arXiv: SkMTEB introduces the first MTEB-style text-embedding benchmark for Slovak

arXiv·low signal

SkMTEB (arXiv 2606.13647, Šuppa et al.) is the first comprehensive MTEB-style embedding benchmark for Slovak, a low-resource West Slavic language, paired with model-adaptation experiments. Beyond the specific language, it offers a reusable methodology for evaluating and adapting embedding models in low-resource settings. Mainly relevant to practitioners working on multilingual retrieval or non-English RAG.