Search papers, labs, and topics across Lattice.
1
0
3
Constraining LLMs' vocabulary ("No-Have" or "E-Prime") can boost ethical reasoning by 19%, and ensembles of these constrained agents can solve debugging problems that standard models miss.