Skip to main content

Objective

This notebook covers how to get started with the Objective vector store.

Objective is an engine for building modern, AI-native search.

Key features include:

  • Semantic, automatically understanding natural language queries, synonyms, typos, and multiple languages.
  • Hybrid, capable of exact matching and approximate matching in one API without requiring you to develop lexical / keyword matching as well as vector-based approximate nearest neighbors (ANN) engines.
  • Deep, surfacing relevant Highlights from different media by going inside the content and pulling out relevant bits.

Setup​

To use Objective be sure to install the latest langchain-community with pip install -qU langchain-community.

Credentials​

Next you'll need to sign up and copy your API key.

The easiest step to configure your key is to store it in the OBJECTIVE_KEY environment variable. Otherwise you may use it directly below

import os

objective_key = os.getenv("OBJECTIVE_KEY")

Initialization​

from langchain_community.vectorstores import Objective

vector_store = Objective(objective_key)
API Reference:Objective

Manage vector store​

Add items to vector store​

from langchain_core.documents import Document

document_1 = Document(page_content="foo", metadata={"source": "https://example.com"})

document_2 = Document(page_content="bar", metadata={"source": "https://example.com"})

document_3 = Document(page_content="baz", metadata={"source": "https://example.com"})

documents = [document_1, document_2, document_3]

vector_store.add_documents(documents=documents, ids=["1", "2", "3"])
API Reference:Document

Update items in vector store​

updated_document = Document(
    id="1", page_content="qux", metadata={"source": "https://another-example.com"}
)

vector_store.upsert(updated_document)

Delete items from vector store​

vector_store.delete(ids=["3"])

Create a searchable index​

index_id = vector_store.create_index()
print(f"Created index ID: {index_id}")
status = vector_store.index_status(index_id)
# status returns an object with UPLOADED, READY, ERROR, and PROCESSING document counts
if status["UPLOADED"] != status["READY"]:
    print(f"Not all documents processed yet, please retry: {status}")
else:
    print("Ready to proceed")

Query vector store​

Once your vector store has been created and the relevant documents have been added you will most likely wish to query it during the running of your chain or agent.

Query directly​

results = vector_store.search(query="thud", search_type="similarity", index_id=index_id)
for doc in results:
    print(doc.page_content)

Query by turning into retriever​

You can also transform the vector store into a retriever for easier usage in your chains.

retriever = vector_store.as_retriever(search_type="mmr", search_kwargs={"k": 1})
retriever.invoke("thud")

Chain usage​

The code below shows how to use the vector store as a retriever in a simple RAG chain:

from langchain_openai import ChatOpenAI
from langchain import hub
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import RunnablePassthrough


llm = ChatOpenAI(model="gpt-3.5-turbo-0125")

prompt = hub.pull("rlm/rag-prompt")


def format_docs(docs):
    return "\n\n".join(doc.page_content for doc in docs)


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

rag_chain.invoke("thud")

API reference​

Was this page helpful?


You can also leave detailed feedback on GitHub.