以编程方式使用文档

本笔记本介绍了如何在 LangChain 中开始使用 Weaviate 向量存储，使用 langchain-weaviate package.

> Weaviate 是一个开源向量数据库。它允许您存储来自您喜爱的 ML 模型的数据对象和向量嵌入，并可无缝扩展到数十亿个数据对象。

要使用此集成，您需要运行一个 Weaviate 数据库实例。

最低版本

此模块需要 Weaviate 1.23.7 或更高版本。但是，我们建议您使用最新版本的 Weaviate。

连接到 Weaviate

在本笔记本中，我们假设您在 http://localhost:8080 上运行了本地 Weaviate 实例，并且 端口 50051 已开放以接收gRPC 流量。我们将使用以下方式连接到 Weaviate：

weaviate_client = weaviate.connect_to_local()

其他部署选项

Weaviate 可以 以多种不同方式部署 例如使用 Weaviate Cloud Services (WCS), Docker or Kubernetes.

如果您的 Weaviate 实例以其他方式部署， 在此处阅读更多 了解有关连接 Weaviate 的不同方法。您可以使用不同的 辅助函数 or 创建自定义实例.

> 请注意，您需要一个 v4 客户端 API，这将创建一个 weaviate.WeaviateClient object.

身份验证

Some Weaviate instances, such as those running on WCS, have authentication enabled, such as API key and/or username+password authentication.

阅读 客户端身份验证指南 获取更多信息，以及 深入的身份验证配置页面.

连接到现有集合（重复使用索引）

如果您已经在本地 Weaviate 实例中创建了集合，可以直接连接到它：

from langchain_weaviate import WeaviateVectorStore

store = WeaviateVectorStore(
    client=weaviate_client,
    index_name="Test",
    text_key="text",
)

安装

# install package
# pip install -Uqq langchain-weaviate
# pip install openai tiktoken langchain

环境设置

本笔记本通过 OpenAIEmbeddings使用 OpenAI API。我们建议获取 OpenAI API 密钥并将其作为环境变量导出，变量名为 OPENAI_API_KEY.

完成此操作后，您的 OpenAI API 密钥将被自动读取。如果您不熟悉环境变量，请在 Python os.environ 文档 or in 本指南.

用法

通过相似性查找对象

以下是一个示例，展示了如何通过相似性查找与查询相似的对象，从数据导入到查询 Weaviate 实例。

步骤 1：数据导入

首先，我们将创建要添加到 Weaviate 的数据，通过加载和分块长文本文件的内容。

from langchain_community.document_loaders import TextLoader
from langchain_openai import OpenAIEmbeddings
from langchain_text_splitters import CharacterTextSplitter
loader = TextLoader("state_of_the_union.txt")
documents = loader.load()
text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
docs = text_splitter.split_documents(documents)

embeddings = OpenAIEmbeddings()
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/langchain_core/_api/deprecation.py:117: LangChainDeprecationWarning: The class `langchain_community.embeddings.openai.OpenAIEmbeddings` was deprecated in langchain-community 0.1.0 and will be removed in 0.2.0. An updated version of the class exists in the langchain-openai package and should be used instead. To use it run `pip install -U langchain-openai` and import as `from langchain_openai import OpenAIEmbeddings`.
  warn_deprecated(

现在，我们可以导入数据。

为此，请连接到 Weaviate 实例并使用生成的 weaviate_client 对象。例如，我们可以按如下方式导入文档：

from langchain_weaviate.vectorstores import WeaviateVectorStore
weaviate_client = weaviate.connect_to_local()
db = WeaviateVectorStore.from_documents(docs, embeddings, client=weaviate_client)
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/pydantic/main.py:1024: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.6/migration/
  warnings.warn('The `dict` method is deprecated; use `model_dump` instead.', category=PydanticDeprecatedSince20)

第 2 步：执行搜索

我们现在可以执行相似性搜索。这将根据存储在 Weaviate 中的嵌入以及从查询文本生成的等效嵌入，返回与查询文本最相似的文档。

query = "What did the president say about Ketanji Brown Jackson"
docs = db.similarity_search(query)

# Print the first 100 characters of each result
for i, doc in enumerate(docs):
    print(f"\nDocument {i + 1}:")
    print(doc.page_content[:100] + "...")
Document 1:
Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Ac...

