Huiqi Zou

Contextualized Evaluation of Vision Language Models through Dynamic, Multi-turn Interactions

Yijiang Li, Huiqi Zou, Bingyang Wang, Ziang Xiao

Multi-modal Large Language Models (MLLMs) have made substantial advances on benchmarks, yet their real-world effectiveness remains uncertain. This gap stems from the fundamental misalignment between benchmarks in controlled, static settings and the dynamic, interactive, and conte…Read more