Engineering ❯ Software Engineering ❯ Algorithm Design ❯ Reinforcement Learning
Researchers say targeted prompting and fine-tuning reduce harmful compliance in controlled tests.