Embedding Models#

This page presents all APIs and classes related to LLM Embedding models.

EmbeddingModel#

class wayflowcore.embeddingmodels.embeddingmodel.EmbeddingModel(__metadata_info__, id=None, name=None, description=None)#

Abstract base class for embedding models.

Implementations should define the ‘embed’ method which returns a list of vector embeddings (each embedding is a list of floats) given a list of text strings.

Parameters:

  • __metadata_info__ (Dict[str, Any] | None) –

  • id (str | None) –

  • name (str | None) –

  • description (str | None) –

abstract embed(data)#

Generate embeddings for the given list of text strings.

Parameters:

data (List[str]) – A list of text strings for which to generate embeddings.

Returns:

A list where each element is a list of floats representing the embedding of the corresponding text.

Return type:

List[List[float]]

All models#

OpenAI Compatible Embedding Models#

class wayflowcore.embeddingmodels.openaicompatiblemodel.OpenAICompatibleEmbeddingModel(model_id, base_url, __metadata_info__=None, id=None, name=None, description=None)#

Base class for OpenAI-compatible embedding models.

Parameters:

  • model_id (str) – The name of the model to use for generating embeddings.

  • base_url (str) – The base URL for the embedding API. Both HTTP and HTTPS protocols are supported.

  • __metadata_info__ (Dict[str, Any] | None) –

  • id (str | None) –

  • name (str | None) –

  • description (str | None) –

embed(data)#

Generate embeddings for the given list of text strings.

Parameters:

data (List[str]) – A list of text strings for which to generate embeddings.

Returns:

A list where each element is a list of floats representing the embedding of the corresponding text.

Return type:

List[List[float]]

async embed_async(data)#
Parameters:

data (List[str]) –

Return type:

List[List[float]]

OpenAI Embedding Models#

class wayflowcore.embeddingmodels.openaimodel.OpenAIEmbeddingModel(model_id, api_key=None, __metadata_info__=None, id=None, name=None, description=None, _validate_api_key=True)#

Embedding model for OpenAI’s embedding API using the requests library.

Parameters:

  • model_id (str) – The name of the OpenAI model to use for generating embeddings.

  • api_key (str | None) – The API key for the service. If not provided, the value of the OPENAI_API_KEY environment variable will be used, if set.

  • Variables (Environment) –

  • --------------------- –

  • OPENAI_API_KEY – The default API key used for authentication with the embedding service when api_key is not explicitly provided.

  • __metadata_info__ (Dict[str, Any] | None) –

  • id (str | None) –

  • name (str | None) –

  • description (str | None) –

  • _validate_api_key (bool) –

Examples

>>> from wayflowcore.embeddingmodels.openaimodel import OpenAIEmbeddingModel  
>>> model = OpenAIEmbeddingModel(model_id="text-embedding-3-small", api_key="<your API key>")  
>>> embeddings = model.embed(["WayFlow is a framework to develop and run LLM-based assistants."])  
>>> # Update API key after initialization
>>> model.api_key = "<your new API key>"  
>>> # If no key is provided, it will try to use the OPENAI_API_KEY environment variable
>>> model.api_key = None  # Will use environment variable if available

Notes

If the API key is not provided and the environment variable OPENAI_API_KEY is not set, a ValueError is raised.

Available embedding models: https://platform.openai.com/docs/guides/embeddings#embedding-models

VLLM Embedding Models#

class wayflowcore.embeddingmodels.vllmmodel.VllmEmbeddingModel(model_id, base_url, __metadata_info__=None, id=None, name=None, description=None)#

Embedding model for self-hosted models via vLLM.

Parameters:

  • base_url (str) – The complete URL of the vLLM server (e.g., “http://localhost:8000” or “https://secure-vllm.example.com”). Both HTTP and HTTPS protocols are supported.

  • model_id (str) – The name of the model to use on the server.

  • __metadata_info__ (Dict[str, Any] | None) –

  • id (str | None) –

  • name (str | None) –

  • description (str | None) –

Examples

>>> from wayflowcore.embeddingmodels.vllmmodel import VllmEmbeddingModel  
>>> # Using HTTP
>>> model = VllmEmbeddingModel(url="http://localhost:8000", model_id="hosted-model-name")  
>>> # Using HTTPS
>>> secure_model = VllmEmbeddingModel(url="https://secure-vllm.example.com", model_id="hosted-model-name")  
>>> embeddings = model.embed(["WayFlow is a framework to develop and run LLM-based assistants."])

Notes

This provider makes HTTP/HTTPS POST requests to the /v1/embeddings endpoint of the specified URL. When using HTTPS, certificate verification is performed by default. For self-signed certificates, additional configuration may be needed at the application level.

OCI GenAI Embedding Models#

class wayflowcore.embeddingmodels.ocigenaimodel.OCIGenAIEmbeddingModel(model_id, config, compartment_id=None, __metadata_info__=None, id=None, name=None, description=None)#

Embedding model for Oracle OCI Generative AI service.

