Corrective/Self-Reflective RAG paper: https://arxiv.org/abs/2401.15884 code: https://github.com/DeEnabler/CRAG Florian blog: https://ai.gopubby.com/advanced-rag-10-corrective-retrieval-augmented-generation-crag-3f5a140796f9 (2024/4/18) - langchain blog: https://blog.langchain.dev/agentic-rag-with-langgraph/ (2024/2/7) https://www.youtube.com/watch?v=E2shqsYwxck langgraph code: https://github.com/langchain-ai/langgraph/blob/main/examples/rag/langgraph_crag.ipynb (有圖) langgraph https://twitter.com/RLanceMartin/status/1771261432870682987 (有圖) ![](https://pbs.twimg.com/media/GJTGrhxacAAXFLX?format=jpg&name=large) 使用輕量級檢索評估器來評估查詢的檢索文件的整體質量,為每個返回一個信心分數。 如果任何文件與主題無關,我們將通過網絡搜索進行補充 使用查詢重寫來優化網頁搜索的查詢 https://twitter.com/bindureddy/status/1752450764956660189 https://twitter.com/omarsar0/status/1752173216942944556 https://twitter.com/sophiamyang/status/1755629945080943030 https://twitter.com/RLanceMartin/status/1755287312295170232 https://twitter.com/ravithejads/status/1754003746831364480 https://twitter.com/RLanceMartin/status/1755287316799828233 langchain 實作 https://twitter.com/LangChainAI/status/1755640272271405347 langchain https://twitter.com/LangChainAI/status/1755281809880027623 langchain https://twitter.com/virattt/status/1755999186103984145 成功案例: 10-K 文件 https://twitter.com/virattt/status/1755340393124946281 https://twitter.com/sophiamyang/status/1755629945080943030 MistralAI cookbook https://twitter.com/RLanceMartin/status/1757456064759705891 https://github.com/langchain-ai/langgraph/blob/main/examples/rag/langgraph_crag_mistral.ipynb https://twitter.com/LangChainAI/status/1775569294241472810 2024/4/4 LangGraph, LLaMA3 and Elasticsearch 的實作 https://www.elastic.co/search-labs/blog/local-rag-agent-elasticsearch-langgraph-llama3 (2024/9)