superlinked/external-benchmarking

这是一个用于向量搜索性能基准测试的数据集，包含来自Amazon Reviews 2023的数据，生成了包含丰富元数据（包括数值、类别、文本和图像）的4154维向量。数据集分为10k、100k、1M和10M四个规模，以解决其他基准数据集中元数据不足、向量大小不够和数据集规模较小的问题。This is a dataset for vector search performance benchmarking, containing data from Amazon Reviews 2023, generating 4154-dimensional vectors with rich metadata (including numerical, categorical, textual, and image data). The dataset is divided into four sizes: 10k, 100k, 1M, and 10M, to address the issues of insufficient metadata, small vector size, and small dataset scale found in other benchmark datasets.