Zhangchen Xu
University of Washington PhD student known for Magpie, a self-synthesis method for instruction-tuning data, and for LLM safety/jailbreak research.
- Role
- grad-student
- Currently at
- University of Washington
- twitter.com/ZhangchenXu
- Scholar
- scholar.google.com/citations
- Papers
- 11
Cite
Notes
Only stored in your browser.
Authored papers
11Visual Aesthetic Benchmark: Can Frontier Models Judge Beauty?
arXiv 2026
Emergent Social Intelligence Risks in Generative Multi-Agent Systems
arXiv 2026
KodCode: A Diverse, Challenging, and Verifiable Synthetic Dataset for Coding
arXiv 2025
TinyV: Reducing False Negatives in Verification Improves RL for LLM Reasoning
arXiv 2025
TOUCAN: Synthesizing 1.5M Tool-Agentic Data from Real-World MCP Environments
arXiv 2025
Building a Foundational Guardrail for General Agentic Systems via Synthetic Data
arXiv 2025
SafeDecoding: Defending against Jailbreak Attacks via Safety-Aware Decoding
arXiv 2024
CleanGen: Mitigating Backdoor Attacks for Generation Tasks in Large Language Models
arXiv 2024
Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing
preprint
ArtPrompt: ASCII Art-based Jailbreak Attacks against Aligned LLMs
arXiv 2024
ChatBug: A Common Vulnerability of Aligned LLMs Induced by Chat Templates
arXiv 2024
Tool contributions
1Affiliations
Frequent co-authors
10from 11 papers
Radha Poovendran
professor
Fengqing Jiang
grad-student
Luyao Niu
researcher
Bhaskar Ramasubramanian
Bill Yuchen Lin
researcher
Xiangliang Zhang
Yue Huang
Yuetai Li
Hang Hua
Pin-Yu Chen