Retrieval-Augmented Generation for Large Language Models: A Survey

Fig 1: 系統樹

https://github.com/Tongji-KGLLM/RAG-Survey/blob/main/images/Survey/Tech_tree.png?raw=true

Fig 2: RAGの例、サム・アルトマン解任

A representative instance of the RAG process applied to question answering

https://github.com/Tongji-KGLLM/RAG-Survey/blob/main/images/Survey/RAG_case.png?raw=true

Fig 3: 類型化

https://github.com/Tongji-KGLLM/RAG-Survey/blob/main/images/Survey/RAG_Compare.png?raw=true

Fig 4:

Fine-tuneの領域になかなか立ち入らずに、プロンプトエンジニアリング -> RAG

naive -> advanced -> modular (-> +fine-tuning)

https://github.com/Tongji-KGLLM/RAG-Survey/blob/main/images/Survey/RAG_FT_eng.png?raw=true

Fig 5: 検索過程 3種

https://github.com/Tongji-KGLLM/RAG-Survey/blob/main/images/Survey/RAG_process.png?raw=true

Table 1

https://github.com/Tongji-KGLLM/RAG-Survey/blob/main/images/Survey/method_sumamry.png?raw=true