activation steering in vision-language models

Latest papers

8 papers4 papers in the last four weeks, against 2 the four weeks before. 0.0% of all new papers.

Nov 8, 2026

Viscot-steer: Disentangling and Transferring Visual Chain-of-Thought via Activation Steering

Visual chain-of-thought (visual-CoT), also known as "thinking with images", equips LVLMs with the ability to locate and use key visual evidence for reasoning. However, acquiring such visual-CoT capabilities requires costly post-training and tightly couples them to a specific backbone, making transfer to other LVLMs difficult. Existing cross-model transfer methods typically treat visual-CoT as a monolithic capability, using a unified representation that ignores the functional distinction between grounding and reasoning and introduces stage-irrelevant interventions. To address this problem, we propose Viscot-steer, a training-free activation steering framework for transferring visual-CoT capabilities across heterogeneous LVLMs. Specifically, Viscot-steer first disentangles visual-CoT into grounding and reasoning vectors and extracts them from the source-model layers associated with the corresponding capabilities. It then aligns the vectors with the activation space of the target model and steers the grounding and reasoning layers during bounding-box prediction and reasoning-trace generation, respectively. Across benchmarks covering high-resolution visual understanding, general visual perception and reasoning, and mathematical visual reasoning, Viscot-steer transfers capabilities from different visual-CoT models to architecturally diverse LVLMs and outperforms the strongest capability-transfer baseline by an average of 1.86 percentage points on the primary evaluation metrics.
Nov 3, 2026

Entropy-Gated Dynamic Rectification for Object Hallucination in Large Vision-Language Models

Large Vision-Language Models (LVLMs) routinely name objects that are absent from the image. Latent steering can reduce this failure in one model forward pass, but a fixed gate treats early and late decoding steps by the same criterion. We observe that predictive entropy varies sharply within free-running generation, while states that commit to short answers occupy a narrow, confident band. We exploit this temporal mismatch by using uncertainty as a causally available, token-local control signal for an existing steering gate. We propose Entropy-Gated Dynamic Rectification (EGDR), which lowers the intervention threshold when entropy derived from the preceding step's already available logits is high and retains the calibrated operating point when it is low. On LLaVA-1.5-7B, the underlying static gate lowers CHAIR CS/CI from 53.00/15.30 to 32.00/14.30; replacing only its constant threshold with EGDR further lowers them to 29.00/13.60. At a matched 58% intervention rate, the corresponding static control obtains 30.50/14.00. The gain over the static gate holds on LLaVA-NeXT-7B and grows with the generation budget, while POPE scores and general multimodal capability remain effectively unchanged. EGDR requires no additional model forward pass and adds roughly 2% measured latency. Our code is available at https://github.com/alvi-uiu/EGDR.
Oct 27, 2026

When Visual Evidence Is Insufficient: How Instructions, Post-Training, and Internal Activations Shape Refusal in Multimodal Language Models

For reliable deployment, multimodal large language models (MLLMs) should recognize when an image provides insufficient evidence to answer a question and withhold unsupported answers. Yet refusal alone is not necessarily evidence of such recognition: models may refuse because of unrelated safety constraints or without identifying the missing evidence. We study how refusal on visually unanswerable questions is shaped by three sources of control: inference-time instructions, post-training, and internal activation interventions. Across three MLLMs on the MoHoBench benchmark, we systematically manipulate the meaning and placement of refusal instructions, compare supervised and preference-based post-training, and intervene on a refusal-related activation direction derived from naturally answered and refused examples. We find that refusal is highly sensitive to both the meaning and placement of instructions, with model-specific changes at the sample level that are obscured by aggregate refusal rates. Evidence-aware and general refusal instructions have stronger effects than an intentionally mismatched safety instruction. Post-training produces larger increases in refusal, but can also induce responses that combine refusal language with attempted answers. In contrast, activation intervention produces smaller yet transferable changes across prompt conditions, without evidence of indiscriminate refusal on tested visually answerable samples. These findings show that refusal is jointly shaped by external instructions, post-training, and internal representations, highlighting the need to disentangle evidence recognition from a model's learned propensity to refuse.
Oct 23, 2026

Depth-Dependent Safety Geometry in Vision-Language Models: Layer-Wise Localization and Bidirectional Refusal Control

Vision-Language Models (VLMs) are widely deployed, yet the internal mechanisms behind their safety behavior remain poorly understood. We present a lightweight inference-time framework for localizing and manipulating safety-sensitive representations in frozen VLMs, without retraining or modifying parameters. On Qwen3-VL-4B, layer-wise activation divergence analysis reveals non-uniform safety geometry across depth. Causal interventions show that early vision layers support strong bidirectional refusal control through activation steering along a learned safety direction, with unsafe rejection spanning 6%-96%, whereas later layers respond more to activation magnitude scaling and exhibit a substantial safety-selectivity trade-off. Deeper layers remain well separated along the safety direction yet are largely insensitive to steering, distinguishing representational separability from causal controllability. InternVL2.5-4B shows qualitatively similar depth-dependent activation geometry despite differences in layer location and magnitude. Our results expose a white-box, representation-level refusal-bypass attack surface in VLM vision encoders and motivate inference-time stabilization of safety-sensitive representations.
Oct 8, 2026

