Track new results across pure and applied mathematics.
Search papers, labs, and topics across Lattice.
Affiliation:
1
0
0
0
Synthesized audio-visual datasets now retain strong cross-modal semantic correspondence, boosting downstream model accuracy by over 15 percentage points and slashing distillation costs.