Search papers, labs, and topics across Lattice.
Peking University
1
0
2
Cold starts for LLMs can be accelerated by over 7 times without sacrificing correctness, thanks to a novel framework that enables safe concurrent execution.