- [ ] https://twitter.com/llama_index/status/1708638824853823893 10/2 - [ ] https://twitter.com/clusteredbytes/status/1708637770762031286 10/2 - [ ] https://twitter.com/llama_index/status/1712129914386993295 10/11 - [ ] https://docs.llamaindex.ai/en/stable/examples/agent/multi_document_agents.html - [ ] https://docs.llamaindex.ai/en/latest/examples/agent/multi_document_agents-v1.html * https://blog.llamaindex.ai/agentic-rag-with-llamaindex-2721b8a49ff6 又命名成 Agentic RAG (2024/1/31) * https://cloud.tencent.com/developer/article/2353463?areaId=106001 * 不同文件拆分不同的 Retriever，包裹成 Tool * 然後用 function calling 去選要用哪一個 tool * https://twitter.com/jerryjliu0/status/1728196122496360683 * 除了 Multi-Document Agents 還有比較簡單的: * Document Summary Index: https://docs.llamaindex.ai/en/stable/examples/index_structs/doc_summary/DocSummary.html 針對文件摘要做索引，然後挑出那個文件做 retriever * 在 langchain 裡放在 [[Multi-Vector Retriever]] * [[Sub Question Query]] engine: https://docs.llamaindex.ai/en/stable/examples/query_engine/sub_question_query_engine.html * Auto-Retrieval https://docs.llamaindex.ai/en/latest/examples/vector_stores/chroma_auto_retriever.html 也就是 [[Self-Querying Retriever]] * 案例 https://blog.lancedb.com/multi-document-agentic-rag-a-walkthrough/ ## OpenAI Assistant Advanced Retrieval Cookbook 使用 OpenAI Assistant API 的範例 * https://twitter.com/jerryjliu0/status/1723508609315996144 * https://twitter.com/llama_index/status/1723430852515037495 * https://github.com/run-llama/llama_index/blob/main/docs/examples/agent/openai_assistant_query_cookbook.ipynb