Creating a New Training Session with Test Split in PrismRCL

Introduction #

PrismRCL, a versatile AI classification tool, allows for efficient training sessions with a feature to automatically create a test split from your dataset. This functionality is crucial for validating the performance of your model on unseen data. This article guides you through the process of starting a new training session with a test split.

Requirements #

• Dataset: Ensure your data is correctly organized and formatted as per PrismRCL guidelines.
• PrismRCL Installation: Verify that PrismRCL is installed on your Windows system.

Command Overview #

The command for initiating a training session with a test split is structured as follows:

C:\PrismRCL\PrismRCL.exe fractal rclticks=15 data=C:\PrismRCL\data\dataset\train-data testsize=0.1 savemodel=C:\PrismRCL\models\model_name.classify log=C:\PrismRCL\logfiles\job_folder stopwhendone

• fractal: Specifies the evaluation method to be used for this training session, tailored for image data.
• rclticks=15: Sets the specific parameter for the fractal evaluation method (applicable only to image data).
• data: The path to your training dataset.
• testsize=0.1: Defines the fraction of the training dataset to be used as a test split. In this example, 10% of the data is reserved for testing.
• savemodel: The path where the trained model will be saved.
• log: The directory where training session logs and result files will be stored.
• stopwhendone: Instructs PrismRCL to shut down automatically after completing the training session.

Steps to Create a Training Session with Test Split #

1. Prepare Your Dataset: Organize your dataset according to PrismRCL’s data structure requirements. Ensure images are in PNG format, and text/tabular data is correctly formatted.
2. Choose a Model Name: Decide on a name for your model. This name will be used in the savemodel path.
3. Set the Test Split Size: Determine the fraction of your dataset you wish to use for testing. A test size of 0.1 to 0.2 is commonly used for a good balance between training and testing.
4. Execute the Command: Open the command line, navigate to the PrismRCL installation directory, and execute the provided command with your specific paths and settings.
5. Monitoring the Process: Check the log files in the specified log directory to monitor the training progress and results.

Conclusion #

Creating a training session with a test split in PrismRCL is a straightforward process that enhances the evaluation of your AI model. By following the steps outlined in this article, you can efficiently train and validate your model, ensuring it performs well on both the training and unseen test data.