Parameters:

  • model_id (str) – The model identifier (e.g., ‘cohere.embed-english-light-v3.0’).

  • config (OCIClientConfig) – OCI client configuration with authentication details.

  • compartment_id (str | None) – The compartment OCID

  • __metadata_info__ (Dict[str, Any] | None) –

  • id (str | None) –

  • name (str | None) –

  • description (str | None) –

Examples

The following examples show how to configure the OCIGenAIEmbeddingModel. In actual use, replace the placeholder values with your real credentials.

# Option 1: User Authentication Config >>> from wayflowcore.models.ociclientconfig import OCIUserAuthenticationConfig, OCIClientConfigWithUserAuthentication # doctest: +SKIP >>> oci_user_config = OCIUserAuthenticationConfig( # doctest: +SKIP … user=”<my_user_ocid>”, … key_content=”<my_key_content>”, … fingerprint=”<fingerprint_of_my_public_key>”, … tenancy=”<my_tenancy_ocid>”, … region=”<my_oci_region>”, … ) >>> client_config = OCIClientConfigWithUserAuthentication( # doctest: +SKIP … service_endpoint=”https://inference.generativeai.us-<your key’s region>.oci.oraclecloud.com”, … compartment_id=”<Please read the ‘Using the API_KEY authentication method’ subsection in the ‘How to Use LLM from Different LLM Sources/Providers’ how-to guide>”, … user_config=oci_user_config … )

# Option 2: Instance Principal >>> from wayflowcore.models.ociclientconfig import OCIClientConfigWithInstancePrincipal # doctest: +SKIP >>> client_config = OCIClientConfigWithInstancePrincipal( # doctest: +SKIP … service_endpoint=”https://inference.generativeai.us-<your key’s region>.oci.oraclecloud.com”, … compartment_id=”<Please read the ‘Using the API_KEY authentication method’ subsection in the ‘How to Use LLM from Different LLM Sources/Providers’ how-to guide>”, … )

# Option 3: API Key from the default config file >>> from wayflowcore.models.ociclientconfig import OCIClientConfigWithApiKey # doctest: +SKIP >>> client_config = OCIClientConfigWithApiKey( # doctest: +SKIP … service_endpoint=”https://inference.generativeai.us-<your key’s region>.oci.oraclecloud.com”, … compartment_id=”<Please read the ‘Using the API_KEY authentication method’ subsection in the ‘How to Use LLM from Different LLM Sources/Providers’ how-to guide>”, … )

# Using the configured client with the embedding model >>> from wayflowcore.embeddingmodels.ocigenaimodel import OCIGenAIEmbeddingModel >>> model = OCIGenAIEmbeddingModel( # doctest: +SKIP … model_id=”cohere.embed-english-light-v3.0”, … config=client_config, # Use whichever client_config option you prefer … ) >>> embeddings = model.embed([“WayFlow is a framework to develop and run LLM-based assistants.”]) # doctest: +SKIP

Notes

The OCI SDK must be installed. For generating the OCI config file, please follow the WayFlow documentation.

Available embedding models: https://docs.oracle.com/en-us/iaas/Content/generative-ai/pretrained-models.htm

embed(data)#

Generate embeddings for the given list of text strings.

Parameters:

data (List[str]) – A list of text strings for which to generate embeddings.

Returns:

A list where each element is a list of floats representing the embedding of the corresponding text.

Return type:

List[List[float]]

Ollama Embedding Models#

class wayflowcore.embeddingmodels.ollamamodel.OllamaEmbeddingModel(model_id, base_url, __metadata_info__=None, id=None, name=None, description=None)#

Embedding model for self-hosted models via Ollama.

Parameters:

  • base_url (str) – The complete URL of the Ollama server (e.g., “http://localhost:11434” or “https://ollama.example.com”). Both HTTP and HTTPS protocols are supported.

  • model_id (str) – The name of the model to use on the Ollama server.

  • __metadata_info__ (Dict[str, Any] | None) –

  • id (str | None) –

  • name (str | None) –

  • description (str | None) –

Examples

>>> from wayflowcore.embeddingmodels.ollamamodel import OllamaEmbeddingModel  
>>> # Using HTTP
>>> model = OllamaEmbeddingModel(url="http://localhost:11434", model_id="nomic-embed-text")  
>>> # Using HTTPS
>>> secure_model = OllamaEmbeddingModel(url="https://ollama.example.com", model_id="nomic-embed-text")  
>>> embeddings = model.embed(["WayFlow is a framework to develop and run LLM-based assistants."])

Notes

This provider makes HTTP/HTTPS POST requests to the /v1/embeddings endpoint of the specified URL. When using HTTPS, certificate verification is performed by default. For self-signed certificates, additional configuration may be needed at the application level.