nemoguardrails.embeddings.providers

Submodules

Package Contents

Functions

Name	Description
`init_embedding_model`	Initialize the embedding model.
`register_embedding_provider`	Register an embedding provider.

Data

_embedding_model_cache

embeddings_executor

API

nemoguardrails.embeddings.providers.init_embedding_model(
    embedding_model: str,
    embedding_engine: str,
    embedding_params: dict = {}
) -> nemoguardrails.embeddings.providers.base.EmbeddingModel

Initialize the embedding model.

Parameters:

embedding_model

str

The path or name of the embedding model.

embedding_engine

str

The name of the embedding engine.

embedding_params

dictDefaults to {}

Additional parameters for the embedding model.

Returns: EmbeddingModel

An instance of the initialized embedding model.

Raises:

ValueError: If the embedding engine is invalid.

nemoguardrails.embeddings.providers.register_embedding_provider(
    model: typing.Type[nemoguardrails.embeddings.providers.base.EmbeddingModel],
    engine_name: typing.Optional[str] = None
)

Parameters:

model

Type[EmbeddingModel]

The embedding model class.

engine_name

strDefaults to None

The name of the embedding engine.

Raises:

ValueError: If the engine name is not provided and the model does not have an engine name.
TypeError: If the model is not an instance of EmbeddingModel.
ValueError: If the model does not have ‘encode’ or ‘encode_async’ methods.

nemoguardrails.embeddings.providers._embedding_model_cache = {}

nemoguardrails.embeddings.providers.embeddings_executor = None