Skip to main content
Version: 0.0.38

Ollama

The Ollama integration with Lakehousecat allows users to interact with a variety of models offered by Ollama, primarily popular in the open-source community. This guide will assist administrators in configuring Ollama within the Lakehousecat environment.

Prerequisites​

Ensure you have the following before starting:

  • Access to Lakehousecat as an admin
  • A valid Ollama Instance URL

Step-by-Step Guide​

  1. Initiation Start by logging in as an admin and navigating to the Admin Workspace. Go to the Models section and select 'Provider Models'. You can choose Ollama by clicking the plus sign.

  2. Model Selection

    • Explore the various available model types, such as Chat and Embedding models.
    • Note: Speech-to-Text and Text-to-Speech models are experimental and may not be available.

  3. Configuration Steps

    • Configuration Name: Enter a clear and distinct name for your Ollama configuration.
    • Instance URL: Provide the Ollama Instance URL. A valid Instance URL facilitates loading additional models from Ollama.com.
    • Model Identification: Consult the Ollama documentation to identify the desired Model ID.

  4. Load Model Utilize the PullModel functionality to load the model from the Ollama library. Once loaded, the model can be selected and persisted within Lakehousecat.

Important Notes​

  • Ollama offers numerous models, but their availability frequently changes. Performance and pre-testing for each model cannot be guaranteed.
  • If working with Lakehousecat, functionality with every Ollama model may not be ensured.

Troubleshooting​

If you face issues during configuration:

  • URL Verification: Double-check the Ollama Instance URL for any errors.
  • Model Loading Errors: Ensure the Model ID provided corresponds with a valid entry in the Ollama documentation.