Skip to main content
Version: v3.2.0

Embedding Model Registration guide

Embedding models are essential for various use cases such as document and query embedding and retrieval. By default, all users are provided with a preloaded embedding model, BAAI/bge-small-en-v1.5. This guide explains how to use the default model and how to register additional embedding models from supported providers: Azure, OpenAI, and Text Embeddings Inference (Hugging Face).

Default Preloaded Model​

All users have access to the BAAI/bge-small-en-v1.5 embedding model by default. This model can be used for embedding tasks immediately after account setup.

Note: Since this model is shared across all users, there may be potential time delays during heavy usage periods.

Registering Additional Embedding Models​

To use embedding models beyond the default, follow these steps to register an embedding model from one of the supported providers.

Step 1: Access the Embedding Model Registration Page​

Embedding model registration can be accessed from both the Account Settings and the Optimization Hub:

  • Account Settings: You can access the registration under the Embedding Models tab.
Launchpad Register Embedding Models

Step 2: Select Provider, Enter API Key, Base URL, to Register Models​

Register Embedding Models

1. Select Your Embedding Model Provider​

On the embedding model registration page, click the dropdown under Model Provider and choose your provider from the available options: Azure, OpenAI, or Text Embedding Inference (Hugging Face).

2. Enter API Key (and Base URL, if required)​

  • OpenAI: Requires only the API Key.
  • Azure: Requires both the API Key and the Base URL.
  • Text Embedding Inference: Most often requires both the API Key and the Base URL.

Enter the required credentials in the respective fields. For more information on how to find your API key and base URL, refer to the section below on Finding Your API Keys and Base URLs.

3. Choose Embedding Model(s) to Register​

After entering your API key and base URL (if applicable), a dropdown menu labeled Embedding Model Name will appear, containing the models available under your selected provider. Select one or more models you wish to register.

Note: The registration process may take some time, as background checks are performed to verify model availability.

Register OpenAI Embedding Models

Step 3: Register the Embedding Model​

Click the Register button next to the selected model(s). Upon successful registration, the model(s) will be linked to your account, and the credentials will be saved securely.

Successful Registration

Finding Your API Keys and Base URLs​

Below are instructions on how to find API keys and base URLs for each embedding model provider.

Azure​

  1. Visit Azure Portal: Navigate to the Azure AI Foundry.
  2. Log In to Your Account: Ensure you are logged in with your Azure account credentials.
  3. Access API Key:
    • Go to your Azure AI Foundry deployments and select the embedding model of interest.
      • Navigate to the Endpoint section under the model details.
      • Copy the provided API Key.
  4. Access Base URL:
    • In the same Endpoint section, locate the Target URI.
      • Copy this URI, which serves as the Base URL for your model.

Note: The base URL for Azure models might need manual reformatting in case of OpenAI models.

  • The base URL for OpenAI Azure models will follow the format:
    https://<your-resource-name>.openai.azure.com//openai/deployments/<model_name>.
    • Replace <your-resource-name> with the name of your Azure AI Foundry resource.
    • Replace <your-model-name> with the name of your deployed model.

OpenAI​

  1. Log in to Your OpenAI Account: Visit the OpenAI platform.
  2. Navigate to API Keys: Once logged in, go to the API Keys section.
  3. Create or View Your Keys: If you don’t have an API key, create a new one or view existing keys.

Text Embedding Inference (Hugging Face)​

  1. Visit Hugging Face Settings: Navigate to Hugging Face tokens.
  2. Log In to Your Account: Ensure you are signed in to your Hugging Face account.
  3. Access Tokens: Create a new token or view your existing tokens.
  4. Access Base URL: The base URL is provided when registering your model with Text Embedding Inference.

Note: Keep your API keys secure and do not share them publicly. Each key provides access to your LLM provider's resources, and misuse can lead to unintended charges or unauthorized access to your data.

Additional Information​

For assistance or more detailed instructions on registering your LLMs, please refer to our user guide or contact our support team at help@trustwise.ai.