# Scaling Representation Diversity: Modulated Attention and Reconstructive Regularization for Visual Grounding

arXiv: https://arxiv.org/abs/2608.12748 · HTML: https://mlsanity.com/paper/2608.12748/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 23 of its 35 references that are in the map.
- First posted on arXiv 13 Aug 2026.

## Builds on (most cited first, 23 of 23)

- [Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detection](https://mlsanity.com/paper/2303.05499/) (2023)
- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations](https://mlsanity.com/paper/1602.07332/) (2016)
- [FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning](https://mlsanity.com/paper/2307.08691/) (2023)
- [Modeling Context in Referring Expressions](https://mlsanity.com/paper/1608.00272/) (2016)
- [Shikra: Unleashing Multimodal LLM's Referential Dialogue Magic](https://mlsanity.com/paper/2306.15195/) (2023)
- [MDETR -- Modulated Detection for End-to-End Multi-Modal Understanding](https://mlsanity.com/paper/2104.12763/) (2021)
- [Generalizing to Unseen Domains: A Survey on Domain Generalization](https://mlsanity.com/paper/2103.03097/) (2021)
- [Prevalence of Neural Collapse during the terminal phase of deep learning training](https://mlsanity.com/paper/2008.08186/) (2020)
- [Ferret: Refer and Ground Anything Anywhere at Any Granularity](https://mlsanity.com/paper/2310.07704/) (2023)
- [GLaMM: Pixel Grounding Large Multimodal Model](https://mlsanity.com/paper/2311.03356/) (2023)
- [MAttNet: Modular Attention Network for Referring Expression Comprehension](https://mlsanity.com/paper/1801.08186/) (2018)
- [Understanding Dimensional Collapse in Contrastive Self-supervised Learning](https://mlsanity.com/paper/2110.09348/) (2021)
- [TransVG: End-to-End Visual Grounding with Transformers](https://mlsanity.com/paper/2104.08541/) (2021)
- [Qwen3-VL-Embedding and Qwen3-VL-Reranker: A Unified Framework for State-of-the-Art Multimodal Retrieval and Ranking](https://mlsanity.com/paper/2601.04720/) (2026)
- [RankMe: Assessing the downstream performance of pretrained self-supervised representations by their rank](https://mlsanity.com/paper/2210.02885/) (2022)
- [VL-JEPA: Joint Embedding Predictive Architecture for Vision-language](https://mlsanity.com/paper/2512.10942/) (2025)
- [HiVG: Hierarchical Multimodal Fine-grained Modulation for Visual Grounding](https://mlsanity.com/paper/2404.13400/) (2024)
- [A Closer Look at Multimodal Representation Collapse](https://mlsanity.com/paper/2505.22483/) (2025)
- [Hierarchical Alignment-enhanced Adaptive Grounding Network for Generalized Referring Expression Comprehension](https://mlsanity.com/paper/2501.01416/) (2025)
- [PropVG: End-to-End Proposal-Driven Visual Grounding with Multi-Granularity Discrimination](https://mlsanity.com/paper/2509.04833/) (2025)
- [Text-Conditional JEPA for Learning Semantically Rich Visual Representations](https://mlsanity.com/paper/2605.03245/) (2026)
- [Comprehension of Multilingual Expressions Referring to Target Objects in Visual Inputs](https://mlsanity.com/paper/2511.11427/) (2025)

## Abstract (arXiv)

Referring Expression Comprehension (REC) is commonly studied under dataset-specific fine-tuning, resulting in specialist models with limited cross-dataset generalization. In this work, we revisit REC from the perspective of unified open-vocabulary grounding and identify representation degeneration as a key obstacle to scaling a single generalist model. To preserve representation diversity, we propose a holistic data-model co-design framework. Architecturally, we introduce the Modulated Attention-Contrastive Head (mACH) for efficient token-level vision-language alignment and a text-conditioned JEPA auxiliary stream that provides complementary gradient support to preserve alignment-active representations without inference overhead. On the data side, we introduce Objects365-Caption, enriching Objects365 with context-aware referring expressions for large-scale language supervision. We further provide a theoretical analysis showing that complementary gradient subspaces preserve alignment capacity and thereby scale representation diversity. Extensive experiments demonstrate that our single-checkpoint framework achieves highly competitive performance on standard REC benchmarks while exhibiting strong generalization across heterogeneous grounding datasets without benchmark-specific adaptation.