Document 2:
And so many families are living paycheck to paycheck, struggling to keep up with the rising cost of ...

Document 3:
Vice President Harris and I ran for office with a new economic vision for America.

Invest in Ameri...

Document 4:
A former top litigator in private practice. A former federal public defender. And from a family of p...

您还可以添加过滤器，这些过滤器将根据过滤条件包含或排除结果。（参见 更多过滤示例.)

from weaviate.classes.query import Filter

for filter_str in ["blah.txt", "state_of_the_union.txt"]:
    search_filter = Filter.by_property("source").equal(filter_str)
    filtered_search_results = db.similarity_search(query, filters=search_filter)
    print(len(filtered_search_results))
    if filter_str == "state_of_the_union.txt":
        assert len(filtered_search_results) > 0  # There should be at least one result
    else:
        assert len(filtered_search_results) == 0  # There should be no results
0
4

也可以提供 k，这是要返回的结果数量的上限。

search_filter = Filter.by_property("source").equal("state_of_the_union.txt")
filtered_search_results = db.similarity_search(query, filters=search_filter, k=3)
assert len(filtered_search_results) <= 3

量化结果相似度

您可以选择检索相关性"分数"。这是一个相对分数，表示特定搜索结果在搜索结果池中的好坏程度。

请注意，这是一个相对分数，意味着它不应被用于确定相关性的阈值。但是，它可以用于比较整个搜索结果集中不同搜索结果的相关性。

docs = db.similarity_search_with_score("country", k=5)

for doc in docs:
    print(f"{doc[1]:.3f}", ":", doc[0].page_content[:100] + "...")
0.935 : For that purpose we’ve mobilized American ground forces, air squadrons, and ship deployments to prot...
0.500 : And built the strongest, freest, and most prosperous nation the world has ever known.

Now is the h...
0.462 : If you travel 20 miles east of Columbus, Ohio, you’ll find 1,000 empty acres of land.

It won’t loo...
0.450 : And my report is this: the State of the Union is strong—because you, the American people, are strong...
0.442 : Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Ac...

搜索机制

similarity_search 使用 Weaviate 的 混合搜索.

混合搜索结合了向量搜索和关键词搜索，其中 alpha 作为向量搜索的权重。 similarity_search 函数允许您将其他参数作为 kwargs 传递。请参阅此 参考文档 了解可用参数。

您可以通过添加 alpha=0 来执行纯关键词搜索，如下所示：

docs = db.similarity_search(query, alpha=0)
docs[0]
Document(page_content='Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Act. And while you’re at it, pass the Disclose Act so Americans can know who is funding our elections. \n\nTonight, I’d like to honor someone who has dedicated his life to serve this country: Justice Stephen Breyer—an Army veteran, Constitutional scholar, and retiring Justice of the United States Supreme Court. Justice Breyer, thank you for your service. \n\nOne of the most serious constitutional responsibilities a President has is nominating someone to serve on the United States Supreme Court. \n\nAnd I did that 4 days ago, when I nominated Circuit Court of Appeals Judge Ketanji Brown Jackson. One of our nation’s top legal minds, who will continue Justice Breyer’s legacy of excellence.', metadata={'source': 'state_of_the_union.txt'})

持久化

通过 langchain-weaviate 添加的任何数据都将根据其配置在 Weaviate 中持久化。

例如，WCS 实例配置为无限期持久化数据，Docker 实例可以设置为在卷中持久化数据。了解更多关于 Weaviate 的持久化.

Multi-tenancy

Multi-tenancy 允许您在单个 Weaviate 实例中拥有大量具有相同集合配置的数据隔离集合。这非常适合多用户环境，例如构建 SaaS 应用程序，其中每个最终用户都将拥有自己的隔离数据集合。

要使用多租户，向量存储需要了解 tenant parameter.

