multilingual-e5-base

multilingual-e5-base is an open-weight text embedding model from Microsoft Research that maps text in up to 100 languages into 768-dimensional vectors, initialized from XLM-RoBERTa-base and trained via weakly-supervised contrastive pre-training. It is designed for multilingual semantic search, cross-lingual retrieval, similarity scoring, and clustering tasks. Source: supabase.