# Multi-Layer Visual Feature Fusion in Multimodal LLMs: Methods, Analysis, and Best Practices

arXiv: https://arxiv.org/abs/2503.06063 · HTML: https://mlsanity.com/paper/2503.06063/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 58 of its 68 references that are in the map.
- First posted on arXiv 8 Mar 2025.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM training and visual instruction tuning](https://mlsanity.com/mllm/mllm-training-and-visual-instruction-tuning/)

## Builds on (most cited first, 30 of 58)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- [Improved Baselines with Visual Instruction Tuning](https://mlsanity.com/paper/2310.03744/) (2023)
- [Flamingo: a Visual Language Model for Few-Shot Learning](https://mlsanity.com/paper/2204.14198/) (2022)
- [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://mlsanity.com/paper/2201.12086/) (2022)
- [LAION-5B: An open large-scale dataset for training next generation image-text models](https://mlsanity.com/paper/2210.08402/) (2022)
- [OPT: Open Pre-trained Transformer Language Models](https://mlsanity.com/paper/2205.01068/) (2022)
- [Mistral 7B](https://mlsanity.com/paper/2310.06825/) (2023)
- [Sigmoid Loss for Language Image Pre-Training](https://mlsanity.com/paper/2303.15343/) (2023)
- [MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Models](https://mlsanity.com/paper/2304.10592/) (2023)
- [Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations](https://mlsanity.com/paper/1602.07332/) (2016)
- [InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks](https://mlsanity.com/paper/2312.14238/) (2023)
- [Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answering](https://mlsanity.com/paper/1612.00837/) (2016)
- [VQA: Visual Question Answering](https://mlsanity.com/paper/1505.00468/) (2015)
- [Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond](https://mlsanity.com/paper/2308.12966/) (2023)
- [MMBench: Is Your Multi-modal Model an All-around Player?](https://mlsanity.com/paper/2307.06281/) (2023)
- [Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone](https://mlsanity.com/paper/2404.14219/) (2024)
- [Evaluating Object Hallucination in Large Vision-Language Models](https://mlsanity.com/paper/2305.10355/) (2023)
- [Towards VQA Models That Can Read](https://mlsanity.com/paper/1904.08920/) (2019)
- [MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models](https://mlsanity.com/paper/2306.13394/) (2023)
- [LAION-400M: Open Dataset of CLIP-Filtered 400 Million Image-Text Pairs](https://mlsanity.com/paper/2111.02114/) (2021)
- [DocVQA: A Dataset for VQA on Document Images](https://mlsanity.com/paper/2007.00398/) (2020)
- [Conceptual 12M: Pushing Web-Scale Image-Text Pre-Training To Recognize Long-Tail Visual Concepts](https://mlsanity.com/paper/2102.08981/) (2021)
- [OK-VQA: A Visual Question Answering Benchmark Requiring External Knowledge](https://mlsanity.com/paper/1906.00067/) (2019)
- [Generation and Comprehension of Unambiguous Object Descriptions](https://mlsanity.com/paper/1511.02283/) (2015)
- [Cambrian-1: A Fully Open, Vision-Centric Exploration of Multimodal LLMs](https://mlsanity.com/paper/2406.16860/) (2024)
- [A Diagram Is Worth A Dozen Images](https://mlsanity.com/paper/1603.07396/) (2016)

