Search papers, labs, and topics across Lattice.
Affiliation:
1
0
3
Safety evaluations of LLMs vary dramatically across languages, revealing that some harmful content is far more susceptible to manipulation than previously understood.