添加任何数据时，提供 tenant 参数，如下所示。

db_with_mt = WeaviateVectorStore.from_documents(
    docs, embeddings, client=weaviate_client, tenant="Foo"
)
2024-Mar-26 03:40 PM - langchain_weaviate.vectorstores - INFO - Tenant Foo does not exist in index LangChain_30b9273d43b3492db4fb2aba2e0d6871. Creating tenant.

执行查询时，也提供 tenant 参数。

db_with_mt.similarity_search(query, tenant="Foo")
[Document(page_content='Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Act. And while you’re at it, pass the Disclose Act so Americans can know who is funding our elections. \n\nTonight, I’d like to honor someone who has dedicated his life to serve this country: Justice Stephen Breyer—an Army veteran, Constitutional scholar, and retiring Justice of the United States Supreme Court. Justice Breyer, thank you for your service. \n\nOne of the most serious constitutional responsibilities a President has is nominating someone to serve on the United States Supreme Court. \n\nAnd I did that 4 days ago, when I nominated Circuit Court of Appeals Judge Ketanji Brown Jackson. One of our nation’s top legal minds, who will continue Justice Breyer’s legacy of excellence.', metadata={'source': 'state_of_the_union.txt'}),
 Document(page_content='And so many families are living paycheck to paycheck, struggling to keep up with the rising cost of food, gas, housing, and so much more. \n\nI understand. \n\nI remember when my Dad had to leave our home in Scranton, Pennsylvania to find work. I grew up in a family where if the price of food went up, you felt it. \n\nThat’s why one of the first things I did as President was fight to pass the American Rescue Plan.  \n\nBecause people were hurting. We needed to act, and we did. \n\nFew pieces of legislation have done more in a critical moment in our history to lift us out of crisis. \n\nIt fueled our efforts to vaccinate the nation and combat COVID-19. It delivered immediate economic relief for tens of millions of Americans.  \n\nHelped put food on their table, keep a roof over their heads, and cut the cost of health insurance. \n\nAnd as my Dad used to say, it gave people a little breathing room.', metadata={'source': 'state_of_the_union.txt'}),
 Document(page_content='He and his Dad both have Type 1 diabetes, which means they need insulin every day. Insulin costs about $10 a vial to make.  \n\nBut drug companies charge families like Joshua and his Dad up to 30 times more. I spoke with Joshua’s mom. \n\nImagine what it’s like to look at your child who needs insulin and have no idea how you’re going to pay for it.  \n\nWhat it does to your dignity, your ability to look your child in the eye, to be the parent you expect to be. \n\nJoshua is here with us tonight. Yesterday was his birthday. Happy birthday, buddy.  \n\nFor Joshua, and for the 200,000 other young people with Type 1 diabetes, let’s cap the cost of insulin at $35 a month so everyone can afford it.  \n\nDrug companies will still do very well. And while we’re at it let Medicare negotiate lower prices for prescription drugs, like the VA already does.', metadata={'source': 'state_of_the_union.txt'}),
 Document(page_content='Putin’s latest attack on Ukraine was premeditated and unprovoked. \n\nHe rejected repeated efforts at diplomacy. \n\nHe thought the West and NATO wouldn’t respond. And he thought he could divide us at home. Putin was wrong. We were ready.  Here is what we did.   \n\nWe prepared extensively and carefully. \n\nWe spent months building a coalition of other freedom-loving nations from Europe and the Americas to Asia and Africa to confront Putin. \n\nI spent countless hours unifying our European allies. We shared with the world in advance what we knew Putin was planning and precisely how he would try to falsely justify his aggression.  \n\nWe countered Russia’s lies with truth.   \n\nAnd now that he has acted the free world is holding him accountable. \n\nAlong with twenty-seven members of the European Union including France, Germany, Italy, as well as countries like the United Kingdom, Canada, Japan, Korea, Australia, New Zealand, and many others, even Switzerland.', metadata={'source': 'state_of_the_union.txt'})]

