Speeding up RAG search without an index: New method "DecisionRAG" traces documents using decision-making AI
On October 26 (local time), a research team including Cornell University released a new method called "DecisionRAG" on arXiv, which finds documents at high speed without using any search engine. Even with Wikipedia's 2.68 million paragraphs, it finds the supporting paragraph within approximately 2 seconds per question.
On October 26 (local time), a research team including Cornell University in the US released a new method called "DecisionRAG" on the paper submission site arXiv, which finds documents at high speed without using any search engine.
Information that search misses cannot be recovered by even the smartest generative AI. An "AI that cannot write text at all" may compensate for this weakness of AI search.
With this method, even when facing Wikipedia with 2.68 million paragraphs, it finds the supporting paragraph within approximately 2 seconds per question.
This item is a summary. The full, checkable version is at the original source:
Read the original at PC Watch ↗