Search papers, labs, and topics across Lattice.
MBZUAI 2 University of Southern California
2
0
4
Orthogonal obfuscation slashes token recovery rates from LLM inference to just 1.3%, safeguarding sensitive data without compromising performance.
LoRA fine-tuning just got a memory-efficient upgrade: LoRSum matches or beats standard LoRA performance by reformulating optimization as a proximal problem and using diagonal K-FAC approximations, all without expensive SVD.
