Search papers, labs, and topics across Lattice.
University of Science and Technology of China
1
0
2
Hallucination rates in MLLMs can be slashed by up to 60% through targeted interventions in attention dynamics, revealing a critical pathway to improving model reliability.