Sanxing

Chen

I am a final-year PhD candidate at Duke University, advised by Bhuwan Dhingra and previously Sam Wiseman. I have also worked at Google Research, Google DeepMind, and Microsoft Research.

My research focuses on learning and exploration in language agents. I am interested in how agents can improve through interaction, building on my earlier work on factuality and knowledge representation.

Learning and exploration. I study how training shapes agents' exploration strategies, using multi-turn interactive tasks to evaluate their behavior. My recent work at Google Research focuses on RL post-training for MLE agents. At Google DeepMind, I worked on continual learning for mobile UI agents.
When Greedy Wins (ICLR 2026) · ChatShop (preprint)
Search agents and factuality. I develop reliable automatic evaluation methods for search agents, with a focus on strategy and factuality.
Real-time Factuality (ACL 2025) · Situated Faithfulness (ICLR 2025, Spotlight) · SearchAtlas (EMNLP 2026 Findings) · Lazy Grounding (EMNLP 2026)
Knowledge representation. I pioneered the adoption of Transformers for knowledge graph representation.
HittER (EMNLP 2021, Oral; 200+ citations) · Pre-training Transformers for KG Completion (preprint)

I'm on the industry job market, seeking RS/MTS roles in agent evaluation, post-training, and self-improvement. CV · Email

Publications

More on Google Scholar.

Planetary Prediction Engine: Autonomous Geospatial Prediction via Intelligent Data Selection and Foundation Model Embeddings

Evelyn Ma, Rama Kumar Pasumarthi, Kishwar Shafin, Mandar Sharma, Mimi Sun, Hamed Sadeghi, Dav M. Ebengo, Mbulayi Onesime, Rouslan Solomakhin, John Wamburu, William Ogallo, Aisha Walcott-Bryant, Sanxing Chen, Arbaaz Muslim, Yael Mayer, Ronald Ho, Roy Lee, Ruth Alcantara, Abdoulaye Diack, Monica Bharel, Lambert Rosique, Jeremy Amez-Droz, Christopher Haire, James Manyika, Yossi Matias, Niv Efron, Gautam Prasad, Shravya Shetty

Preprint, 2026

AbsBibarXiv Blog

🕹️ When Greedy Wins: Emergent Exploitation Bias in Meta-Bandit LLM Training

Sanxing Chen, Xiaoyin Chen, Yukun Huang, Roy Xie, Bhuwan Dhingra

In ICLR'2026

AbsBibarXiv Code

Cite Pretrain: Retrieval-Free Knowledge Attribution for Large Language Models

Yukun Huang, Sanxing Chen, Jian Pei, Manzil Zaheer, Bhuwan Dhingra

In ICLR'2026

AbsBibarXiv Code

📰 Real-time Factuality Assessment from Adversarial Feedback

Sanxing Chen, Yukun Huang, Bhuwan Dhingra

In ACL'2025 (Main)

AbsBibarXiv Code

Enhancing Large Language Models' Situated Faithfulness to External Contexts

Yukun Huang, Sanxing Chen, Hongyi Cai, Bhuwan Dhingra

In ICLR'2025 (Spotlight)

AbsBibarXiv Code

🛍️ ChatShop: Interactive Information Seeking with Language Agents

Sanxing Chen, Sam Wiseman, Bhuwan Dhingra

Preprint, 2024

AbsBibarXiv Code

Pre-training Transformers for Knowledge Graph Completion

Sanxing Chen, Hao Cheng, Xiaodong Liu, Jian Jiao, Yangfeng Ji, Jianfeng Gao

Preprint, 2023

AbsBibarXiv Code

⚾️ HittER: Hierarchical Transformers for Knowledge Graph Embeddings

Sanxing Chen, Xiaodong Liu, Jianfeng Gao, Jian Jiao, Ruofei Zhang, Yangfeng Ji

In EMNLP'2021 (Oral)

AbsBibLink arXiv Code

A Tale of Two Linkings: Dynamically Gating between Schema Linking and Structural Linking for Text-to-SQL Parsing

Sanxing Chen, Aidan San, Xiaodong Liu, Yangfeng Ji

In COLING'2020 (Oral)

AbsBibLink arXiv Code

Service

Reviewer: COLING'2020, NAACL'2021, ACL'2021, EMNLP'2021, NLPCC'2021, CoLM'2024, COLING'2024, ARR (regularly)
Volunteer: ACL'2020, EMNLP'2020

Misc

I am originally from Quanzhou, China. It was one of the world's largest and most cosmopolitan seaports and now a World Heritage Site. (Interestingly, UVA is also a World Heritage Site.)
I started playing the violin when I was seven.
I enjoy photography and traveling.