VertexAIEmbeddings class

Wrapper around GCP Vertex AI text embedding models API

Example:

final embeddings = VertexAIEmbeddings(
  httpClient: authClient,
  project: 'your-project-id',
);
final result = await embeddings.embedQuery('Hello world');

Vertex AI documentation: https://cloud.google.com/vertex-ai/docs/generative-ai/language-model-overview

Set up your Google Cloud Platform project

  1. Select or create a Google Cloud project.
  2. Make sure that billing is enabled for your project.
  3. Enable the Vertex AI API.
  4. Configure the Vertex AI location.

Authentication

To create an instance of VertexAIEmbeddings you need to provide an HTTP client that handles authentication. The easiest way to do this is to use AuthClient from the googleapis_auth package.

To create an instance of VertexAIEmbeddings you need to provide an AuthClient instance.

There are several ways to obtain an AuthClient depending on your use case. Check out the googleapis_auth package documentation for more details.

Example using a service account JSON:

final serviceAccountCredentials = ServiceAccountCredentials.fromJson(
  json.decode(serviceAccountJson),
);
final authClient = await clientViaServiceAccount(
  serviceAccountCredentials,
  [VertexAIEmbeddings.cloudPlatformScope],
);
final vertexAi = VertexAIEmbeddings(
  httpClient: authClient,
  project: 'your-project-id',
);

The service account should have the following permission:

  • aiplatform.endpoints.predict

The requiredOAuth2 scope is:

  • https://www.googleapis.com/auth/cloud-platform (you can use the constant VertexAIEmbeddings.cloudPlatformScope)

See: https://cloud.google.com/vertex-ai/docs/generative-ai/access-control

Available models

  • textembedding-gecko
    • Max input token: 3072
    • Output: 768-dimensional vector embeddings
  • textembedding-gecko-multilingual: support over 100 non-English languages
    • Max input token: 3072
    • Output: 768-dimensional vector embeddings

The previous list of models may not be exhaustive or up-to-date. Check out the Vertex AI documentation for the latest list of available models.

Task type

Embedding models released after August 2023 support specifying a 'task type' when embedding documents. The task type is then used by the model to improve the quality of the embeddings.

This class uses the specifies the following task type:

  • retrievalDocument: for embedding documents
  • retrievalQuery: for embedding queries

Title

Embedding models released after August 2023 support specifying a document title when embedding documents. The title is then used by the model to improve the quality of the embeddings.

To specify a document title, add the title to the document's metadata. Then, specify the metadata key in the docTitleKey parameter.

Example:

final embeddings = VertexAIEmbeddings(
  httpClient: authClient,
  project: 'your-project-id',
  docTitleKey: 'title',
);
final result = await embeddings.embedDocuments([
  Document(
    pageContent: 'Hello world',
    metadata: {'title': 'Hello!'},
  ),
]);

Constructors

VertexAIEmbeddings.new({required Client httpClient, required String project, String location = 'us-central1', String? rootUrl, String publisher = 'google', String model = 'textembedding-gecko', int batchSize = 5, String docTitleKey = 'title'})
Wrapper around GCP Vertex AI text embedding models API

Properties

batchSize → int
The maximum number of documents to embed in a single request.
final
client → VertexAIGenAIClient
A client for interacting with Vertex AI API.
final
docTitleKey → String
The metadata key used to store the document's (optional) title.
final
hashCode → int
The hash code for this object.
no setterinherited
model → String
The embeddings model to use.
final
publisher → String
The publisher of the model.
final
runtimeType → Type
A representation of the runtime type of the object.
no setterinherited

Methods

embedDocuments(List<Document> documents) → Future<List<List<double>>>
Embed search docs.
embedQuery(String query) → Future<List<double>>
Embed query text.
noSuchMethod(Invocation invocation) → dynamic
Invoked when a nonexistent method or property is accessed.
inherited
toString() → String
A string representation of this object.
inherited

Operators

operator ==(Object other) → bool
The equality operator.
inherited

Constants

cloudPlatformScope → const String
Scope required for Vertex AI API calls.