Search papers, labs, and topics across Lattice.
Microsoft, Beijing, China
1
16
3
Achieving up to 3.2x speedup in recommender system inference across multiple hardware platforms without manual tuning could revolutionize deployment strategies in heterogeneous environments.
