Scholar Feed

Retrieval-augmented generation

DPR

Superseded baseline#32 of 1,179 most-superseded

Superseded — cited as a baseline and beaten by newer methods

4 papers critique it · 3 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites DPR as a baseline.

Although classical retrieval methods like BM25 and DPR conclusively demonstrated that RAG could increase the factual accuracy of LLM-generated responses, they occasionally struggle with the deep semantic understanding of specialised domains
— ModernBERT + ColBERT: Enhancing biomedical RAG through an advanced re-ranking retriever
“While dense retrievers like Dense Passage Retrieval (DPR) and late-interaction models like ColBERT improve recall on single-hop queries, they often struggle with multi-hop reasoning, where the answer depends on composing information from multiple disjoint documents”
— Replace, Don't Expand: Mitigating Context Dilution in Multi-Hop RAG via Fixed-Budget Evidence Assembly
“While these systems demonstrate impressive performance on knowledge-intensive tasks, they primarily optimize for single objective functions under the implicit assumption that retrieved context should always be prioritized.”
— Do Retrieval-Augmented Language Models Adapt to Varying User Needs?
“Dense Passage Retrieval (DPR) models fail to retrieve statements requiring reasoning beyond surface-level similarity.”
— Masking or Mitigating? Deconstructing the Impact of Query Rewriting on Retriever Biases in RAG

Beaten on benchmarks

Head-to-head results where a newer method reports beating DPR. Values are copied from the source paper's tables — verify against the cited paper.

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.