Steering at the Source: Style Modulation Heads for Robust Persona Control
arXiv:2603.13249v1 Announce Type: new Abstract: Activation steering offers a computationally efficient mechanism for controlling Large Language Models (LLMs) without fine-tuning. While effectively controlling target traits …
Yoshihiro Izawa, Gouki Minegishi, Koshi Eguchi, Sosuke Hosokawa, Kenjiro Taura
16 views