Haoxiang Sun

V-Zero achieves fine-grained visual reasoning without any annotated answer labels, outperforming traditional methods in both speed and accuracy.

Unified vision-language perception in MLLMs is not just an evolution; it’s a critical leap toward achieving artificial general intelligence.