NUSRI-CQedit

NUSRI-CQ refers to the NUS (Chongqing) Research Institute, where Xinbao Qiao worked as a full-time research intern from June to December 2025. This internship followed the Zhejiang University master's phase and preceded the CUHK doctoral phase.

Programedit

Qiao's internship work focused on trustworthy LLM systems and synthetic-data evaluation. NUSRI-CQ presents itself as a platform for research, technology transfer, education, and innovation in Chongqing.¹

Academic contextedit

Public NUS material describes NUSRI-CQ as a joint effort between the National University of Singapore and Chongqing Liangjiang New Area, while the institute's own English site frames it as NUS's cooperation window in western China.²

Connection to Qiaoedit

The NUSRI-CQ period involved research on synthetic data, collaborative evaluation, Wasserstein geometry, and LLM reliability. Related work includes the ICML 2026 paper When Sample Selection Bias Precipitates Model Collapse, which examines how distributed participants with limited resources can diagnose data-generation failures without exposing all local data. The NeurIPS 2026 paper Illusory Pattern Perception Drives Spurious Inference in Large Language Models compares how LLMs and humans infer relations from random, unrelated, or insufficient evidence.

See alsoedit

• Xinbao Qiao
• Experience
• Synthetic Data and Model Collapse
• When Sample Selection Bias Precipitates Model Collapse
• Illusory Pattern Perception Drives Spurious Inference in Large Language Models
• The Chinese University of Hong Kong

Footnotesedit

1. The NUSRI-CQ English institute introduction describes the institute as a research, incubation, education, and training platform focused on industry-academia-research integration. ↩

2. NUS News reported in September 2018 that NUS signed a strategic framework agreement with Chongqing Liangjiang New Area Administrative Committee to establish NUSRI@CQ. ↩