Search papers, labs, and topics across Lattice.
1
0
3
2
LLMs can be efficiently aligned with human preferences using only a small, carefully selected subset of training data, without sacrificing generalization ability.
