2020.10 Third Prize Academic Scholarship, NCEPU, Awarded to top 10% of students

• 2019.10 Second Prize Academic Scholarship, NCEPU, Awarded to top 5% of students

📖 Educations

• 2022.09 – 2023.11, MSc in Artificial Intelligence, University of Edinburgh, Distinction Degree, ranked top ~10%

• 2020.09 – 2022.05, Bachelor in Electronics and Electrical Engineering, University of Edinburgh, First-Class Honour Degree, ranked top ~10%

• 2018.09 – 2020.06, Bachelor in Electrical Engineering and Its Automation, North China Electric Power University (NCEPU) Ranked ~15%

💻 Internships

• 2024.04 - 2024.08, Research Intern at 01.AI.

🌍 Visitor Analytics