MIKE
收藏资源简介:
MIKE是一个专为细粒度多模态实体知识编辑设计的数据集,由未明确的研究机构创建。该数据集包含1103个细粒度实体,每个实体至少有5张图像,涵盖9个超级类别,主要用于评估多模态大型语言模型在细粒度实体识别和编辑方面的能力。数据集的创建过程涉及从OVEN数据集中选择实体,并通过搜索引擎收集图像,随后由经验丰富的注释者进行筛选。MIKE数据集的应用领域主要集中在提升多模态大型语言模型在实际场景中的部署和效果,特别是在需要精确信息处理的领域。
MIKE is a dataset specifically designed for fine-grained multimodal entity knowledge editing, created by an unspecified research institution. This dataset comprises 1103 fine-grained entities, each with at least 5 images, covering 9 super categories, and is primarily used to evaluate the capabilities of multimodal large language models in fine-grained entity recognition and editing. The dataset construction process involves selecting entities from the OVEN dataset, collecting images via search engines, and then conducting screening by experienced annotators. The application scenarios of the MIKE dataset mainly focus on enhancing the deployment and performance of multimodal large language models in real-world scenarios, particularly in fields requiring precise information processing.

- 1MIKE: A New Benchmark for Fine-grained Multimodal Entity Knowledge Editing未提及 · 2024年



