Search papers, labs, and topics across Lattice.
1
0
3
8
State-of-the-art LLM agents face a staggering performance decline in multilingual workflows, revealing critical gaps in current evaluation methods.
