Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
1
0
3
4
Interleaved stacking preserves critical layer-specific knowledge in speech model distillation, leading to faster training without sacrificing performance.