Search papers, labs, and topics across Lattice.
Affiliation:
1
0
LLMs routinely overstate evidence and prematurely close investigations, but a novel dataset and RL approach enable them to match human expert judgment on evidence sufficiency.
