Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
1
0
2
0
ToolFailBench uncovers that even top-performing LLMs struggle with tool use, revealing a staggering 89-point accuracy gap between models on control tasks.