Search papers, labs, and topics across Lattice.
4
0
9
0
LALMs are shockingly vulnerable to inaudible audio prompts that can make them execute unauthorized actions, even on commercial systems like Mistral AI and Microsoft Azure.
Fuzzy Private Set Intersection just got a whole lot faster: new protocols slash runtime by up to 145x and communication costs by 19x compared to existing methods.
Automatically exploiting web application vulnerabilities is now significantly more feasible, with AutoEG achieving over 80% success where previous methods struggled to reach 33%.
Training speculative decoding models just got an order of magnitude faster, unlocking real-world deployment with a new open-source framework and a suite of production-ready draft models.
