Creating AI Models from the Gallery

OneTrust AI Governance offers a guided flow when creating a new model record to easily register model details by leveraging the Hugging Face platform. This Model Gallery allows you to populate additional metadata for the model record plus associate the model record to an existing inventory record using the relationship type feature.

The Model Gallery is populated by Hugging Face, an open-source platform for AI training and learning. Learn more here. The gallery returns the 10 most downloaded models from the following authors, as well as the top 50 most downloaded models overall:

  • Google

  • Mistral AI

  • Stability AI

  • OpenAI

  • Meta

  • Microsoft

  • BigCode

For the generally available model creation flow, see Managing Models in AI Governance.

To create a model from the Gallery

  1. On the AI Governance menu, select Inventory > Models.

    The Models screen appears.

  2. Click the Add New button.

    The Set Model Type screen appears.

  3. Use the fields to filter and choose an available model and its respective details from Hugging Face.

    Add new model - set model type

    Field

    Options

    Task

    Natural Language Processing

    Audio

    Computer Vision

    Reinforcement Learning

    Tabular

    Multimodal

    Other

    Provider

    Big Code

    Google

    Meta

    Microsoft

    Mistral AI

    Open AI

    Stability AI

    Sort

    Newest to Oldest

    Oldest to Newest

    A to Z

    Z to A
  4. Click the Continue button.

    The Add Details screen appears.

  5. Update the model name from Hugging Face to a preferred naming.

    Add new model - add details
  6. Expand the Add and Review Additional Details (optional) section to manage more fields.

    Add and review additional details

    Field

    Description/Option

    Owner

    Choose from the available users in the application.

    Description

    Enter text to supply details.

    Internal or 3rd Party

    Choose from the options:

    • Self-managed open source software

    • Third-party managed open source software

    • Self-managed proprietary

    • Third-party managed proprietary

    Autonomy Level

    Choose from the options:

    • Human Support

    • Human-In-The-Loop

    • Human-On-The-Loop

    • Human-Out-Of-The-Loop

    Reference Link

    Enter text to supply details.

    Biases

    Enter text to supply details.

    Limitations

    Enter text to supply details.

    Demographic Parity Difference

    Enter text to supply details.

    Demographic Parity Ratio

    Enter text to supply details.

    Equalized Odds Difference

    Enter text to supply details.

    Equalized Odds Ratio

    Enter text to supply details.

    Generative AI

    Select Yes or No.

    Framework

    Choose from the options:

    • AccelOnnx

    • AutoML

    • Chainer

    • H20

    • JAX

    • Keras

    Hosting Provider

    Choose from the options:

    • Amazon Web Services (AWS)

    • Google Cloud Platform (GCP)

    • Microsoft Azure

    • IBM Cloud

    • Algorithmia

    • Heroku

    Model Version

    Enter text to supply details.

    Programming Language

    Choose from the options:

    • Python

    • R Programming

    • JavaScript / Java

    • Julia

    • Lisp

    • Scala

    Type of Learning

    Choose from the options:

    • Supervised

    • Unsupervised

    • Reinforcement

    Task Type

    This field is pre-populated based on the selected model type.
  7. Click the Continue button.
  8. Here you can add relationships to link this new model to another inventory that already exists in your tenant.
  9. Define the relationship type, target type, and target record, then click the Add button.

    Add new model relationship type
  10. When finished linking relationships, click the Continue button.

    The Review screen appears.

  11. Review the details before creating your AI model.

    Add new model review
  12. Click the Create model button.

    The Inventory Details screen for the new model appears.