大模型---LangGraph的循环与迭代

目录1.LangGraph的循环2.典型的迭代方式3.循环退出条件1.LangGraph的循环LangGraph的循环可以分为3个部分：① State：保存迭代过程中的状态，② node：每轮执行的逻辑，③conditional edge：判断继续循环还是结束。最简单的循环方法就是固定迭代次数，即让一个节点不断更新状态，直到达到最大迭代次数。from typing import TypedDict, Literal from langgraph.graph import StateGraph, START, END class State(TypedDict): count: int max_iter: int history: list[str] def iterate(state: State) - State: count = state["count"] + 1 return { "count": count, "max_iter": state["max_iter"], "history": state["history"] + [f"第 {count} 次迭代"], } def should_continue(state: State) - Literal["continue", "stop"]: if state["count"] state["max_iter"]: return "continue" return "stop" builder = StateGraph(State) builder.add_node("iterate", iterate) builder.add_edge(START, "iterate") builder.add_conditional_edges( "iterate