Beyond Linear Activation Steering: Invertible Latent Transformations for Controlling LLM Behavior

• URL: http://arxiv.org/abs/2606.08454v1
• Date: Sun, 07 Jun 2026 05:01:25 GMT
• Title: Beyond Linear Activation Steering: Invertible Latent Transformations for Controlling LLM Behavior
• Abstract summary: INNSteer is a nonlinear activation steering framework based on invertible latent transformations.<n>It consistently improves model control over linear, transport-based, and nonlinear steering baselines.
• Score: 13.073472989807675
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Activation steering provides a lightweight inference-time mechanism for controlling large language models (LLMs) by modifying their internal activation vectors toward desired behaviors. Most existing methods compute a fixed steering direction in the original activation space, typically from pairs of contrastive examples using mean differences, linear probes, or arbitrary separability criteria. While effective to a certain extent, these methods treat behavioral control as a global, linear, additive offset: the same direction is applied across inputs, and behaviors are linearly separable. This can be restrictive when behavioral features vary nonlinearly across the activation space or lie on curved and anisotropic manifolds, where the optimal intervention may be input-dependent. To address this limitation, we propose INNSteer, a nonlinear activation steering framework based on invertible latent transformations. Rather than searching for a better steering vector in the original representation space, INNSteer learns a lightweight invertible neural network $φ$ that maps an LLM's activations into a latent space where behavioral classes are more amenable to linear control. At inference time, activations are mapped through $φ$, steered in the latent space, and mapped back through the exact inverse transformation $φ^{-1}$. This makes a simple latent-space translation become a nonlinear, input-dependent intervention in the original activation space. Across experiment settings on multiple LLM families, scales, behavioral traits, and safety benchmarks, INNSteer consistently improves model control over linear, transport-based, and nonlinear steering baselines while largely preserving generation fluency.

Related papers

• Local Linearity of LLMs Enables Activation Steering via Model-Based Linear Optimal Control [2.752817022620644]
  Inference-time LLM alignment methods offer an alternative to fine-tuning by directly modifying activations during generation.<n>We show that layer-wise dynamics across multiple LLM architectures and scales are well-approximated by locally-linear models.<n>We derive theoretical bounds on setpoint tracking error, enabling formal guarantees on steering performance.
  arXiv  Detail & Related papers  (2026-04-21T03:09:46Z)
• Where to Steer: Input-Dependent Layer Selection for Steering Improves LLM Alignment [17.001194019333052]
  We show that different inputs can require steering at different layers to achieve alignment with a desirable model behavior.<n>We introduce Where to Steer, a framework that adaptively selects the intervention layer conditioned on the input, by learning a mapping from input embeddings to optimal steering layers.
  arXiv  Detail & Related papers  (2026-04-04T21:16:47Z)
• Curveball Steering: The Right Direction To Steer Isn't Always Linear [7.264147948470562]
  Activation steering is a widely used approach for controlling large language model (LLM) behavior by intervening on internal representations.<n>We propose "Curveball steering", a nonlinear steering method based on PCA kernel that performs interventions in a feature space.
  arXiv  Detail & Related papers  (2026-03-10T07:45:35Z)
• FreeAct: Freeing Activations for LLM Quantization [89.97086263978058]
  Quantization is pivotal for mitigating the significant memory and computational overhead of Large Language Models.<n>FreeAct is a novel quantization framework that relaxes the static one-to-one constraint to accommodate dynamic activation disparities.<n>Experiments across dLLMs and MLLMs demonstrate that FreeAct significantly outperforms baselines, up to 5.3% performance improvement.
  arXiv  Detail & Related papers  (2026-03-02T12:02:17Z)
• ODESteer: A Unified ODE-Based Steering Framework for LLM Alignment [49.68063561145927]
  We propose a unified ordinary differential equations (ODEs)-based theoretical framework for activation steering.<n>We introduce ODESteer, a kind of ODE-based steering guided by barrier functions.<n>Compared to state-of-the-art activation steering methods, ODESteer achieves consistent empirical improvements.
  arXiv  Detail & Related papers  (2026-02-19T17:13:44Z)
• MirrorLA: Reflecting Feature Map for Vision Linear Attention [49.41670925034762]
  Linear attention significantly reduces the computational complexity of Transformers from quadratic to linear, yet it consistently lags behind softmax-based attention in performance.<n>We propose MirrorLA, a geometric framework that substitutes passive truncation with active reorientation.<n>MirrorLA achieves state-of-the-art performance across standard benchmarks, demonstrating that strictly linear efficiency can be achieved without compromising representational fidelity.
  arXiv  Detail & Related papers  (2026-02-04T09:14:09Z)
• GrAInS: Gradient-based Attribution for Inference-Time Steering of LLMs and VLMs [56.93583799109029]
  GrAInS is an inference-time steering approach that operates across both language-only and vision-language models and tasks.<n>During inference, GrAInS hidden activations at transformer layers guided by token-level attribution signals, and normalizes activations to preserve representational scale.<n>It consistently outperforms both fine-tuning and existing steering baselines.
  arXiv  Detail & Related papers  (2025-07-24T02:34:13Z)
• Guiding Giants: Lightweight Controllers for Weighted Activation Steering in LLMs [8.085475675888045]
  Activation steering provides an alternative for inference-time control.<n>We introduce a novel approach using a lightweight, trainable controller network integrated during inference.
  arXiv  Detail & Related papers  (2025-05-22T01:48:38Z)
• Exploring Linear Feature Disentanglement For Neural Networks [63.20827189693117]
  Non-linear activation functions, e.g., Sigmoid, ReLU, and Tanh, have achieved great success in neural networks (NNs) Due to the complex non-linear characteristic of samples, the objective of those activation functions is to project samples from their original feature space to a linear separable feature space. This phenomenon ignites our interest in exploring whether all features need to be transformed by all non-linear functions in current typical NNs.
  arXiv  Detail & Related papers  (2022-03-22T13:09:17Z)
• Orthogonal Jacobian Regularization for Unsupervised Disentanglement in Image Generation [64.92152574895111]
  We propose a simple Orthogonal Jacobian Regularization (OroJaR) to encourage deep generative model to learn disentangled representations. Our method is effective in disentangled and controllable image generation, and performs favorably against the state-of-the-art methods.
  arXiv  Detail & Related papers  (2021-08-17T15:01:46Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.