Search papers, labs, and topics across Lattice.
1
0
0
1
MLLMs can be made significantly safer in multi-turn dialogues with a new framework that combines cold-start refusal and turn-aware policy optimization, achieving a 10% drop in attack success rate.
