ruMTEB

该数据集名为ruMTEB基准，是大规模文本嵌入基准（MTEB）的俄语版本，涵盖了七个类别的任务，包括语义文本相似度、文本分类、重新排序和检索等。该基准评估了一系列俄语及多语言模型的代表性集合，研究结果显示，新型模型ru-en-RoSBERTa在俄语任务中的表现与现有最先进模型持平。该数据集对多种嵌入模型在各项任务上的表现进行了评估，这些任务涵盖了语义文本相似度、文本分类、重新排序和检索等多个领域。The dataset is named ruMTEB Benchmark, which is the Russian-language variant of the Massive Text Embedding Benchmark (MTEB). It encompasses tasks across seven categories, including semantic textual similarity, text classification, re-ranking, retrieval, and more. This benchmark evaluates a representative collection of both monolingual Russian and multilingual embedding models. Research results show that the newly developed model ru-en-RoSBERTa achieves performance on par with current state-of-the-art models on Russian-language tasks. Additionally, this dataset assesses the performance of various embedding models across a diverse range of tasks, covering semantic textual similarity, text classification, re-ranking, retrieval, and other related fields.

搜集汇总

背景与挑战

背景概述

该内容描述了Massive Text Embedding Benchmark (MTEB)，这是一个用于评估文本嵌入模型的综合性基准测试平台，支持多种任务如分类、聚类和检索，并涵盖多语言数据集。MTEB提供了安装、使用和贡献的详细指南，以及一个交互式排行榜，用于比较不同模型的性能。