Xuan Wang

The paper investigates generalized Wall-Sun-Sun primes, $WSS(d)$, and uses them to study the weight distributions of specific cyclic codes defined over $ ext{F}_p$ and $ ext{Z}_{p^2}$.

The paper introduces NICE, a novel, theory-grounded diagnostic benchmark for assessing the social intelligence of LLMs, which reveals that current frontier models consistently struggle with specific facets of communication.

The paper introduces a structured multi-level visual perturbation framework to systematically analyze how dependent VLA-based driving behavior is on visual information, revealing uneven visual grounding across different abstraction levels.

The paper introduces SURE, a unified framework designed to standardize and improve the comparability and reproducibility of evaluations for advanced speech understanding models.

The paper introduces GenPT, a Generative Projective Testing framework, which demonstrates superior reliability and resistance to social-desirability bias compared to traditional self-report questionnaires when assessing LLM psychological states.

The paper introduces an A*-inspired framework to generate highly effective and efficient adversarial prompts that cause LLMs to hallucinate commonsense errors while maintaining the original prompt's intent.

MOSS-Audio is a unified audio-language model designed for comprehensive understanding of speech, environmental sounds, and music, achieving strong performance across various audio-grounded tasks.

This paper introduces a novel full-space quantization-driven architecture (FQA) to create highly efficient and accurate hardware approximations of nonlinear activation functions using piecewise polynomial approximations (PPAs).