RAGQ
A retrieval-augmented generation method that generates citations indicating which input contexts support an answer, with the aim of improving answer verifiability and trustworthiness.
Timeline (last 90 days)
Attributable RAG (RAGQ) improves trustworthiness by generating citations that indicate which input context supports the answer.
Existing RAGQ faces three major pain points when handling questions that require multi-hop reasoning and multimodal understanding: over-citation, failure in implicit entity recognition, and poor generalization.
All Facts (2)
Attributable RAG (RAGQ) improves trustworthiness by generating citations that indicate which input context supports the answer.
50%UnverifiedExisting RAGQ faces three major pain points when handling questions that require multi-hop reasoning and multimodal understanding: over-citation, failure in implicit entity recognition, and poor generalization.
50%