Embeddings Module API Reference Documentation

register_embeddings_provider

Register a provider for embedding models.

Function Signature

def register_embeddings_provider(
    provider_name: str,
    embeddings_model: EmbeddingsType,
    base_url: Optional[str] = None,
) -> None:

Parameters

Parameter	Type	Required	Default	Description
provider_name	str	Yes	-	Custom provider name
embeddings_model	EmbeddingsType	Yes	-	Embedding model class or supported provider string type
base_url	Optional[str]	No	None	BaseURL of the provider

Example

register_embeddings_provider("fakeembeddings", FakeEmbeddings)
register_embeddings_provider("vllm", "openai-compatible", base_url="http://localhost:8000/v1")

---

batch_register_embeddings_provider

Batch register embedding model providers.

Function Signature

def batch_register_embeddings_provider(
    providers: list[EmbeddingProvider]
) -> None:

Parameters

Parameter	Type	Required	Default	Description
providers	list[EmbeddingProvider]	Yes	-	List of provider configurations

Example

batch_register_embeddings_provider([
    {"provider_name": "fakeembeddings", "embeddings_model": FakeEmbeddings},
    {"provider_name": "vllm", "embeddings_model": "openai-compatible", "base_url": "http://localhost:8000/v1"},
])

---

load_embeddings

Load an embedding model from registered providers.

Function Signature

def load_embeddings(
    model: str,
    *,
    provider: Optional[str] = None,
    **kwargs: Any,
) -> Embeddings:

Parameters

Parameter	Type	Required	Default	Description
model	str	Yes	-	Model name, format: model_name or provider_name:model_name
provider	Optional[str]	No	None	Model provider name
**kwargs	Any	No	-	Additional model parameters

Example

embeddings = load_embeddings("vllm:qwen3-embedding-4b")

---

create_openai_compatible_embedding

Create an OpenAI compatible embedding model class.

Function Signature

def create_openai_compatible_embedding(
    embedding_provider: str,
    base_url: Optional[str] = None,
    embedding_model_cls_name: Optional[str] = None,
) -> type[Embeddings]:

Parameters

Parameter	Type	Required	Default	Description
embedding_provider	str	Yes	-	Embedding model provider name
base_url	Optional[str]	No	None	BaseURL of the model provider
embedding_model_cls_name	Optional[str]	No	None	Custom embedding model class name

Return Value

Type	Description
type[Embeddings]	Dynamically created OpenAI compatible embedding model class

Example

VLLMEmbeddings = create_openai_compatible_embedding(
    embedding_provider="vllm",
    base_url="http://localhost:8000/v1",
    embedding_model_cls_name="VLLMEmbeddings",
)

EmbeddingsType

Types supported by the embeddings_model parameter when registering embedding providers.

Type Definition

EmbeddingsType = Union[type[Embeddings], Literal["openai-compatible"]]

---

EmbeddingProvider

Embedding model provider configuration type.

Class Definition

class EmbeddingProvider(TypedDict):
    provider_name: str
    embeddings_model: EmbeddingsType
    base_url: NotRequired[str]

Field Description

Field	Type	Required	Description
provider_name	str	Yes	Provider name
embeddings_model	EmbeddingsType	Yes	Embedding model class or string
base_url	NotRequired[str]	No	Base URL