Skip to main content

ThirdAI NeuralDB

NeuralDB 是由 ThirdAI 开发的一个对CPU友好且可微调的向量存储。

初始化

有两种初始化方法:

  • 从头开始:基本模型
  • 从检查点:加载之前保存的模型

对于所有以下初始化方法,如果设置了 THIRDAI_KEY 环境变量,则可以省略 thirdai_key 参数。

可以在 https://www.thirdai.com/try-bolt/ 获取 ThirdAI API 密钥。

您需要使用 pip install -qU langchain-community 安装 langchain-community 才能使用此集成。

<!--IMPORTS:[{"imported": "NeuralDBVectorStore", "source": "langchain_community.vectorstores", "docs": "https://python.langchain.com/api_reference/community/vectorstores/langchain_community.vectorstores.thirdai_neuraldb.NeuralDBVectorStore.html", "title": "ThirdAI NeuralDB"}]-->
from langchain_community.vectorstores import NeuralDBVectorStore

# From scratch
vectorstore = NeuralDBVectorStore.from_scratch(thirdai_key="your-thirdai-key")

# From checkpoint
vectorstore = NeuralDBVectorStore.from_checkpoint(
# Path to a NeuralDB checkpoint. For example, if you call
# vectorstore.save("/path/to/checkpoint.ndb") in one script, then you can
# call NeuralDBVectorStore.from_checkpoint("/path/to/checkpoint.ndb") in
# another script to load the saved model.
checkpoint="/path/to/checkpoint.ndb",
thirdai_key="your-thirdai-key",
)

插入文档源

vectorstore.insert(
# If you have PDF, DOCX, or CSV files, you can directly pass the paths to the documents
sources=["/path/to/doc.pdf", "/path/to/doc.docx", "/path/to/doc.csv"],
# When True this means that the underlying model in the NeuralDB will
# undergo unsupervised pretraining on the inserted files. Defaults to True.
train=True,
# Much faster insertion with a slight drop in performance. Defaults to True.
fast_mode=True,
)

from thirdai import neural_db as ndb

vectorstore.insert(
# If you have files in other formats, or prefer to configure how
# your files are parsed, then you can pass in NeuralDB document objects
# like this.
sources=[
ndb.PDF(
"/path/to/doc.pdf",
version="v2",
chunk_size=100,
metadata={"published": 2022},
),
ndb.Unstructured("/path/to/deck.pptx"),
]
)

相似性搜索

要查询向量存储,您可以使用标准的 LangChain 向量存储方法 similarity_search,该方法返回一个 LangChain 文档对象的列表。每个文档对象代表来自索引文件的一段文本。例如,它可能包含来自某个索引 PDF 文件的段落。除了文本之外,文档的元数据字段还包含信息,例如文档的 ID、该文档的来源(来自哪个文件)以及文档的得分。

# This returns a list of LangChain Document objects
documents = vectorstore.similarity_search("query", k=10)

微调

NeuralDBVectorStore 可以根据用户行为和特定领域知识进行微调。可以通过两种方式进行微调:

  1. 关联:向量存储将源短语与目标短语关联。当向量存储看到源短语时,它还会考虑与目标短语相关的结果。
  2. 赞成:向量存储会提高特定查询的文档得分。这在您想要根据用户行为微调向量存储时非常有用。例如,如果用户搜索“汽车是如何制造的”并喜欢返回的 ID 为 52 的文档,那么我们可以为查询“汽车是如何制造的”对 ID 为 52 的文档进行赞成。
vectorstore.associate(source="source phrase", target="target phrase")
vectorstore.associate_batch(
[
("source phrase 1", "target phrase 1"),
("source phrase 2", "target phrase 2"),
]
)

vectorstore.upvote(query="how is a car manufactured", document_id=52)
vectorstore.upvote_batch(
[
("query 1", 52),
("query 2", 20),
]
)

相关


Was this page helpful?


You can also leave detailed feedback on GitHub.

扫我,入群扫我,找书