Hybrid attention dominates frontier LLMs, yet Vision Transformers (ViTs) in multimodal LLMs lack a satisfactory hybrid design, with no consensus on why certain attention patterns work better. To fill this gap, we study ViT attention heads and find they differentiate into object- and background-specialist roles, a pattern most pronounced under full attention; we call this Semantic Head Specialization (SHS). We propose SHS-Index to quantify this specialization, show that it distinguishes full-attention from chunk-window ViTs, and find that it strongly tracks downstream benchmark performance. We then identify three structural factors that shape SHS---window interaction, token serialization, and local softmax allocation---and use them as design principles for hybrid attention. Guided by these factors, we design Ariadne Attention, a hybrid that matches full attention on 22 image and video tasks at 6.5x less attention compute. Our findings establish head specialization as a measurable property for diagnosing and designing principled hybrid ViT attention at the multimodal-LLM scale.
Vision-language models (VLMs) can locate an image region referred to by a text prompt and route the corresponding visual evidence to the output, yet the internal mechanism behind this behavior is not understood. Inspired by retrieval heads in large language models, we ask whether VLMs contain an analogous mechanism for visual retrieval. We answer affirmatively by introducing Visual Retrieval Heads (VRHs), a small subset of attention heads (about 1.7-2.6%) that are causally responsible for grounding text descriptions to image regions. To find them, we recast existing head-scoring methods under a unified design space over query tokens, key aggregation, and cross-sample aggregation. We then show that scoring attention from output prediction tokens with a sum over the ground-truth referent region most reliably identifies causal heads. Across eleven VLMs and five referring-expression benchmarks, masking only the top 20 VRHs reduces grounding accuracy by up to 80 percentage points, while masking the same number of random heads has little effect. Beyond replicating the causal-sparse-universal triad established for text retrieval heads, VRHs exhibit several properties not previously reported: they generalize across visual reference tasks, remaining causal on attribute, spatial, counting, and visual-math benchmarks despite being discovered through bounding-box prediction; they are functionally specific, preserving output format while corrupting localization; and they are architecturally shared, transferring causally across VLMs that share an LLM backbone but differ in vision encoder, projector, and instruction tuning.
Vision-Language Models (VLMs) have exhibited impressive performance across diverse visual scenarios. However, this success comes at the cost of explosive growth in visual tokens, which imposes substantial memory and computational overhead during inference, ultimately increasing latency. To improve VLM inference efficiency, a typical class of visual token pruning methods estimates token importance by aggregating attention scores across all heads in the pruning layer of the Large Language Model (LLM) backbone and prunes tokens based on aggregated scores. However, in this paper, we reveal a compelling phenomenon: the capability to pinpoint critical visual tokens is concentrated within a small fraction of heads. Aggregation exclusively on these heads can improve task performance. Inspired by this observation, we propose ProViP, a training-free progressive visual token pruning framework. ProViP first removes redundant visual tokens based on the embedding similarity of input tokens before reasoning of the LLM backbone, and then further prunes tokens during reasoning via head-aware pruning. Experiments demonstrate that ProViP delivers outstanding task performance and inference efficiency. For instance, when applied to LLaVA-1.5-7B, ProViP retains 95.9% of the original performance and achieves 1.62x inference speedup under an 88.9% pruning ratio.
Vision-language models such as LLaVA-1.5-7B often hallucinate objects absent from the image when generating captions. We ask whether an interpretability diagnosis of this failure can guide a targeted fix, and we measure what that fix actually changes. We rank attention heads by how much their image attention drops around hallucinated object words, then screen the shortlist by ablating candidate heads and measuring the change in hallucination-token log probability, yielding a 32-head set. We restrict two interventions to these heads: a head-sliced LoRA adapter and an inference-time grounding controller. On 400 held-out COCO images, the combined method lowers CHAIRs (the fraction of captions with a hallucinated object) from 0.370 to 0.230 and CHAIRi (the fraction of hallucinated object mentions) from 0.156 to 0.096 (p < 0.001, paired sign-flip tests). Two controls sharpen attribution. A random-head LoRA control, matched layer-for-layer and trained identically, performs no better than the matched baseline on a separate 200-image control split, supporting the role of head selection rather than LoRA capacity. Under fixed decoding budgets, the CHAIR reduction persists and grows with budget (23% at 64 tokens to 58% at 128), arguing against a pure max-token or truncation artifact, although the method remains shorter and more conservative. The resulting behavior reduces unsupported object mentions while also lowering object recall (0.78 to 0.70). We present a diagnosis-to-intervention pipeline for object hallucination, and, more importantly, a controlled account of what acting on the diagnostic signal actually does: it localizes intervention sites with real, non-random leverage, reported as a behavioral profile rather than a single score.
