A Korean text embedding model based on SRoBERTa trained with multiple negative ranking loss for semantic search.