使用 Milvus 作為 LangChain 向量儲存庫

本筆記展示如何使用Milvus作為LangChain 向量儲存庫的相關功能。

安裝

您需要安裝langchain-milvus 及其他必要的相依性。

$ pip install -qU langchain-milvus milvus-lite langchain-openai

最新版本的 pymilvus 隨附一個本機向量資料庫 Milvus Lite，很適合做為原型。如果您有大規模的資料，例如超過一百萬份的文件，我們建議您在docker 或 kubernetes 上架設效能更高的 Milvus 伺服器。

初始化

from langchain_openai import OpenAIEmbeddings

embeddings = OpenAIEmbeddings(model="text-embedding-3-large")

from langchain_milvus import Milvus

# The easiest way is to use Milvus Lite where everything is stored in a local file.
# If you have a Milvus server you can use the server URI such as "http://localhost:19530".
URI = "./milvus_example.db"

vector_store = Milvus(
    embedding_function=embeddings,
    connection_args={"uri": URI},
)

使用 Milvus Collections 區隔資料

您可以在同一個 Milvus 實例中，將不同的不相關文件存放在不同的集合中，以維護上下文。

以下是如何從文件建立一個新的向量儲存集合：

from langchain_core.documents import Document

vector_store_saved = Milvus.from_documents(
    [Document(page_content="foo!")],
    embeddings,
    collection_name="langchain_example",
    connection_args={"uri": URI},
)

以下是如何擷取儲存的集合

vector_store_loaded = Milvus(
    embeddings,
    connection_args={"uri": URI},
    collection_name="langchain_example",
)

管理向量儲存

一旦你創建了你的向量商店，我們可以通過添加和刪除不同的項目與它互動。

新增項目至向量儲存

我們可以使用add_documents 功能將項目新增至向量儲存空間。

from uuid import uuid4

from langchain_core.documents import Document

document_1 = Document(
    page_content="I had chocalate chip pancakes and scrambled eggs for breakfast this morning.",
    metadata={"source": "tweet"},
)

document_2 = Document(
    page_content="The weather forecast for tomorrow is cloudy and overcast, with a high of 62 degrees.",
    metadata={"source": "news"},
)

document_3 = Document(
    page_content="Building an exciting new project with LangChain - come check it out!",
    metadata={"source": "tweet"},
)

document_4 = Document(
    page_content="Robbers broke into the city bank and stole $1 million in cash.",
    metadata={"source": "news"},
)

document_5 = Document(
    page_content="Wow! That was an amazing movie. I can't wait to see it again.",
    metadata={"source": "tweet"},
)

document_6 = Document(
    page_content="Is the new iPhone worth the price? Read this review to find out.",
    metadata={"source": "website"},
)

document_7 = Document(
    page_content="The top 10 soccer players in the world right now.",
    metadata={"source": "website"},
)

document_8 = Document(
    page_content="LangGraph is the best framework for building stateful, agentic applications!",
    metadata={"source": "tweet"},
)

document_9 = Document(
    page_content="The stock market is down 500 points today due to fears of a recession.",
    metadata={"source": "news"},
)

document_10 = Document(
    page_content="I have a bad feeling I am going to get deleted :(",
    metadata={"source": "tweet"},
)

documents = [
    document_1,
    document_2,
    document_3,
    document_4,
    document_5,
    document_6,
    document_7,
    document_8,
    document_9,
    document_10,
]
uuids = [str(uuid4()) for _ in range(len(documents))]

vector_store.add_documents(documents=documents, ids=uuids)

['31915e2d-55fd-4bfb-ae08-d441252b8e08',
 'dbf6560a-1487-4a6e-8797-245d57874f5b',
 'e991a253-5f37-46ae-850a-82a660e33013',
 '2818c051-5a1a-44cb-9deb-aaaac709f616',
 '91c7ef07-26d1-4319-b48c-9261df9ce8d7',
 'fb258085-6400-4cd7-aa92-fc5e32ca243e',
 'ffea9a9f-460d-4d8d-ba07-c45e9cfa1e33',
 'eb149e29-239a-4e2c-9f99-751cb7207abf',
 '119d4a42-fd6b-433d-842b-1e0be5df81e5',
 '5b099eb0-98fe-40a3-b13a-300c10250960']

從向量庫中刪除項目

vector_store.delete(ids=[uuids[-1]])

