这将帮助您开始使用 LangChain 进行 Nomic 嵌入模型学习。若要获取详细文档 NomicEmbeddings 功能和配置选项，请参阅 API 参考.

概述

集成详情

设置

To access Nomic embedding models you'll need to create a/an Nomic account, get an API key, and install the langchain-nomic 集成包。

凭证

前往 https://atlas.nomic.ai/ 注册 Nomic 并生成 API 密钥。完成后设置 NOMIC_API_KEY 环境变量：

if not os.getenv("NOMIC_API_KEY"):
    os.environ["NOMIC_API_KEY"] = getpass.getpass("Enter your Nomic API key: ")

要启用模型调用的自动跟踪，请设置您的 LangSmith API 密钥：

os.environ["LANGSMITH_TRACING"] = "true"
os.environ["LANGSMITH_API_KEY"] = getpass.getpass("Enter your LangSmith API key: ")

安装

LangChain Nomic 集成位于 langchain-nomic package:

pip install -qU langchain-nomic

实例化

现在我们可以实例化模型对象并生成聊天补全：

from langchain_nomic import NomicEmbeddings

embeddings = NomicEmbeddings(
    model="nomic-embed-text-v1.5",
    # dimensionality=256,
    # Nomic's `nomic-embed-text-v1.5` model was [trained with Matryoshka learning](https://blog.nomic.ai/posts/nomic-embed-matryoshka)
    # to enable variable-length embeddings with a single model.
    # This means that you can specify the dimensionality of the embeddings at inference time.
    # The model supports dimensionality from 64 to 768.
    # inference_mode="remote",
    # One of `remote`, `local` (Embed4All), or `dynamic` (automatic). Defaults to `remote`.
    # api_key=... , # if using remote inference,
    # device="cpu",
    # The device to use for local embeddings. Choices include
    # `cpu`, `gpu`, `nvidia`, `amd`, or a specific device name. See
    # the docstring for `GPT4All.__init__` for more info. Typically
    # defaults to CPU. Do not use on macOS.
)

索引和检索

嵌入模型常用于检索增强生成（RAG）流程，既用于索引数据，也用于后续检索。如需更详细的说明，请参阅我们的 RAG 教程.

下方演示如何使用 embeddings 对象对数据进行索引和检索。在此示例中，我们将在 InMemoryVectorStore.

# Create a vector store with a sample text
from langchain_core.vectorstores import InMemoryVectorStore

text = "LangChain is the framework for building context-aware reasoning applications"

vectorstore = InMemoryVectorStore.from_texts(
    [text],
    embedding=embeddings,
)

# Use the vectorstore as a retriever
retriever = vectorstore.as_retriever()

# Retrieve the most similar text
retrieved_documents = retriever.invoke("What is LangChain?")

# show the retrieved document's content
retrieved_documents[0].page_content

'LangChain is the framework for building context-aware reasoning applications'

直接使用

在底层，向量存储和检索器的实现调用 embeddings.embed_documents(...) 和 embeddings.embed_query(...) 为中使用的文本创建嵌入 from_texts 和检索 invoke 操作。

您可以直接调用这些方法来获取适合您用例的嵌入。

嵌入单个文本

您可以使用以下方式嵌入单个文本或文档 embed_query:

single_vector = embeddings.embed_query(text)
print(str(single_vector)[:100])  # Show the first 100 characters of the vector

[0.024642944, 0.029083252, -0.14013672, -0.09082031, 0.058898926, -0.07489014, -0.0138168335, 0.0037

嵌入多个文本

您可以使用以下方式嵌入多个文本 embed_documents:

text2 = (
    "LangGraph is a library for building stateful, multi-actor applications with LLMs"
)
two_vectors = embeddings.embed_documents([text, text2])
for vector in two_vectors:
    print(str(vector)[:100])  # Show the first 100 characters of the vector

[0.012771606, 0.023727417, -0.12365723, -0.083740234, 0.06530762, -0.07110596, -0.021896362, -0.0068
[-0.019058228, 0.04058838, -0.15222168, -0.06842041, -0.012130737, -0.07128906, -0.04534912, 0.00522

API 参考

有关详细文档 NomicEmbeddings 功能和配置选项，请参阅 API 参考.