从无状态到类人思维：为什么记忆是 Agentic AI 的关键？ 原创

from langchain.document_loaders import PyPDFLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.vectorstores import FAISS
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.chains import RetrievalQA
from langchain.chat_models import ChatOpenAI

# 1. 加载 PDF
loader = PyPDFLoader("employee_handbook.pdf")
documents = loader.load()

# 2. 分块
splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=100)
docs = splitter.split_documents(documents)

# 3. 向量化并存储
embeddings = OpenAIEmbeddings()
vectorstore = FAISS.from_documents(docs, embeddings)

# 4. 构建 RAG 检索链
retriever = vectorstore.as_retriever()
qa = RetrievalQA.from_chain_type(
    llm=ChatOpenAI(),
    retriever=retriever,
    chain_type="stuff"
)

# 5. 提问
query = "公司试用期员工可以请年假吗？"
print(qa.run(query))

import os
from langchain.embeddings import OpenAIEmbeddings
from langchain.vectorstores import Chroma
from langchain.text_splitter import RecursiveCharacterTextSplitter

# 加载本地 API 文档
with open("api_doc.md", "r") as f:
    text = f.read()

splitter = RecursiveCharacterTextSplitter(chunk_size=800, chunk_overlap=100)
docs = splitter.create_documents([text])

# 存储到 Chroma
embeddings = OpenAIEmbeddings()
db = Chroma.from_documents(docs, embeddings, persist_directory="chroma_store")
db.persist()

# 检索
query = "用户登录接口需要哪些参数？"
docs = db.similarity_search(query, k=2)
for d in docs:
    print(d.page_content[:200])  # 打印部分内容

from pymilvus import connections, Collection, utility, FieldSchema, CollectionSchema, DataType
from sentence_transformers import SentenceTransformer
import numpy as np

# 1. 连接 Milvus
connections.connect("default", host="localhost", port="19530")

# 2. 定义 schema
fields = [
    FieldSchema(name="id", dtype=DataType.INT64, is_primary=True, auto_id=True),
    FieldSchema(name="embedding", dtype=DataType.FLOAT_VECTOR, dim=384)
]
schema = CollectionSchema(fields, descriptinotallow="log search")
collection = Collection("log_data", schema)

# 3. 加载日志数据并向量化
model = SentenceTransformer('all-MiniLM-L6-v2')
logs = [
    "ERROR: Database connection failed at 10:23",
    "INFO: User login success user_id=123",
    "WARN: High memory usage detected on server-2"
]
embeddings = model.encode(logs).tolist()

# 插入数据
collection.insert([embeddings])
collection.load()

# 4. 查询
query = "数据库无法连接"
q_emb = model.encode([query]).tolist()
res = collection.search(q_emb, anns_field="embedding", param={"metric_type": "IP", "params": {"nprobe": 10}}, limit=2)

for hit in res[0]:
    print("相关日志：", logs[hit.id])

import weaviate
from sentence_transformers import SentenceTransformer

# 连接 Weaviate
client = weaviate.Client("http://localhost:8080")
model = SentenceTransformer("all-MiniLM-L6-v2")

faq_data = [
    {"q": "如何退货？", "a": "登录订单中心，选择退货申请"},
    {"q": "运费如何计算？", "a": "运费根据重量和地区自动计算"}
]

# 向量化并写入
for item in faq_data:
    emb = model.encode(item["q"]).tolist()
    client.data_object.create({
        "question": item["q"],
        "answer": item["a"],
        "embedding": emb
    }, "FAQ")

# 用户提问
query = "我要退货怎么操作？"
q_emb = model.encode(query).tolist()
res = client.query.get("FAQ", ["question", "answer"]).with_near_vector({"vector": q_emb}).with_limit(1).do()

print("客服答案：", res["data"]["Get"]["FAQ"][0]["answer"])