GAZE:Governance-Aware pre-annotation for Zero-shot World Model Environments

Leela Krishna, Mengyang Zhao, Saicharithreddy Pasula +2

Training robust world models requires large-scale, precisely labeled multimodal datasets, a process historically bottlenecked by slow and expensive manual annotation. We present a…