Welcome to DeepAnalysis Homepage

We are the team of Deep Analysis Research at Alibaba Cloud.

We study LLM’s mechanistic interpretability and agents.

We are excited to expand our team and are currently seeking Research Interns to join us on our journey. For more details and to apply, please visit Learn More.

---

Selected Blogs

• 结构化
  A talk on what is structurization, why structurization is important for humans and LLMs and how to do structurization.
  blogpost

---

Selected work

• [New!🎉🎉] Interpret and Improve In-Context Learning via the Lens of Input-Label Mappings
  Chenghao Sun, Zhen Huang, Yonggang Zhang, Le Lu, Houqiang Li, Xinmei Tian, Xu Shen, Jieping Ye
  ACL 2025
  paper | repo

• [New!🎉🎉] Visual Evidence Prompting Mitigates Hallucinations in Large Vision-Language Models
  Wei Li, Zhen Huang, Houqiang Li, Le Lu, Yang Lu, Xinmei Tian, Xu Shen, Jieping Ye
  ACL 2025
  paper | repo

• [New!🎉🎉] Tracing and Dissecting How LLMs Recall Factual Knowledge for Real World Questions
  Yiqun Wang, Chaoqun Wan, Sile Hu, Yonggang Zhang, Xiang Tian, Yaowu Chen, Xu Shen, Jieping Ye
  ACL 2025
  paper | repo

• [New!🎉🎉] NeuronMerge: Merging Models via Functional Neuron Groups
  Wangyun Gu, Qianghua Gao, Xu Shen, Jieping Ye
  ACL 2025
  paper | repo

• Interpreting and Improving Large Language Models in Arithmetic Calculation
  Wei Zhang, Chaoqun Wan, Yonggang Zhang, Yiu-ming Cheung, Xinmei Tian, Xu Shen, Jieping Ye
  ICML 2024 (Oral, top5%)
  paper

• From Yes-Men to Truth-Tellers: Addressing Sycophancy in Large Language Models with Pinpoint Tuning
  Wei Chen, Zhen Huang, Liang Xie, Binbin Lin, Houqiang Li, Le Lu, Xinmei Tian, Deng Cai, Yonggang Zhang, Wenxiao Wang, Xu Shen, Jieping Ye
  ICML 2024
  paper | repo

• Enhancing Multiple Dimensions of Trustworthiness in LLMs via Sparse Activation Control
  Yuxin Xiao, Chaoqun Wan, Yonggang Zhang, Wenxiao Wang, Binbin Lin, Xiaofei He, Xu Shen, Jieping Ye
  NeurIPS 2024
  paper

• Leveraging Submodule Linearity Enhances Task Arithmetic Performance in LLMs
  Rui Dai, Sile Hu, Xu Shen, Yonggang Zhang, Xinmei Tian, Jieping Ye
  ICLR 2025
  paper | repo

• Knowledge Graph Finetuning Enhances Knowledge Manipulation in Large Language Models
  Hanzhu Chen, Xu Shen, Jie Wang, Zehao Wang, Qitan Lv, Junjie He, Rong Wu, Feng Wu, Jieping Ye
  ICLR 2025
  paper

• Interpretable Composition Attribution Enhancement for Visio-linguistic Compositional Understanding
  Wei Li, Zhen Huang, Xinmei Tian, Le Lu, Houqiang Li, Xu Shen, Jieping Ye
  EMNLP 2024
  paper

• SAC-KG: Exploiting Large Language Models as Skilled Automatic Constructors for Domain Knowledge Graphs
  Hanzhu Chen, Xu Shen, Qitan Lv, Jie Wang, Xiaoqi Ni, Jieping Ye
  ACL 2024
  paper

---

Favorite talks

• An Observation on Generalization

---

(last updated: Apr 2025)