xuhaike/bge-reasoner_embedding_bright

该数据集包含使用BGE-Reasoner模型为BRIGHT基准测试预计算的嵌入向量。数据集包含23个不同领域的嵌入文件，每个领域都有对应的段落嵌入和查询嵌入（部分领域只有段落嵌入）。这些嵌入文件涵盖了多个学科领域，包括数学（aops）、生物学、地球科学、经济学、编程（leetcode）、心理学、机器人学、技术问答（stackoverflow）、可持续生活以及数学定理相关的内容。所有嵌入都是1024维的，并经过L2归一化处理。数据集主要用于特征提取和句子相似性任务。

This dataset contains pre-computed embeddings for the BRIGHT benchmark using the BGE-Reasoner model. It includes 23 embedding files across various domains, with both passage and query embeddings (some domains have only passage embeddings). The domains cover multiple disciplines including mathematics (aops), biology, earth science, economics, programming (leetcode), psychology, robotics, technical Q&A (stackoverflow), sustainable living, and mathematical theorem-related content. All embeddings are 1024-dimensional and L2-normalized. The dataset is primarily intended for feature extraction and sentence similarity tasks.