Search papers, labs, and topics across Lattice.
1
5
3
48
Using preference data from stronger models to align LLMs via DPO can backfire, dramatically worsening safety by making models more susceptible to jailbreaking.
