paraphrase-multilingual-mpnet-base-v2

A multilingual sentence-embedding model from the Sentence Transformers project that encodes text from 50+ languages into 768-dimensional dense vectors for semantic similarity, paraphrase detection, clustering, and cross-lingual semantic search. It extends the English-only paraphrase-mpnet-base-v2 to multilingual settings via knowledge distillation. Source: supabase.