Search papers, labs, and topics across Lattice.
1
0
3
LLMs struggle to understand nuanced values across languages, with accuracy dropping below 77% and varying by over 20% between languages, as revealed by the new X-Value benchmark.
