Pith. sign in

REVIEW 9 cited by

Seeing Clearly by Layer Two: Enhancing Attention Heads to Alleviate Hallucination in LVLMs

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2411.09968 v1 pith:TQVV4L6W submitted 2024-11-15 cs.CV cs.AI

classification cs.CVcs.AI
keywords attentionimagelayertokensheadslayerssinkshallucinations
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

The hallucination problem in multimodal large language models (MLLMs) remains a common issue. Although image tokens occupy a majority of the input sequence of MLLMs, there is limited research to explore the relationship between image tokens and hallucinations. In this paper, we analyze the distribution of attention scores for image tokens across each layer and head of the model, revealing an intriguing and common phenomenon: most hallucinations are closely linked to the pattern of attention sinks in the self-attention matrix of image tokens, where shallow layers exhibit dense attention sinks and deeper layers show sparse attention sinks. We further analyze the attention heads of different layers and find that heads with high-density attention sink in the image part play a positive role in alleviating hallucinations. In this paper, we propose a training-free method named \textcolor{red}{\textbf{E}}nhancing \textcolor{red}{\textbf{A}}ttention \textcolor{red}{\textbf{H}}eads (EAH), an approach designed to enhance the convergence of image tokens attention sinks in the shallow layers. EAH identifies the attention head that shows the vision sink in a shallow layer and extracts its attention matrix. This attention map is then broadcast to other heads in the layer, thereby strengthening the layer to pay more attention to the image itself. With extensive experiments, EAH shows significant hallucination-mitigating performance on different MLLMs and metrics, proving its effectiveness and generality.

Discussion (0). Sign in to comment.

Forward citations

Cited by 9 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. The Ebb and Flow of Multimodal Focus: Scheduling Visual Relay Windows for Grounded VLM Reasoning

    cs.AI 2026-07 conditional novelty 6.5 of 10

    Visual evidence in VLMs flows through a depth-wise Visual Relay Window; scheduling that window with the lightweight TRACE controller yields +4.33 points on grounding benchmarks and +3.05 on MathVista across four open-...

  2. Role-Break in Attention Heads: Understanding and Detecting Hallucinations in VLMs

    cs.CV 2026-07 conditional novelty 6.0 of 10

    Hallucinated tokens in vision-language models are detectable from each attention head's deviation from its own faithful source-allocation role, reaching ~93% AUROC with a linear probe.

  3. ST-Veto: Spatio-Temporal Token Veto for Diffusion MLLMs via Taylor Prediction and Visual Grounding

    cs.AI 2026-07 conditional novelty 6.0 of 10

    ST-Veto improves reasoning in diffusion MLLMs by vetoing temporally unstable tokens and tokens with weak image grounding, swapping in safer near-boundary candidates.

  4. METEOR: Multi-Encoder Collaborative Token Pruning for Efficient Vision Language Models

    cs.CV 2025-07 conditional novelty 6.0 of 10

    METEOR is a three-stage token pruning framework that reduces visual tokens in multi-encoder MLLMs by 76% with only a 0.3% average accuracy drop.

  5. Kwai Keye-VL Technical Report

    cs.CV 2025-07 conditional novelty 6.0 of 10

    Kwai Keye-VL shows that a five-mode chain-of-thought cold-start plus mix-mode reinforcement learning can push an 8B multimodal model to strong short-video and general vision-language performance.

  6. DenseWorld-1M: Towards Detailed Dense Grounded Caption in the Real World

    cs.CV 2025-06 conditional novelty 6.0 of 10

    DenseWorld-1M provides one million images with detailed object captions, pixel masks, and spatial relations by chaining SAM, APE, RAM++, and VLMs through a three-stage labeling pipeline.

  7. Dense360: Dense Understanding from Omnidirectional Panoramas

    cs.CV 2025-06 reject novelty 6.0 of 10

    Introduces a 160K-panorama auto-annotated dataset, a dense captioning and grounding benchmark, and ERP-RoPE; fine-tuning Qwen2.5VL on the data lifts benchmark scores.

  8. Mitigating Object Hallucination via Robust Local Perception Search

    cs.CV 2025-06 conditional novelty 6.0 of 10

    A training-free decoding method that uses an MLLM's own local object descriptions as a reward prior, combined with CLIP similarity, to cut object hallucination, especially under adversarial image noise.

  9. MCA-LLaVA: Manhattan Causal Attention for Reducing Hallucination in Large Vision-Language Models

    cs.CV 2025-07 conditional novelty 4.0 of 10

    MCA-LLaVA reindexes image tokens by sums of mirrored 2D coordinates so instruction tokens attend across the whole image, reducing hallucination on POPE, CHAIR, and MME.

Pith tools