CausalLM/Retrievatar

Retrievatar 是一个多模态数据集，旨在增强视觉语言模型的检索增强生成能力，特别关注虚构的动漫角色和现实世界的各界名人。该数据集包含 100,000 个样本，具有多语言描述（英语、中文、日语和德语）。描述是通过 Gemini-2.5-pro GA 模型生成的，利用了 Google 搜索落地和反向图像搜索结果的元数据。数据集的目标是通过提供更全面的实体表示，解决传统视觉语言模型训练中的局限性。

Retrievatar is a multimodal dataset designed to enhance the retrieval-augmented generation capabilities of vision-language models, specifically focusing on fictional anime characters and real-world celebrities across various fields. This release represents a subset of 100,000 samples with multilingual captions in English, Chinese, Japanese, and German. The captions were generated using the Gemini-2.5-pro GA model, leveraging Grounding with Google Search and metadata from reverse image search results. The dataset aims to address limitations in traditional Vision-Language Model training by providing more holistic representations of entities.