Search papers, labs, and topics across Lattice.
University College London
1
0
3
Stop chasing moving targets: verifiable model identity, not just safety claims, is the key to robust AI evaluation.