Xin Sun (孙鑫)

PhD Student, University of Science and Technology of China (USTC)
Institute of Automation, Chinese Academy of Sciences (CASIA)

I am a PhD student at USTC, jointly supervised by Prof. Liang Wang (IEEE Fellow) and Associate Prof. Qiang Liu at the National Laboratory of Pattern Recognition (NLPR), CASIA. Since May 2026, I have been an Ant Star Intern in the Machine Intelligence Division at Ant Group, where I lead projects on Retrieval-Augmented Generation and LLM Agent Training.

My research focuses on Retrieval-Augmented Generation (RAG) and LLM Agent Training, spanning two core directions: (1) RAG & Deep Search — building reliable and adaptive retrieval-augmented systems that know when, what, and how to retrieve, toward autonomous deep research agents; (2) LLM Agent Training — empowering LLM-based agents through reinforcement learning and post-training to perform multi-turn exploration, tool use, structured reasoning, and learning from environment feedback. During my PhD, I have published 7 first-/co-first-author papers in top-tier conferences and journals, including 6 rated A by CCF or TH-CPL.

Email Google Scholar GitHub CV

News

2026.08 One paper accepted to Findings of EMNLP 2026: GAPD! [arXiv]
2026.08 Two new co-first-author papers under review at AAAI 2027: EviSD and HopRefusalBench. [EviSD] [HopRefusalBench]
2026.05 One paper accepted at ICML 2026: KBQA-R1!
2026.01 One paper accepted at ICASSP 2026: Predict the Retrieval!
2025.12 New preprint: KBQA-R1 — Reinforcing LLMs for Knowledge Base Question Answering. [arXiv]
2025.05 One paper accepted at ACL 2025 (Main Conference): Divide-Then-Align.
2024.11 Joined Ant Group's Machine Intelligence Division, working on RAG.
2024.10 Received the National Scholarship for Graduate Students (2024).

Publications

RAG, Deep Search & Agentic RL (Core Research)

KBQA-R1: Reinforcing Large Language Models for Knowledge Base Question Answering

Xin Sun, Zhongqi Chen, Xing Zheng, Qiang Liu, Shu Wu, Bowen Song, Zilei Wang, Weiqiang Wang, Liang Wang

ICML 2026 CCF-A

arXiv Code
GAPD: Gold-Action Policy Distillation for Agentic Reinforcement Learning in Knowledge Base Question Answering

Xin Sun, Zhongqi Chen, Xing Zheng, Qiang Liu, Shu Wu, Bowen Song, Zilei Wang, Weiqiang Wang, Liang Wang

EMNLP 2026, Findings TH-CPL A

arXiv
EviSD: Evidence-Conditioned Self-Distillation for Search-Augmented Agents

Jianan Xie*, Xin Sun*, Zhongqi Chen, Xing Zheng, Shu Wu, Bowen Song, Liang Wang

Under Review at AAAI 2027 Co-first Author

arXiv
HopRefusalBench: Diagnosing Refusal Failures in Search-Augmented Agents for Multi-Hop Reasoning

Jianan Xie*, Xin Sun*, Zhongqi Chen, Xing Zheng, Qiang Liu, Bowen Song

Under Review at AAAI 2027 Co-first Author

arXiv
Divide-Then-Align: Honest Alignment based on the Knowledge Boundary of RAG

Xin Sun, Jianan Xie, Zhongqi Chen, Qiang Liu, Shu Wu, Yuehe Chen, Bowen Song, Zilei Wang, Weiqiang Wang, Liang Wang

ACL 2025, Main Conference CCF-A

Paper Code
Predict the Retrieval! Test Time Adaptation for Retrieval Augmented Generation

Xin Sun, Zhongqi Chen, Qiang Liu, Shu Wu, Bowen Song, Weiqiang Wang, Zilei Wang, Liang Wang

ICASSP 2026 CCF-B

arXiv Code
ToolWeaver: Weaving Collaborative Semantics for Scalable Tool Use in Large Language Models

Bowen Fang, Qiang Liu, Yesheng Liu, Jiabing Yang, Xin Sun, Shu Wu, Baole Wei, Liang Wang

ICLR 2026 CAAI-A

arXiv
Multimodal Adaptive Retrieval Augmented Generation through Internal Representation Learning

Ruoshuang Du, Xin Sun, Qiang Liu, Bowen Song, Zhongqi Chen, Weiqiang Wang, Liang Wang

IJCNN 2026 CCF-C

arXiv

Other Works: Graph Learning & Trustworthy ML

DIVE: Subgraph Disagreement for Graph Out-of-Distribution Generalization

Xin Sun, Liang Wang, Qiang Liu, Shu Wu, Zilei Wang, Liang Wang

KDD 2024 CCF-A
Noise-Robust Semi-Supervised Learning for Distant Supervision Relation Extraction

Xin Sun, Qiang Liu, Shu Wu, Zilei Wang, Liang Wang

EMNLP 2023, Findings TH-CPL A
Uncertainty Calibration for Counterfactual Propensity Estimation in Recommendation

Wenbo Hu*, Xin Sun* (First Student Author), Qiang Liu, Liang Wang

IEEE TKDE CCF-A SCI Q1
Pin-Tuning: Parameter-Efficient In-Context Tuning for Few-shot Molecular Property Prediction

Liang Wang, Qiang Liu, Shaozhen Liu, Xin Sun, Shu Wu, Liang Wang

NeurIPS 2024 CCF-A
GSLB: The Graph Structure Learning Benchmark

Zhixun Li, Liang Wang, Xin Sun, Yifan Luo, Yanqiao Zhu, Dingshuo Chen, Yingtao Luo, Xiangxin Zhou, Qiang Liu, Shu Wu, Liang Wang, Jeffrey Xu Yu

NeurIPS 2023 CCF-A

Code

Experience

Ant Group
Nov 2024 – Present

Machine Intelligence Division — Hangzhou, China

Joined as a research intern in Nov 2024 and became an Ant Star Intern in May 2026. Led core RAG and agent research: honest alignment for RAG (ACL 2025), multimodal adaptive RAG (IJCNN 2026), test-time adaptation for RAG (ICASSP 2026), agentic RL for knowledge base QA (ICML 2026; EMNLP 2026 Findings), and search-agent credit assignment and refusal evaluation (AAAI 2027 under review).
ByteDance
Feb 2022 – Aug 2022

Machine Learning Engineer Intern, AML — Beijing, China

Developed an AutoML platform for multivariate time series forecasting.

Education

University of Science and Technology of China (USTC)
2022 – 2027 (Expected)

Ph.D. in Control Science and Engineering (Joint with CASIA)
- National Laboratory of Pattern Recognition (NLPR), CASIA
- Advisors: Prof. Liang Wang (IEEE Fellow), Assoc. Prof. Qiang Liu
- National Scholarship for Graduate Students (2024)
- First-class Academic Scholarship (2022–2025, 3 consecutive years)
Shanghai Jiao Tong University (SJTU)
2018 – 2022

B.Eng. in Computer Science and Technology
- National Encouragement Scholarship (2019–2022)
- SJTU Academic Scholarship

Academic Service

Reviewer: ICML 2026, NeurIPS 2025, ACL 2025, EMNLP 2025, KDD 2024

Xin Sun (孙鑫)

News

Publications

Experience

Ant Group

ByteDance

Education

University of Science and Technology of China (USTC)

Shanghai Jiao Tong University (SJTU)

Academic Service