RAG - Retrieval-Augmented Generation

Cross-Encoder Reranking

A relevance evaluation method where the model processes the query and candidate document simultaneously, unlike the standard bi-encoder approach.

← Tillbaka