Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Huawei Cloud BU
3
0
4
5
Targeted data augmentation and joint RL/OPD training unlock significantly stronger long-horizon multimodal search capabilities for agents.
Visual evidence can now actively drive long-horizon multimodal search, transforming how agents interact with complex information.
Current multimodal agents are surprisingly bad at web browsing, achieving only 36% accuracy on a new benchmark designed to test deep, multi-modal reasoning across web pages.