Search papers, labs, and topics across Lattice.
1
25
3
6
DPO's rise as a computationally efficient alternative to RLHF for LLM alignment has spurred a diverse range of research, now systematically organized and analyzed in this comprehensive survey.
