Tag
This paper introduces WLDS, a large-model-driven system for simulating and deducing emergency instances by leveraging controllable randomness and cross-domain knowledge. It presents the Emergency Instances Deduction (EID) benchmark and demonstrates high-fidelity simulation capabilities across multiple domains.