AI21Embeddings

This will help you get started with AI21 embedding models using LangChain. For detailed documentation on AI21Embeddings features and configuration options, please refer to the API reference.

Overview

Integration details

Provider	Package
AI21	langchain-ai21

Setup

To access AI21 embedding models you'll need to create an AI21 account, get an API key, and install the langchain-ai21 integration package.

Credentials

Head to https://docs.ai21.com/ to sign up to AI21 and generate an API key. Once you've done this set the AI21_API_KEY environment variable:

import getpass
import os

if not os.getenv("AI21_API_KEY"):
    os.environ["AI21_API_KEY"] = getpass.getpass("Enter your AI21 API key: ")

If you want to get automated tracing of your model calls you can also set your LangSmith API key by uncommenting below:

# os.environ["LANGCHAIN_TRACING_V2"] = "true"
# os.environ["LANGCHAIN_API_KEY"] = getpass.getpass("Enter your LangSmith API key: ")

Installation

The LangChain AI21 integration lives in the langchain-ai21 package:

%pip install -qU langchain-ai21

Instantiation

Now we can instantiate our model object and generate chat completions:

from langchain_ai21 import AI21Embeddings

embeddings = AI21Embeddings(
    # Can optionally increase or decrease the batch_size
    # to improve latency.
    # Use larger batch sizes with smaller documents, and
    # smaller batch sizes with larger documents.
    # batch_size=256,
)

Indexing and Retrieval

Embedding models are often used in retrieval-augmented generation (RAG) flows, both as part of indexing data as well as later retrieving it. For more detailed instructions, please see our RAG tutorials under the working with external knowledge tutorials.

Below, see how to index and retrieve data using the embeddings object we initialized above. In this example, we will index and retrieve a sample document in the 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

API Reference:InMemoryVectorStore

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

Direct Usage

Under the hood, the vectorstore and retriever implementations are calling embeddings.embed_documents(...) and embeddings.embed_query(...) to create embeddings for the text(s) used in from_texts and retrieval invoke operations, respectively.

You can directly call these methods to get embeddings for your own use cases.

Embed single texts

You can embed single texts or documents with embed_query:

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

[0.01913362182676792, 0.004960147198289633, -0.01582135073840618, -0.042474791407585144, 0.040200788

Embed multiple texts

You can embed multiple texts with 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

[0.03029559925198555, 0.002908500377088785, -0.02700909972190857, -0.04616579785943031, 0.0382771529
[0.018214847892522812, 0.011460083536803722, -0.03329407051205635, -0.04951060563325882, 0.032756105

API Reference

For detailed documentation on AI21Embeddings features and configuration options, please refer to the API reference.

Related

• Embedding model conceptual guide
• Embedding model how-to guides