Russian

Russian text embedding quality across classification, clustering, reranking, pair classification, retrieval, and semantic similarity. In v1.1, MIRACLRetrieval and RiaNewsRetrieval were replaced with their HardNegatives variants (v2), which include improved default prompts.

Reference paper →

Cite this benchmark

Citation (BibTeX)


@misc{snegirev2024russianfocusedembeddersexplorationrumteb,
  archiveprefix = {arXiv},
  author = {Artem Snegirev and Maria Tikhonova and Anna Maksimova and Alena Fenogenova and Alexander Abramov},
  eprint = {2408.12503},
  primaryclass = {cs.CL},
  title = {The Russian-focused embedders' exploration: ruMTEB benchmark and Russian embedding model design},
  url = {https://arxiv.org/abs/2408.12503},
  year = {2024},
}

Languages 1

Tasks 23

Task Types 7

Models 0