A specialist that thinks in pairs — this model encodes Vietnamese text into dense vector representations using a dual-encoder architecture, making it well-suited for semantic similarity and retrieval tasks. It processes two pieces of text independently into comparable embeddings, excelling at matching queries to relevant passages in Vietnamese. Its focus is narrow but deliberate: Vietnamese language understanding, not multilingual generalization.