Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Liu, Zheng, Liu, Mengjie, Chen, Jingzhou, Xu, Jingwei, Cui, Bin, He, Conghui, Zhang, Wentao
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2504.09925
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918472241381376
author Liu, Zheng
Liu, Mengjie
Chen, Jingzhou
Xu, Jingwei
Cui, Bin
He, Conghui
Zhang, Wentao
author_facet Liu, Zheng
Liu, Mengjie
Chen, Jingzhou
Xu, Jingwei
Cui, Bin
He, Conghui
Zhang, Wentao
contents We introduce FLARE, a family of vision language models (VLMs) with a fully vision-language alignment and integration paradigm. Unlike existing approaches that rely on single MLP projectors for modality alignment and defer cross-modal interaction to LLM decoding, FLARE achieves deep, dynamic integration throughout the pipeline. Our key contributions include: (1) Text-Guided Vision Encoding that incorporates textual information during vision encoding to achieve pixel-level alignment; (2) Context-Aware Alignment Decoding that aggregates visual features conditioned on textual context during decoding for query-level integration; (3) Dual-Semantic Mapping Loss to supervise feature mapping from both modalities and enable modality-level bridging; and (4) Text-Driven VQA Synthesis that leverages high-quality text to generate VQA pairs and synthesize corresponding images, enabling data-level optimization. We train FLARE at 3B and 8B scales under both fixed and dynamic resolution settings, demonstrating that our full-modality alignment significantly outperforms existing methods while maintaining strong generalizability. FLARE 3B surpasses Cambrian-1 8B and Florence-VL 8B using only 630 vision tokens. Ablation studies reveal that FLARE achieves superior performance over existing methods with minimal computational cost. Even without dynamic resolution, FLARE outperforms LLaVA-NeXT, validating the effectiveness of our approach. We release our code, model weights, and dataset in https://github.com/starriver030515/FLARE.
format Preprint
id arxiv_https___arxiv_org_abs_2504_09925
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle FLARE: Fully Integration of Vision-Language Representations for Deep Cross-Modal Understanding
Liu, Zheng
Liu, Mengjie
Chen, Jingzhou
Xu, Jingwei
Cui, Bin
He, Conghui
Zhang, Wentao
Computer Vision and Pattern Recognition
We introduce FLARE, a family of vision language models (VLMs) with a fully vision-language alignment and integration paradigm. Unlike existing approaches that rely on single MLP projectors for modality alignment and defer cross-modal interaction to LLM decoding, FLARE achieves deep, dynamic integration throughout the pipeline. Our key contributions include: (1) Text-Guided Vision Encoding that incorporates textual information during vision encoding to achieve pixel-level alignment; (2) Context-Aware Alignment Decoding that aggregates visual features conditioned on textual context during decoding for query-level integration; (3) Dual-Semantic Mapping Loss to supervise feature mapping from both modalities and enable modality-level bridging; and (4) Text-Driven VQA Synthesis that leverages high-quality text to generate VQA pairs and synthesize corresponding images, enabling data-level optimization. We train FLARE at 3B and 8B scales under both fixed and dynamic resolution settings, demonstrating that our full-modality alignment significantly outperforms existing methods while maintaining strong generalizability. FLARE 3B surpasses Cambrian-1 8B and Florence-VL 8B using only 630 vision tokens. Ablation studies reveal that FLARE achieves superior performance over existing methods with minimal computational cost. Even without dynamic resolution, FLARE outperforms LLaVA-NeXT, validating the effectiveness of our approach. We release our code, model weights, and dataset in https://github.com/starriver030515/FLARE.
title FLARE: Fully Integration of Vision-Language Representations for Deep Cross-Modal Understanding
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.09925