让开源模型学会用图片回答视觉问题，效果逼近闭源大模型。
VQ-VA World: Towards High-Quality Visual Question-Visual Answering
- 构建自动化数据流水线，生成180万条高质量图文对用于训练。
- 训练后模型在新基准上得分53.06，远超此前开源模型表现。
- 适合关注多模态生成、开放域视觉理解的研究者使用。
本文研究视觉问答生成图像（VQ-VA）任务：以图像而非文本回应视觉问题，该能力已见于纳米香蕉（NanoBanana）和GPT-Image等闭源系统。为使开源模型也具备此能力，我们提出VQ-VA World，一个以智能体驱动的规模化数据构建框架。通过网络规模部署，该流水线爬取约180万条高质量、交错排列的图文样本用于训练。为评估性能，我们进一步发布IntelligentBench，一个由人工标注的基准，系统性评估模型在世界知识、设计知识和推理能力方面的表现。使用VQ-VA World数据训练的LightFusion模型在IntelligentBench上取得53.06分，显著超越现有开源基线（如原始LightFusion为7.78，UniWorld-V1为1.94），并大幅缩小与领先闭源系统（如NanoBanana为81.67，GPT-Image为82.64）的差距。我们公开全部模型权重、数据集及流水线，旨在推动该方向研究发展。
原文摘要 · Abstract (English)
This paper studies Visual Question-Visual Answering (VQ-VA): generating an image, rather than text, in response to a visual question -- an ability that has recently emerged in proprietary systems such as NanoBanana and GPT-Image. To also bring this capability to open-source models, we introduce VQ-VA World, a data-centric framework built around an agentic pipeline for large-scale, targeted data construction. Leveraging web-scale deployment, this pipeline crawls a massive amount of ~1.8M high-quality, interleaved image-text samples for model training. For evaluation, we further release IntelligentBench, a human-curated benchmark that systematically assesses VQ-VA along the aspects of world knowledge, design knowledge, and reasoning. Training with VQ-VA World data yields strong empirical gains: it helps LightFusion attain 53.06 on IntelligentBench, substantially surpassing the best prior open-source baselines (i.e., 7.78 from vanilla LightFusion; 1.94 from UniWorld-V1), and significantly narrowing the gap toward leading proprietary systems (e.g., 81.67 from NanoBanana; 82.64 from GPT-Image). By releasing the full suite of model weights, datasets, and pipelines, we hope to stimulate future research on VQ-VA.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。