agi

langchain based agi

RAG

• history_chain = RunnableLambda(self._enter_history, self._aenter_history).with_config(run_name="load_history")
• history_chain = RunnablePassthrough.assign(**{"chat_history": history_chain}).with_config(run_name="insert_history")
• retrieval_docs = (lambda x: x["input"]) | retriever
• retriever = (lambda x: x["input"]) | retriever or prompt | llm | StrOutputParser() | retriever, run_name="chat_retriever_chain"
• context=retrieval_docs.with_config(run_name="retrieve_documents")
• "context": format_docs run_name="format_inputs"
• answer = "context" | prompt | llm | _output_parser run_name="stuff_documents_chain"