Search papers, labs, and topics across Lattice.
1
0
2
A judge's benchmark accuracy can significantly overstate its true evaluative competence, revealing critical insights for optimizing skill evaluation in AI systems.