FearCaut-Qwen: Affective Steering in a Vision-Language Model Shifts the Decision Criterion for Hazard Assessment

Vision-language models (VLMs) show great potential for damage assessment after a disaster, but a recurring deficiency is that they are reluctant to declare a hazard; that is, recall is low even when overall accuracy appears adequate. This study examines that deficiency by using signal detection theory to decompose the decision behavior into perceptual capability and decision-criterion placement. We then propose a novel method for correcting the over-conservative decision policy, inspired by the finding that fear makes humans risk-averse, and ask whether an affective representation associated with fear can be causally manipulated to similarly alter a VLM's decision tendency. Using mechanistic interpretability, we localize a causally implicated affective circuit in the model and use activation steering to manipulate it while observing the effect on downstream prediction. The method is tested on a two-stage SeisMLLM pipeline built on Qwen2.5-VL-7B-Instruct, which flags only 27.0% of genuinely unsafe buildings on the SeisMLLM-1K test split and never issues a false Red, an SDT criterion of c = +1.354, despite adequate evidence quality (d' = 1.521). An affective direction is localized on emotion-rich natural scenes, causally validated by sparse-neuron knockout and distributed steering on held-out emotion data, and then injected into the building task. Fear-direction injection raises Red recall to 75.7% (p<0.001), and subtracting the same direction suppresses Red predictions entirely, whereas norm-matched random and matched happiness directions show no significant effect. The mechanism is a shift in criterion (c=-1.515) while discrimination is not improved (d'=-0.493). These results show that VLM decisions can be adjusted at inference time without retraining and demonstrate how mechanistic interpretability can be used to diagnose and control VLM decision behaviors in engineering applications.
Oct 5, 2026

ROT: Rotating Hidden States towards Contextual Vectors for Hallucination Mitigation in LVLMs

Large Vision-Language Models (LVLMs) frequently suffer from object hallucination. Existing training-free interventions primarily manipulate attention weights, which indirectly affect the deep semantics reaching the final predictive layers. In this work, we shift our focus to the hidden state vectors extracted after self-attention and residual addition. Empirical analysis reveals that hallucinated tokens do not simply over-rely on linguistic priors; instead, they exhibit an anomalous contextual deviation, showing significantly lower similarities to both textual and visual contexts in intermediate layers. Motivated by this, we propose ROT, a layer-specific, training-free framework. ROT dynamically detects semantic deviation in the middle layers and applies a norm-preserving rotation to steer the hidden states back toward the local multimodal context plane spanned by the contexts. For subsequent layers, a representational smoothing mechanism is introduced to stabilize the calibrated trajectory. Extensive experiments on multiple benchmarks demonstrate that ROT consistently reduces hallucinations across various model architectures and scales, offering an efficient, geometry-driven solution for grounded generation.
Apr 9, 2026

Mitigating Entangled Steering in Large Vision-Language Models for Hallucination Reduction

Large Vision-Language Models (LVLMs) have achieved remarkable success across cross-modal tasks but remain hindered by hallucinations, producing textual outputs inconsistent with visual content. Existing methods mitigate hallucinations but often alter generation behavior, resulting in shorter outputs and shifted token distributions, especially in latent space steering approaches. We identify that this issue stems from entangled steering signals, where suppressing hallucinations inadvertently disrupts the model's intrinsic generation behavior. To address this, we propose MESA, an effective plug-and-play framework that performs controlled and selective latent intervention for hallucination mitigation. Specifically, MESA targets hallucination-relevant responses while preserving the model's original token distribution, enabling effective hallucination reduction without compromising generation behavior. Extensive experiments across diverse generative and discriminative benchmarks demonstrate that MESA consistently reduces hallucinations while better preserving generation behavior, outperforming prior methods across multiple LVLM families.
Feb 25, 2026

Dynamic Multimodal Activation Steering for Hallucination Mitigation in Large Vision-Language Models

Large Vision-Language Models (LVLMs) exhibit outstanding performance on vision-language tasks but struggle with hallucination problems. Through in-depth analysis of LVLM activation patterns, we reveal two key findings: 1) truthfulness and visual perception capabilities predominantly engage different subsets of attention heads within the model architecture; and 2) truthfulness steering vectors vary significantly across different semantic contexts. Based on these observations, we propose Dynamic Multimodal Activation Steering, a training-free approach for hallucination mitigation. Our method constructs a semantic-based truthfulness steering vector database and computes visual perception steering vectors, enabling context-aware interventions during inference by dynamically selecting the most relevant steering vectors based on input semantic similarity and applying them to the most influential attention heads. We conduct comprehensive experiments across multiple models and datasets, demonstrating that our approach significantly enhances model performance, outperforming existing state-of-the-art methods.