Despite remarkable progress in vision-language generation, Vision-Language Models (VLMs) remain prone to hallucinations, producing content that is inconsistent with or unsupported by the input image. Existing works largely design detection or mitigation methods around one specific hallucination pattern, such as visual-textual imbalance, but real VLM hallucinations arise from a mixture of multiple patterns, so signals bound to a single pattern struggle to remain stable across models and tasks. Under a unified head-level view, we find that hallucination-induced changes manifest as localized deviations from each head's faithful contextual behavior, a phenomenon we term Role-Break. Detailed analysis reveals that these deviations are systematically organized across attention heads, contextual sources, and deviation directions, and that the resulting signal is linearly readable once head identity is preserved. Based on these findings, we build a lightweight linear detector on top of Role-Break that requires no fine-tuning of the VLM, whose feature dimension stays below 5,000 and reaches an average AUROC of 93.23 across six VLMs and four benchmarks. A small-scale intervention experiment further shows that the detected tokens can be directly acted upon in the discriminative setting.
Davide Caffagni, Alberto Compagnoni, Federico Melis +5cs.CV cs.AI cs.CL cs.MM
Representation alignment has emerged as an effective approach to improve Multimodal Large Language Models (MLLMs) by regularizing their internal representations toward those of an external vision encoder. However, existing methods typically align a fixed layer of the language backbone, overlooking the fine-grained structure of Transformer models. In this work, we propose Head-Wise Representation Alignment (HeRA), a method that enforces cross-modal alignment at the level of individual attention heads. Our approach is grounded in the Platonic Representation Hypothesis, focusing on preserving the topological structure of representations (i.e., their local neighborhood relationships) across modalities. Following the Mutual K-Nearest Neighbor (MKNN) alignment metric, we introduce a contrastive objective that acts as a differentiable proxy for matching local structures. HeRA applies this objective during multimodal training to specific attention heads in the LLM, selected by their alignment score according to the MKNN metric. Counterintuitively, we find that aligning the least aligned heads yields the largest gains. Extensive evaluations across multiple MLLMs and 18 benchmarks demonstrate that HeRA consistently improves performance on challenging vision-centric tasks and serves as an effective regularizer against visual hallucinations by naturally curbing the over-reliance on linguistic priors. Our code is publicly released.
How a vision-language model internally solves the task of describing an image is far from obvious. We find that the model develops a specific mechanism for this: a small set of attention heads in its language-model backbone, which we call gaze heads, whose attention tracks the image region the model is currently describing. We find them with a simple correlation score from a few forward passes, using comic strips as a controlled testbed where narrative order is laid out spatially. These gaze heads do not just track the image tokens being described: redirecting their attention to a chosen region forces the VLM to describe that region instead. A single attention-mask intervention on the top-100 gaze heads, fewer than 9% of all heads, steers the model's answer to any chosen comic panel at 83.1% accuracy, while the same intervention on random heads fails to redirect the answer, and intervening on all heads destroys generation. The same lever also extends to continuous control: switching the gaze target mid-generation makes the model wrap up its current panel description and move to the new one within a few tokens. Beyond comics, the same intervention redirects answers to chosen regions in natural COCO images. The mechanism further recurs across model sizes from 2B to 32B parameters and across other VLM architectures, although some frozen-encoder families show no comparable head set. More broadly, this shows that targeted edits identified through mechanistic analysis can serve as practical inference-time levers for steering multimodal model behavior, without any retraining. Our code, interactive demo, and datasets are available at https://gaze.baulab.info/
While Multimodal Large Language Models (MLLMs) demonstrate remarkable proficiency on complex vision-language tasks, the mechanisms by which they extract query-relevant visual features from complex, noisy contexts remain opaque. In this paper, we present an in-depth interpretability study that uncovers a profound structural property within MLLMs: functional sparsity in cross-modal retrieval. Leveraging a token-level metric termed Retrieval Attention Mass (RAM), we identify and characterize a highly specialized subset of attention heads, referred to as Context-aware Retrieval (CoRe) heads. Across diverse visual domains and model scales, we observe a clear functional division: CoRe heads act as dedicated information extractors, while most other heads distribute attention over broader contextual regions. Causal interventions further demonstrate the necessity of these specialized heads. Ablating only the top 5% of CoRe heads causes significant degradation in multimodal reasoning performance, whereas ablating lower-ranked heads has minimal effect. Moreover, acceleration experiments validate the utility of CoRe heads, showing that leveraging this localized sparsity significantly accelerates inference while maintaining robust task performance. Our findings reveal a structural principle of functional sparsity within MLLMs, refining the current understanding of mechanistic interpretability and laying a theoretical foundation that can inspire future architecture design and model optimization.