Search papers, labs, and topics across Lattice.
1
0
3
3
LLM agents in high-stakes domains can be verified more reliably by accumulating evidence grounded in expert guidelines, achieving a 12% AUROC improvement and 50% Brier score reduction over existing methods.
