In Defense of RAG in the Era of Long-Context Language Models

Tan Yu, Anbang Xu, Rama Akkiraju

Overcoming the limited context limitations in early-generation LLMs, retrieval-augmented generation (RAG) has been a reliable solution for context-based answer generation in the pa…