Search papers, labs, and topics across Lattice.
1
0
2
BEST-RQ-2 achieves superior audio representation transfer by decoupling contextualization from prediction, challenging the efficiency of traditional one-stage models.