LatentRAG performs agentic RAG by generating latent tokens for thoughts and subqueries in one forward pass, matching explicit methods' accuracy on seven benchmarks while reducing latency by ~90%.
Bring your own knowledge: A survey of methods for LLM knowledge expansion
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CL 2years
2026 2verdicts
UNVERDICTED 2roles
background 1polarities
background 1representative citing papers
A new real-world benchmark shows RAG and continual learning methods fail at continuous knowledge drift in LLMs due to forgetting and inconsistency, while a time-aware retrieval baseline using event evolution graphs improves consistency.
citing papers explorer
-
LatentRAG: Latent Reasoning and Retrieval for Efficient Agentic RAG
LatentRAG performs agentic RAG by generating latent tokens for thoughts and subqueries in one forward pass, matching explicit methods' accuracy on seven benchmarks while reducing latency by ~90%.
-
RAG or Learning? Understanding the Limits of LLM Adaptation under Continuous Knowledge Drift in the Real World
A new real-world benchmark shows RAG and continual learning methods fail at continuous knowledge drift in LLMs due to forgetting and inconsistency, while a time-aware retrieval baseline using event evolution graphs improves consistency.