kafked/anycrap

ANYCRAP是一个包含超过120,000个AI生成的荒诞和创意产品概念的数据集，收集自ANYCRAP平台，时间跨度约一年。产品范围从温和的文字游戏到超现实荒诞主义，具有多语言名称、AI生成的描述和图像、参与度数据、手动类别标签以及AI驱动的质量评分。数据集提供三种配置：full包含所有产品（126K个），featured为手动策划的子集（35K个，带类别标签），evaluated为通过Crap-O-Meter系统评分的子集（28K个，评估文本连贯性、创造力和图像相关性）。潜在用例包括创意写作、幽默研究、创造力基准测试、内容生成、占位数据、游戏项目名称生成以及NSFW内容研究。ANYCRAP is a dataset of over 120,000 AI-generated absurdist and creative product concepts, collected over approximately one year from the ANYCRAP catalog platform. Products range from gentle wordplay to surreal absurdism, with multilingual names, AI-generated descriptions and images, engagement data, manual category labels, and AI-powered quality scores. The dataset includes three configurations: full with all products (126K), featured as a manually curated subset (35K with category labels), and evaluated as a subset scored by the Crap-O-Meter system (28K, assessing text coherence, creativity, and image relevance). Potential use cases include creative writing, humor research, creativity benchmarking, content generation, placeholder data, game item name generation, and NSFW content research.