检索器选项

Weaviate 也可以用作检索器

最大边际相关性搜索 (MMR)

除了在检索器对象中使用相似性搜索外，您还可以使用 mmr.

retriever = db.as_retriever(search_type="mmr")
retriever.invoke(query)[0]
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/pydantic/main.py:1024: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.6/migration/
  warnings.warn('The `dict` method is deprecated; use `model_dump` instead.', category=PydanticDeprecatedSince20)
Document(page_content='Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Act. And while you’re at it, pass the Disclose Act so Americans can know who is funding our elections. \n\nTonight, I’d like to honor someone who has dedicated his life to serve this country: Justice Stephen Breyer—an Army veteran, Constitutional scholar, and retiring Justice of the United States Supreme Court. Justice Breyer, thank you for your service. \n\nOne of the most serious constitutional responsibilities a President has is nominating someone to serve on the United States Supreme Court. \n\nAnd I did that 4 days ago, when I nominated Circuit Court of Appeals Judge Ketanji Brown Jackson. One of our nation’s top legal minds, who will continue Justice Breyer’s legacy of excellence.', metadata={'source': 'state_of_the_union.txt'})

与 LangChain 配合使用

大型语言模型 (LLM) 的一个已知局限性是，它们的训练数据可能已过时，或者不包含您需要的特定领域知识。

请看下面的示例：

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=0)
llm.predict("What did the president say about Justice Breyer")
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/langchain_core/_api/deprecation.py:117: LangChainDeprecationWarning: The class `langchain_community.chat_models.openai.ChatOpenAI` was deprecated in langchain-community 0.0.10 and will be removed in 0.2.0. An updated version of the class exists in the langchain-openai package and should be used instead. To use it run `pip install -U langchain-openai` and import as `from langchain_openai import ChatOpenAI`.
  warn_deprecated(
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/langchain_core/_api/deprecation.py:117: LangChainDeprecationWarning: The function `predict` was deprecated in LangChain 0.1.7 and will be removed in 0.2.0. Use invoke instead.
  warn_deprecated(
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/pydantic/main.py:1024: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.6/migration/
  warnings.warn('The `dict` method is deprecated; use `model_dump` instead.', category=PydanticDeprecatedSince20)
"I'm sorry, I cannot provide real-time information as my responses are generated based on a mixture of licensed data, data created by human trainers, and publicly available data. The last update was in October 2021."

向量存储通过提供存储和检索相关信息的方式来补充 LLM。这使您能够将 LLM 的推理和语言能力与向量存储检索相关信息的能力相结合，从而充分发挥两者的优势。

将LLM与向量存储结合的两个知名应用是：

  • - 问答
  • - 检索增强生成 (RAG)

带来源的问答

LangChain中的问答可以通过使用向量存储来增强。让我们看看如何实现。

本节使用 RetrievalQAWithSourcesChain，它从索引中查找文档。

首先，我们将再次对文本进行分块，并将其导入到Weaviate向量存储中。

from langchain_classic.chains import RetrievalQAWithSourcesChain
from langchain_openai import OpenAI
with open("state_of_the_union.txt") as f:
    state_of_the_union = f.read()
text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
texts = text_splitter.split_text(state_of_the_union)
docsearch = WeaviateVectorStore.from_texts(
    texts,
    embeddings,
    client=weaviate_client,
    metadatas=[{"source": f"{i}-pl"} for i in range(len(texts))],
)

现在我们可以构建链，并指定检索器：

chain = RetrievalQAWithSourcesChain.from_chain_type(
    OpenAI(temperature=0), chain_type="stuff", retriever=docsearch.as_retriever()
)
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/langchain_core/_api/deprecation.py:117: LangChainDeprecationWarning: The class `langchain_community.llms.openai.OpenAI` was deprecated in langchain-community 0.0.10 and will be removed in 0.2.0. An updated version of the class exists in the langchain-openai package and should be used instead. To use it run `pip install -U langchain-openai` and import as `from langchain_openai import OpenAI`.
  warn_deprecated(

