Search papers, labs, and topics across Lattice.
1
0
3
0
LLMs can achieve state-of-the-art coreference resolution in task-based dialogue by reasoning over object metadata at test time, even outperforming supervised methods in cross-domain generalization.
