TLDR
激活引导系数设定干预强度，但请求特定程度的人设表达需要一个行为量表。我们研究人设剂量：通过特性描述和请求的平均强度来控制语言模型。PersonaDose 在人设响应上特化一个共享的、由描述条件化的 FLAS 控制器，然后根据测得的特性表达对其流时间进行校准。训练响应未与请求的目标强度配对。在 Llama-3.1-8B、Qwen3-8B 和 Gemma-3-4B 上，PersonaDose 提升了 Persona Vect 上的核心特性表达……An activation-steering coefficient sets intervention strength, but requesting a particular degree of persona expression requires a behavioral scale. We study persona dosing: controlling a language model through a trait description and a requested mean intensity. PersonaDose specializes a shared, description-conditioned FLAS controller on persona responses, then calibrates its flow time against measured trait expression. Training responses are not paired with requested target intensities. Across Llama-3.1-8B, Qwen3-8B, and Gemma-3-4B, PersonaDose raises core-trait expression at the Persona Vect