Search papers, labs, and topics across Lattice.
1
0
3
5
Current adversarial evaluation practices are misleading: stronger LLMs don't necessarily produce more effective adversarial claims when you actually check if the facts are still true.
