Search papers, labs, and topics across Lattice.
Bilibili Inc.
1
0
3
Achieve DLRM-level performance in recommender systems with up to 26x fewer parameters and FLOPs by compressing feature crosses hierarchically.
