Search papers, labs, and topics across Lattice.
Edinburgh Napier University
1
0
3
LLMs can be made significantly safer at inference time, without any retraining, by using RL to dynamically optimize prompts based on dialogue context.
