nemoguardrails.embeddings.providers.google

Module Contents

Name	Description
`GoogleEmbeddingModel`	Embedding model using Gemini API.

class nemoguardrails.embeddings.providers.google.GoogleEmbeddingModel(
    embedding_model: str,
    kwargs = {}
)

Bases: EmbeddingModel

Embedding model using Gemini API.

This class is a wrapper for using embedding models powered by Gemini API.

To use, you must have either:

Parameters:

embedding_model

str

The name of the embedding model to be used.

**kwargs

Defaults to {}

Additional keyword arguments. Supports:

output_dimensionality (int, optional): Desired output dimensions (128-3072 for gemini-embedding-001). Recommended values: 768, 1536, or 3072. If not specified, API defaults to 3072.
api_key (str, optional): API key for authentication (or use GOOGLE_API_KEY env var).
Other arguments passed to genai.Client() constructor.

_embedding_size

client

= genai.Client(**kwargs)

embedding_size

int

engine_name

= 'google'

output_dimensionality

= kwargs.pop('output_dimensionality', None)

nemoguardrails.embeddings.providers.google.GoogleEmbeddingModel.encode(
    documents: typing.List[str]
) -> typing.List[typing.List[float]]

Encode a list of documents into their corresponding sentence embeddings.

Parameters:

documents

List[str]

The list of documents to be encoded.

Returns: List[List[float]]

List[List[float]]: The list of sentence embeddings, where each embedding is a list of floats.

Raises:

nemoguardrails.embeddings.providers.google.GoogleEmbeddingModel.encode_async(
    documents: typing.List[str]
) -> typing.List[typing.List[float]]

async

Encode a list of documents into their corresponding sentence embeddings.

Parameters:

documents

List[str]

The list of documents to be encoded.

Returns: List[List[float]]

List[List[float]]: The list of sentence embeddings, where each embedding is a list of floats.