# GLIMPSE: Holistic Cross-Modal Explainability for Large Vision-Language Models

arXiv: https://arxiv.org/abs/2506.18985 · HTML: https://mlsanity.com/paper/2506.18985/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Placed in VLM benchmarks and visual reasoning, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 9 of its 28 references that are in the map.
- First posted on arXiv 23 Jun 2025.

## Placed in

- Multimodal LLMs / vision-language models / [VLM benchmarks and visual reasoning](https://mlsanity.com/mllm/vlm-benchmarks-and-visual-reasoning/)

## Builds on (most cited first, 9 of 9)

- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond](https://mlsanity.com/paper/2308.12966/) (2023)
- [AttnLRP: Attention-Aware Layer-Wise Relevance Propagation for Transformers](https://mlsanity.com/paper/2402.05602/) (2024)
- [AtMan: Understanding Transformer Predictions Through Memory Efficient Attention Manipulation](https://mlsanity.com/paper/2301.08110/) (2023)
- [LVLM-Interpret: An Interpretability Tool for Large Vision-Language Models](https://mlsanity.com/paper/2404.03118/) (2024)
- [Visual Explanations via Iterated Integrated Attributions](https://mlsanity.com/paper/2310.18585/) (2023)
- [Attention, Please! PixelSHAP Reveals What Vision-Language Models Actually Focus On](https://mlsanity.com/paper/2503.06670/) (2025)
- [Uncertainty estimates for semantic segmentation: providing enhanced reliability for automated motor claims handling](https://mlsanity.com/paper/2401.09245/) (2024)

## Built on by (most cited first, 5 of 5)

- [Saliency-R1: Enforcing Interpretable and Faithful Vision-language Reasoning via Saliency-map Alignment Reward](https://mlsanity.com/paper/2604.04500/) (2026)
- [Through Their Eyes: Fixation-aligned Tuning for Personalized User Emulation](https://mlsanity.com/paper/2604.09368/) (2026)
- [Real-Time Visual Attribution Streaming in Thinking Model](https://mlsanity.com/paper/2604.16587/) (2026)
- [AnswerMap: Faithful Spatial Interpretability of VLMs from Answer Posteriors](https://mlsanity.com/paper/2609.35247/) (2026)
- [Tracing the Evidence: Faithful Token Attribution Through Vision-Language Reasoning](https://mlsanity.com/paper/2609.37656/) (2026)

## Abstract (arXiv)

Recent large vision-language models (LVLMs) have advanced capabilities in visual question answering (VQA). However, interpreting where LVLMs direct their visual attention remains a significant challenge, yet is essential for understanding model behavior. We introduce GLIMPSE (Gradient-Layer Importance Mapping for Prompted Visual Saliency Explanation), a lightweight, model-agnostic framework that jointly attributes LVLM outputs to the most relevant visual evidence and textual signals that support open-ended generation. GLIMPSE fuses gradient-weighted attention, adaptive layer propagation, and relevance-weighted token aggregation to produce holistic response-level heat maps for interpreting cross-modal reasoning, outperforming prior methods in faithfulness and pushing the state-of-the-art in human-attention alignment. We demonstrate an analytic approach to uncover fine-grained insights into LVLM cross-modal attribution, trace reasoning dynamics, analyze systematic misalignment, diagnose hallucination and bias, and ensure transparency.