Scholar
Hengyi Cai
Google Scholar ID: Kz-r34UAAAAJ
Institute of Computing Technology, Chinese Academy of Sciences
Citations & Impact
Citations
1,566
H-index
12
i10-index
17
Publications
20
Co-authors
10
list available
Contact
Publications
Resume (English only)
Academic Achievements
- Multiple publications in top-tier venues such as ACL, EMNLP, NeurIPS, KDD, and SIGIR. Selected publications include:
- - MARA: A Multimodal Adaptive Retrieval-Augmented Framework for Document Question Answering
- - Towards AI Search Paradigm
- - Enhancing Retrieval-Augmented Generation via Evidence Tree Search
- - Multi-Agent Proactive Information Seeking with Adaptive LLM Orchestration for Non-Factoid Question Answering
- - Tool Learning with Large Language Models: A Survey
- - From Exploration to Mastery: Enabling LLMs to Master Tools via Self-Driven Interactions
- - From Prompting to Alignment: A Generative Framework for Query Recommendation
- - PA-RAG: RAG Alignment via Multi-Perspective Preference Optimization
- - Explainability for Large Language Models: A Survey
- - Towards Completeness-Oriented Tool Retrieval for Large Language Models
- - AdaSwitch: Adaptive Switching between Small and Large Agents for Effective Cloud-Local Collaborative Learning
- - Towards Verifiable Text Generation with Evolving Memory and Self-Reflection
- - Cross-model Control: Improving Multiple Large Language Models in One-time Training
- - Text-Video Retrieval via Multi-Modal Hypergraph Networks
Research Experience
- Currently a Research Scientist at Baidu, leading the Baidu Search Foundation Model Team. Focuses on developing an efficient foundation model system for search applications.
Education
- Ph.D. from the Institute of Computing Technology, Chinese Academy of Sciences in 2021, graduating as an Outstanding Graduate.
Background
- Research Scientist at Baidu Inc., leading the Baidu Search Foundation Model Team. Specializes in the development and optimization of large-scale language models, particularly in MoE sparsification strategies, pre-training task design, post-training optimization, and reinforcement learning-based reasoning enhancement. His research spans text generation, question answering, and retrieval-augmented language models.
Miscellany
- Actively serves on the program committees of leading conferences.
