starriver030515/FUSION-LLaMA3.1-8B

Model Card for FUSION

This is the checkpoint after Stage 1, Stage1.5 and Stage2 training of FUSION-LLaMA3.1-8B.

This repository contains the model described in the paper FUSION: Fully Integration of Vision-Language Representations for Deep Cross-Modal Understanding.

Model Details

Model Description

FUSION is a family of multimodal large language models that adopts a fully integrated vision-language architecture, enabling comprehensive and fine-grained cross-modal understanding. In contrast to prior approaches that primarily perform shallow or late-stage modality fusion during the LLM decoding phase, FUSION achieves deep, dynamic integration across the entire vision-language processing pipeline.

To enable this, FUSION utilizes Text-Guided Unified Vision Encoding, which incorporates textual context directly into the vision encoder. This design allows for pixel-level vision-language alignment and facilitates early-stage cross-modal interaction.

During decoding, FUSION employs Context-Aware Recursive Alignment Decoding strategy. This component dynamically aggregates and refines visual features based on the evolving textual context at each decoding step, allowing the model to capture question-level semantics with high precision.

To further enhance alignment and reduce the semantic gap between modalities, FUSION integrates Dual-Supervised Semantic Mapping Loss, which provides simultaneous supervision in both visual and textual embedding spaces. This dual-path guidance strengthens the consistency and semantic coherence of the fused representations.

Base Model

LLM: meta-llama/Llama-3.1-8B-Instruct

Vision Encoder: google/siglip-so400m-patch14-384

Training Details

Training Strategies

FUSION is trained with a three-stage training framework, ensuring comprehensive alignment and integration between visual and linguistic modalities.

• Stage1: Foundational Semantic Alignment: We pretrain the vision encoder using extensive image-caption datasets to establish precise semantic alignment be- tween visual and textual representations.
• Stage1.5: Contextual Multimodal Fusion: In contrast to Stage 1, this intermediate stage incorporates various types of QA data along with image-caption pairs. This phase is designed to enhance the model’s adaptability in aligning vision and language representations across a broad spectrum of scenarios.
• Stage2: Visual Instruction Tuning: At this stage, we expose the model to various visual tasks, enabling it to answer downstream vision-related questions effectively.

Training Data

• 10M FUSION Alignment Data For Stage1
• 12M FUSION Curated Instruction Tuning Data For Stage1.5 and Stage2

Performance

Where to send questions or comments about the model:

https://github.com/starriver030515/FUSION/issues

Paper or resources for more information

• https://arxiv.org/abs/2504.09925
• https://github.com/starriver030515/FUSION

Citation

If you find FUSION useful for your research and applications, please cite using this BibTeX:

@misc{liu2025fusionfullyintegrationvisionlanguage,
      title={FUSION: Fully Integration of Vision-Language Representations for Deep Cross-Modal Understanding}, 
      author={Zheng Liu and Mengjie Liu and Jingzhou Chen and Jingwei Xu and Bin Cui and Conghui He and Wentao Zhang},
      year={2025},
      eprint={2504.09925},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2504.09925}, 
}