Hybrid Retrieval

Cross-Encoder

Neural model architecture that simultaneously encodes queries and documents to predict their mutual relevance. More accurate but slower than bi-encoders, often used for re-ranking hybrid results.

← Kembali