## Built on by (most cited first, 22 of 22)

- [VisiPruner: Decoding Discontinuous Cross-Modal Dynamics for Efficient Multimodal LLMs](https://mlsanity.com/paper/2510.17205/) (2025)
- [Do Audio LLMs Listen or Read? Analyzing and Mitigating Paralinguistic Failures with VoxParadox](https://mlsanity.com/paper/2605.27772/) (2026)
- [Seeing Beyond Words: Self-Supervised Visual Learning for Multimodal Large Language Models](https://mlsanity.com/paper/2512.15885/) (2025)
- [Recurrence Meets Transformers for Universal Multimodal Retrieval](https://mlsanity.com/paper/2509.08897/) (2025)
- [HERMAN: Hierarchical Representation Matching for CLIP-based Class-Incremental Learning](https://mlsanity.com/paper/2509.22645/) (2025)
- [RETLLM: Training and Data-Free MLLMs for Multimodal Information Retrieval](https://mlsanity.com/paper/2602.22278/) (2026)
- [Unleashing the Intrinsic Visual Representation Capability of Multimodal Large Language Models](https://mlsanity.com/paper/2512.06281/) (2025)
- [TGIF: Text-Guided Layer Fusion Mitigates Hallucination in Multimodal LLMs](https://mlsanity.com/paper/2601.03100/) (2026)
- [VAT: Vision Action Transformer by Unlocking Full Representation of ViT](https://mlsanity.com/paper/2512.06013/) (2025)
- [Dynamic Image Prompt Adapter for Scalable Zero-shot Personalized Text-to-Image Generation](https://mlsanity.com/paper/2512.09814/) (2025)
- [From One-to-One to Many-to-Many: Dynamic Cross-Layer Injection for Deep Vision-Language Fusion](https://mlsanity.com/paper/2601.10710/) (2026)
- [E.M.Ground: A Temporal Grounding Vid-LLM with Holistic Event Perception and Matching](https://mlsanity.com/paper/2602.05215/) (2026)
- [Boosting Visual Instruction Tuning with Self-Supervised Guidance](https://mlsanity.com/paper/2604.12966/) (2026)
- [InstructPLM-mu: 1-Hour Fine-Tuning of ESM2 Beats ESM3 in Protein Mutation Predictions](https://mlsanity.com/paper/2510.03370/) (2025)
- [Detached Skip-Links and R-Probe: Decoupling Feature Aggregation from Gradient Propagation for MLLM OCR](https://mlsanity.com/paper/2603.20020/) (2026)
- [CLASP: Class-Adaptive Layer Fusion and Dual-Stage Pruning for Multimodal Large Language Models](https://mlsanity.com/paper/2604.12767/) (2026)
- [The Hidden Evolution of Disguised Visual Context inside the VLM](https://mlsanity.com/paper/2606.20077/) (2026)
- [Seeing Before Answering: Training-Free Visual Layer Profiling for Vision-Language Models](https://mlsanity.com/paper/2608.16263/) (2026)
- [PACER: Progressive Availability-Conditioned Evidence Routing for Radiology Report Generation under Incomplete Clinical Context](https://mlsanity.com/paper/2609.34487/) (2026)
- [eRLT: Efficient VLA Reinforcement Learning via Action-Relevant Token Routing](https://mlsanity.com/paper/2610.00913/) (2026)
- [Where-OPD: Spatially Guided On-Policy Self-Distillation of MLLMs with Synthetic Scenes](https://mlsanity.com/paper/2610.02117/) (2026)
- [Sparse All-Layer Connector for Domain Generalised Semantic Segmentation](https://mlsanity.com/paper/2610.15660/) (2026)

## Abstract (arXiv)

Multimodal Large Language Models (MLLMs) have made significant advancements in recent years, with visual features playing an increasingly critical role in enhancing model performance. However, the integration of multi-layer visual features in MLLMs remains underexplored, particularly with regard to optimal layer selection and fusion strategies. Existing methods often rely on arbitrary design choices, leading to suboptimal outcomes. In this paper, we systematically investigate two core aspects of multi-layer visual feature fusion: (1) selecting the most effective visual layers and (2) identifying the best fusion approach with the language model. Our experiments reveal that while combining visual features from multiple stages improves generalization, incorporating additional features from the same stage typically leads to diminished performance. Furthermore, we find that direct fusion of multi-layer visual features at the input stage consistently yields superior and more stable performance across various configurations. We make all our code publicly available: https://github.com/EIT-NLP/Layer_Select_Fuse_for_MLLM.