Search papers, labs, and topics across Lattice.
Affiliation:
2
0
2
5
SALA achieves superior reasoning alignment by learning flexible, task-specific operations, outperforming traditional methods that rely on rigid matching criteria.
CoMerge not only preserves task-specific capabilities but also achieves near-optimal performance with just 1,445 scalar coefficients, challenging the need for full-parameter fine-tuning.
