Search papers, labs, and topics across Lattice.
IPAI, Seoul National University
1
2
Current function-calling benchmarks are too simple: DICE-BENCH reveals that LLMs still fail at realistic, multi-turn tool use.
