Search papers, labs, and topics across Lattice.
3
0
5
1
Current LLMs can autonomously penetrate systems with success rates up to 69.3%, revealing alarming implications for cybersecurity.
Achieving competitive predictive performance with significantly fewer parameters, TI-Adapter redefines efficiency in multimodal learning.
Prevent LLM agents from leaking data or causing financial loss with AgentGuard, a new access control framework that requires only ~10 lines of code to integrate.
