Steering Large Language Models using Conceptors: Improving Addition-Based Activation Engineering

Open in new window