Search papers, labs, and topics across Lattice.
1
0
3
5
Encrypted LLM inference just got a whole lot faster: TIGER accelerates TFHE-based nonlinear layers by up to 17x using GPUs.