True

查詢向量儲存

一旦您的向量儲存庫建立完成，並加入相關文件後，您很可能希望在連鎖或代理程式執行期間查詢它。

直接查詢

相似性搜尋

執行簡單的類似性搜尋並過濾元資料的方法如下：

results = vector_store.similarity_search(
    "LangChain provides abstractions to make working with LLMs easy",
    k=2,
    expr='source == "tweet"',
    # param=...  # Search params for the index type
)
for res in results:
    print(f"* {res.page_content} [{res.metadata}]")

WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
I0000 00:00:1761298048.354308 7886403 fork_posix.cc:71] Other threads are currently calling into gRPC, skipping fork() handlers


* Building an exciting new project with LangChain - come check it out! [{'source': 'tweet', 'pk': 'e991a253-5f37-46ae-850a-82a660e33013'}]
* LangGraph is the best framework for building stateful, agentic applications! [{'source': 'tweet', 'pk': 'eb149e29-239a-4e2c-9f99-751cb7207abf'}]

使用分數進行相似性搜尋

您也可以使用分數進行搜尋：

results = vector_store.similarity_search_with_score(
    "Will it be hot tomorrow?", k=1, expr='source == "news"'
)
for res, score in results:
    print(f"* [SIM={score:3f}] {res.page_content} [{res.metadata}]")

* [SIM=0.893776] The weather forecast for tomorrow is cloudy and overcast, with a high of 62 degrees. [{'source': 'news', 'pk': 'dbf6560a-1487-4a6e-8797-245d57874f5b'}]

如需使用Milvus 向量商店時可用的所有搜尋選項的完整清單，您可以造訪API 參考資料。

透過轉換成retriever進行查詢

您也可以將向量儲存轉換成retriever，以便在您的鏈中更容易使用。

retriever = vector_store.as_retriever(search_type="mmr", search_kwargs={"k": 1})
retriever.invoke("Stealing from the bank is a crime", expr='source == "news"')

I0000 00:00:1761298049.275354 7886403 fork_posix.cc:71] Other threads are currently calling into gRPC, skipping fork() handlers





[Document(metadata={'source': 'news', 'pk': '2818c051-5a1a-44cb-9deb-aaaac709f616'}, page_content='Robbers broke into the city bank and stole $1 million in cash.')]

擷取增強世代的使用方式

有關如何使用此向量儲存器進行retrieval-augmented generation (RAG)，請參閱此RAG 指南。

按使用者擷取

在建立擷取應用程式時，您通常必須以多位使用者為考量。這表示您可能不只為一個使用者儲存資料，而是為許多不同的使用者儲存資料，而且他們應該無法看到彼此的資料。

Milvus 建議使用partition_key來實現多重租用，這裡是一個範例。

現在 Milvus Lite 沒有分區鑰匙的功能，如果你想使用它，你需要從docker 或 kubernetes 啟動 Milvus 伺服器。

from langchain_core.documents import Document

docs = [
    Document(page_content="i worked at kensho", metadata={"namespace": "harrison"}),
    Document(page_content="i worked at facebook", metadata={"namespace": "ankush"}),
]
vectorstore = Milvus.from_documents(
    docs,
    embeddings,
    collection_name="partitioned_collection",  # Use a different collection name
    connection_args={"uri": URI},
    # drop_old=True,
    partition_key_field="namespace",  # Use the "namespace" field as the partition key
)

若要使用分割區金鑰進行搜尋，您應該在搜尋請求的布林表達式中包含下列任一項：

search_kwargs={"expr": '<partition_key> == "xxxx"'}

search_kwargs={"expr": '<partition_key> == in ["xxx", "xxx"]'}

請以指定為分割區金鑰的欄位名稱取代<partition_key> 。

Milvus 會根據指定的分割區金鑰變更為分割區，根據分割區金鑰過濾實體，並在過濾的實體中進行搜尋。

# This will only get documents for Ankush
vectorstore.as_retriever(search_kwargs={"expr": 'namespace == "ankush"'}).invoke(
    "where did i work?"
)

[Document(metadata={'namespace': 'ankush', 'pk': 460829372217788296}, page_content='i worked at facebook')]

# This will only get documents for Harrison
vectorstore.as_retriever(search_kwargs={"expr": 'namespace == "harrison"'}).invoke(
    "where did i work?"
)

[Document(metadata={'namespace': 'harrison', 'pk': 460829372217788295}, page_content='i worked at kensho')]

API 參考

如需詳細說明文件，請前往 API 參考： https://reference.langchain.com/python/integrations/langchain_milvus/