Search papers, labs, and topics across Lattice.
1
0
2
Semantic similarity is a poor proxy for agent performance: ranking agents based on execution-aware probing beats description-based retrieval by a wide margin.
