Search papers, labs, and topics across Lattice.
Affiliation:
2
0
4
Even after targeted fine-tuning, LLMs peak at an F1 score of just 0.58 on cyber threat level determination, revealing that current models remain far too brittle for operational SecOps workflows.
Multilingual LLMs exhibit a consistent and concerning language bias when resolving conflicting information, favoring certain languages (like Chinese) while disfavoring others (like Russian), even when the information content is equivalent.
