Skip to main content
Version: 0.0.38

General

The General Tab under Custom Models is crucial for setting foundational parameters that define the behavior and configuration of your models. Accessible via Workspace > Models > Custom Models, this tab presents options essential for model identification, functionality, and security. Here's a breakdown of the available settings:

Tab Overview​

1. Name​

  • Display Name: This name appears in the Lakehousecat menu, enabling users to locate and identify the model easily. Selecting a clear, descriptive name ensures that the model's purpose is understandable at a glance.

2. Foundation Model​

  • Underlying Model: Chooses the computational model that serves as the base for responding to inquiries. The Foundation Model handles the logical operations and data processing necessary for Custom Model interactions.

3. System Prompt​

  • Prompt Creation: Develop a System Prompt using the chat or voice dictation feature to optimize the model's response behavior. Dictation is recommended for accurate text conversion, enhancing the prompt's effectiveness.

4. Description (Optional)​

  • User-Focused Description: Offer a concise description that elucidates the model's purpose and use-case scenarios. This optional field aids users in understanding the model's functionality and scope.

5. Standard Fields​

These fields affect the operability and visibility of your Custom Model:

  • Enabled: Activate this switch to make the model operational. Turning off the Enabled option will deactivate all its processes and interactions.

  • Visible: Determine whether the model is viewable by users. If set to invisible, the model remains listed but inaccessible for standard user interaction, allowing for controlled testing or private configurations.

  • IsLocked: Protects the model configuration by preventing accidental changes. When enabled, the settings remain static, especially useful during extensive training sessions where changes could disrupt accuracy.

Best Practices​

  • Clear Naming Conventions: Ensure the name reflects the model’s purpose or functionality, aiding administrators in managing multiple models effectively.
  • Effective Prompts: Carefully constructed System Prompts influence the interactions and responses, so aim for precision and clarity.
  • Visibility and Accessibility: Regularly assess the Visible and Enabled settings based on the current need for model testing or full deployment.
  • Configuration Security: Use the IsLocked feature judiciously during crucial operations to safeguard model configurations against inadvertent modifications.

By managing these settings wisely, administrators can ensure that Custom Models within Lakehousecat are configured optimally to meet organizational requirements and deliver precise analytical insights.