Search papers, labs, and topics across Lattice.
Zhejiang University
1
0
3
7
LLMs can overcome "tunnel vision" in multi-turn search scenarios by using information gain to guide dynamic prompting interventions, leading to more efficient and accurate reasoning.
