Search papers, labs, and topics across Lattice.
Huazhong University of Science and Technology
1
12
3
0
System-aware optimizations in KV cache design could dramatically reduce the memory footprint and costs of serving large language models.
