Create or reuse an Example Store instance

Stay organized with collections Save and categorize content based on your preferences.

This page shows you how to create a new Example Store instance or reuse an existing Example Store instance. You can store your examples in an Example Store when developing your LLM application and dynamically retrieve them to use in your LLM prompts.

To teach an LLM or an agent using few-shot examples, you must first create or reuse an Example Store instance for your project and location, and then upload examples to it.

For each project and location, you can have a maximum of 50 Example Store instances. After you create an Example Store instance, you can share it across multiple LLM applications and agents.

There are two ways to provision an Example Store instance:

• Create a new Example Store instance: When you create a new Example Store instance, you need to specify the embedding model, which Example Store uses to determine which examples are relevant to users' queries. Example Store supports the following embedding models:

  • text-embedding-005

  • text-multilingual-embedding-002

  You can't change an embedding model after you create the Example Store instance. If you want to use a different embedding model, you must create another example store. For more information about text embeddings, see Get text embeddings.

• Reuse an existing Example Store instance: Example Store instances are designed to be used by multiple agents, so you can access the stored examples across LLM applications. You can't change the embedding model when you reuse an existing Example Store instance.

Prerequisites

Before you use the Python samples on this page, install and initialize the Vertex AI SDK for Python in your local Python environment.

1. Run the following command to install the Vertex AI SDK for Python for Example Store.

   pip install --upgrade google-cloud-aiplatform>=1.87.0

2. Use the following code sample to import and initialize the SDK for Example Store.

   import vertexai
   from vertexai.preview import example_stores
   
   vertexai.init(
     project="PROJECT_ID",
     location="LOCATION"
   )
   

   Replace the following:

   • PROJECT_ID: Your project ID.

   • LOCATION: Your region. Only us-central1 is supported.

Create an Example Store instance

Use the following samples to create an Example Store instance for a specified project and location. Note that creating an Example Store instance takes a few minutes.

import vertexai
from vertexai.preview import example_stores

vertexai.init(
    project="PROJECT_ID",
    location="LOCATION"
)

my_example_store = example_stores.ExampleStore.create(
    example_store_config=example_stores.ExampleStoreConfig(
        vertex_embedding_model="EMBEDDING_MODEL"
    )
)

POST https://LOCATION-aiplatform.googleapis.com/v1beta1/projects/PROJECT_ID/locations/LOCATION/exampleStores

{
  "display_name": "DISPLAY_NAME",
  "example_store_config": {"vertex_embedding_model": EMBEDDING_MODEL}
}

curl -X POST \
     -H "Authorization: Bearer $(gcloud auth print-access-token)" \
     -H "Content-Type: application/json; charset=utf-8" \
     -d @request.json \
     "https://LOCATION-aiplatform.googleapis.com/v1beta1/projects/PROJECT_ID/locations/LOCATION/exampleStores"

$cred = gcloud auth print-access-token
$headers = @{ "Authorization" = "Bearer $cred" }

Invoke-WebRequest `
    -Method POST `
    -Headers $headers `
    -ContentType: "application/json; charset=utf-8" `
    -InFile request.json `
    -Uri "https://LOCATION-aiplatform.googleapis.com/v1beta1/projects/PROJECT_ID/locations/LOCATION/exampleStores" | Select-Object -Expand Content

{
  "name": "projects/PROJECT_ID/locations/LOCATION/exampleStores/EXAMPLE_STORE_ID/operations/",
  "metadata": {
    "@type": "type.googleapis.com/google.cloud.aiplatform.v1beta1.CreateExampleStoreOperationMetadata",
    "genericMetadata": {
      "createTime": "2024-10-10T02:06:10.417111Z",
      "updateTime": "2024-10-10T02:06:10.417111Z"
    }
  }
}

Reuse an existing Example Store instance

Use the following sample to reuse an existing Example Store instance for a specified project and location.

import vertexai
from vertexai.preview import example_stores

vertexai.init(
    project="PROJECT_ID",
    location="LOCATION"
)

example_store = example_stores.ExampleStore(
    "EXAMPLE_STORE_NAME")

What's next

• Upload examples to the Example Store instance.