1 paper
Enjun Du, Hange Zhou, Chenxu Du +4
The paper introduces LedgerMind, a framework that records and constrains the evidence used by multimodal agents during visual question answering, ensuring that each reasoning step…