Search papers, labs, and topics across Lattice.
2
0
5
0
LLMs exhibit significant geographical performance disparities and task-specific gaps when evaluated on the new GaoYao benchmark, highlighting the need for more nuanced multilingual and multicultural training.
LLM agent performance hinges on maximizing decision-relevant information density within context, not just context length, and GenericAgent proves it.