然后运行链来提问：

chain(
    {"question": "What did the president say about Justice Breyer"},
    return_only_outputs=True,
)
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/langchain_core/_api/deprecation.py:117: LangChainDeprecationWarning: The function `__call__` was deprecated in LangChain 0.1.0 and will be removed in 0.2.0. Use invoke instead.
  warn_deprecated(
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/pydantic/main.py:1024: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.6/migration/
  warnings.warn('The `dict` method is deprecated; use `model_dump` instead.', category=PydanticDeprecatedSince20)
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/pydantic/main.py:1024: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.6/migration/
  warnings.warn('The `dict` method is deprecated; use `model_dump` instead.', category=PydanticDeprecatedSince20)
{'answer': ' The president thanked Justice Stephen Breyer for his service and announced his nomination of Judge Ketanji Brown Jackson to the Supreme Court.\n',
 'sources': '31-pl'}

检索增强生成

将LLM与向量存储结合的另一个非常流行的应用是检索增强生成（RAG）。这是一种使用检索器从向量存储中查找相关信息，然后使用LLM根据检索到的数据和提示词提供输出的技术。

我们从类似的设置开始：

with open("state_of_the_union.txt") as f:
    state_of_the_union = f.read()
text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
texts = text_splitter.split_text(state_of_the_union)
docsearch = WeaviateVectorStore.from_texts(
    texts,
    embeddings,
    client=weaviate_client,
    metadatas=[{"source": f"{i}-pl"} for i in range(len(texts))],
)

retriever = docsearch.as_retriever()

我们需要为RAG模型构建一个模板，以便将检索到的信息填充到模板中。

from langchain_core.prompts import ChatPromptTemplate

template = """You are an assistant for question-answering tasks. Use the following pieces of retrieved context to answer the question. If you don't know the answer, just say that you don't know. Use three sentences maximum and keep the answer concise.
Question: {question}
Context: {context}
Answer:
"""
prompt = ChatPromptTemplate.from_template(template)

print(prompt)
input_variables=['context', 'question'] messages=[HumanMessagePromptTemplate(prompt=PromptTemplate(input_variables=['context', 'question'], template="You are an assistant for question-answering tasks. Use the following pieces of retrieved context to answer the question. If you don't know the answer, just say that you don't know. Use three sentences maximum and keep the answer concise.\nQuestion: {question}\nContext: {context}\nAnswer:\n"))]
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=0)

运行单元格后，我们得到了非常相似的输出。

from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import RunnablePassthrough

rag_chain = (
    {"context": retriever, "question": RunnablePassthrough()}
    | prompt
    | llm
    | StrOutputParser()
)

rag_chain.invoke("What did the president say about Justice Breyer")
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/pydantic/main.py:1024: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.6/migration/
  warnings.warn('The `dict` method is deprecated; use `model_dump` instead.', category=PydanticDeprecatedSince20)
/workspaces/langchain-weaviate/.venv/lib/python3.12/site-packages/pydantic/main.py:1024: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.6/migration/
  warnings.warn('The `dict` method is deprecated; use `model_dump` instead.', category=PydanticDeprecatedSince20)
"The president honored Justice Stephen Breyer for his service to the country as an Army veteran, Constitutional scholar, and retiring Justice of the United States Supreme Court. The president also mentioned nominating Circuit Court of Appeals Judge Ketanji Brown Jackson to continue Justice Breyer's legacy of excellence. The president expressed gratitude towards Justice Breyer and highlighted the importance of nominating someone to serve on the United States Supreme Court."

请注意，由于模板需要由您自己构建，您可以根据自己的需求进行定制。

总结与资源

Weaviate是一个可扩展的、生产就绪的向量存储。

此集成允许Weaviate与LangChain结合使用，通过强大的数据存储来增强大型语言模型的能力。其可扩展性和生产就绪性使其成为LangChain应用程序向量存储的绝佳选择，并且将缩短您的生产时间。