Search papers, labs, and topics across Lattice.
1
0
2
Confidence estimates from LLMs can be misleading when evaluating many candidates, but a new framework ensures high-probability agreement with human judgments.
