无需线性假设，用单个模型实现多属性灵活控制。
Beyond Linear Steering: Unified Multi-Attribute Control for Language Models
- 训练非线性分类器，通过梯度动态计算干预方向。
- 在3个模型族上均优于基线，实现多属性精准控制。
- 适合需要灵活组合行为的场景，无需重调参。
在推理阶段控制大语言模型的多个行为属性是一项挑战，源于属性间的干扰以及线性引导方法的局限——后者假设激活空间中行为可加，并需为每项属性单独调参。我们提出K-Steering，一种统一且灵活的方法：在隐藏激活上训练单一非线性多标签分类器，并在推理时通过梯度计算干预方向。该方法避免了线性假设，无需存储或调参独立属性向量，支持无需重训练的行为动态组合。为评估该方法，我们设计两个新基准ToneBank和DebateMix，聚焦组合式行为控制。在3个模型族上的实证结果，经基于激活的分类器和大语言模型判官双重验证，表明K-Steering在准确引导多个行为方面显著优于强基线。
原文摘要 · Abstract (English)
Controlling multiple behavioral attributes in large language models (LLMs) at inference time is a challenging problem due to interference between attributes and the limitations of linear steering methods, which assume additive behavior in activation space and require per-attribute tuning. We introduce K-Steering, a unified and flexible approach that trains a single non-linear multi-label classifier on hidden activations and computes intervention directions via gradients at inference time. This avoids linearity assumptions, removes the need for storing and tuning separate attribute vectors, and allows dynamic composition of behaviors without retraining. To evaluate our method, we propose two new benchmarks, ToneBank and DebateMix, targeting compositional behavioral control. Empirical results across 3 model families, validated by both activation-based classifiers and LLM-based judges, demonstrate that K-Steering outperforms strong baselines in accurately steering multiple behaviors.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。