Training

Model Training Documentation¶

This documentation provides a detailed guide on how to set up and initiate a custom model training session using the interface

Interface Overview¶

This interface allows users to configure and start the training of a custom model, with several components to facilitate the setup process.

Project and Asset Configuration¶

Team Project ID: Enter the name of your project (e.g., agency-brand).
Model Name: Specify the name of the model (e.g., demo_selection_1).
Set Asset Type: Choose the asset type from available options: Object, Subject or Style.

Status Notifications¶

Receive real-time notifications confirming actions such as setting asset types and folder statuses.

Config Options¶

Upload Image:
Upload your training dataset file.
Ensure the file size and format comply with the requirements.
Folder Status: Verify the image directory status is updated successfully.

Class Descriptor¶

Select Category: If the asset type is Objector Subject, choose an appropriate category for class descriptor (e.g., cat, mascot).

Step Multiplier Settings¶

Adjust Multiplier: The default value is set to 60. Adjust this based on the size and complexity of your dataset for optimal training outcomes.

Training Controls¶

Check Model Status: Before training, use this feature to monitor the status and readiness of the model before the training has started.
Train Model: Click this button to begin the model training process. Ensure all fields are correctly filled out before starting.

By following this guide, users can effectively configure and train their custom models, ensuring all necessary parameters are set for optimal performance.