Understand What LLM Needs: Dual Preference Alignment for Retrieval-Augmented Generation
Open paper
Benchmark
Model
Rank
Results
knowledge-base-question-answering-on-1
DPA-RAG
#2
Hits@1: 80.28
F1: 81.74
question-answering-on-natural-questions
DPA-RAG
#3
EM: 59.19
question-answering-on-triviaqa
DPA-RAG
#31
F1: 80.1
Rank counts only results with a code link.