Blended RAG

Reported on 5 benchmarks across 2 tasks · 1 paper · 2 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Natural Language Processing5 results

Question AnsweringonNQ (BEIR)
nDCG@10· 2024-03-22
0.67
SOTA
Blended RAG: Improving RAG (Retriever-Augmented Generation) Accuracy with Semantic Search and Hybrid Query-Based Retrievers arXiv:2404.07220
Question AnsweringonSQuAD
Exact Match· 2024-03-22
57.63
SOTA
Blended RAG: Improving RAG (Retriever-Augmented Generation) Accuracy with Semantic Search and Hybrid Query-Based Retrievers arXiv:2404.07220
Question AnsweringonNatural Questions
EM· 2024-03-22
42.63
best: 64 (Atlas (full, Wiki-dec-2018 index))
Blended RAG: Improving RAG (Retriever-Augmented Generation) Accuracy with Semantic Search and Hybrid Query-Based Retrievers arXiv:2404.07220
Question AnsweringonSQuAD1.1 dev
EM· 2024-03-22
57.63
best: 90.06 (T5-11B)
Blended RAG: Improving RAG (Retriever-Augmented Generation) Accuracy with Semantic Search and Hybrid Query-Based Retrievers arXiv:2404.07220
Open-Domain Question AnsweringonSQuAD1.1 dev
EM· 2024-03-22
57.63
best: 59.3 (SPARTA)
Blended RAG: Improving RAG (Retriever-Augmented Generation) Accuracy with Semantic Search and Hybrid Query-Based Retrievers arXiv:2404.07220