New SAP HANA NLP Service Features Cross Encoder and Dimensionality Reduction for Enhanced Text Embedding

SAP HANA Cloud Predictive Analysis Library has introduced a new cross-encoder model for NLP service, enhancing the existing embedding model. The cross-encoder offers high accuracy and precision, but is computationally expensive and limited to processing input pairs. It can be combined with the bi-encoder model to form an Advanced RAG architecture for retrieve and rerank pipelines. The library also supports dimensionality reduction for embedding models.