Utility and Fine-Tuning

[home]

Utility and Fine-Tuning contains tools to help you create and manage your fine-tuned models. You can access the page by clicking on Utility on the main menu.

The Utility page consists of the following sections:

Fine Tuning

• Displays any Fine-Tune models you have created on OpenAI.
• Edit button - Allows you to rename the file name to a more easily understandable name.
• Delete button - Allows you to delete the Fine-Tune model. This will actually delete this model on OpenAI.

Create Training Data

• Allows you to select an existing Story from the dropdown.
• Allows you to select a section in the left-hand window, and edit the content of the text in the (Training Data) and (Instruction) tab on the right.
• Upload to OpenAI button - Sends the selected Story to OpenAI as training data and starts processing it as a fine-tuned model.
• Export Excel button - Allows you to export the training data as an Excel file.

JsonL Utility

• Download Template - Provides an Excel file to use as a template for your Fine-Tune model training data.
• Upload Excel file and create JsonL file - Allows you to upload the Excel file and return the data in the JsonL format that can then be uploaded to OpenAI to Fine-Tune your model.

Guide to Automatically Fine-Tuning Your AI Model with OpenAI

Creating the training data to create a fine-tuned model is simple and easy when using the Create Training Data tab.

1. Navigate to the Create Training Data tab
2. Select an existing Story where all the existing prose has been edited to match your writing style
3. Click the Upload to OpenAI button

You will see status messages that the contents of the Story have been uploaded to OpenAI as training data and that fine-tuning has started.

You can navigate to: https://platform.openai.com/finetune and see the progress.

When fine-tuning is complete you will receive an email from OpenAI.

Click on the Home button and then navigate back to the Utility page by clicking the Utility button.

On the Fine Tuning tab you will see your new fine-tuned model.

OpenAI assigns a unique identifier to each fine-tuned model. If desired, you can rename your model to something more memorable or descriptive. This can be done by selecting the Edit button next to the model's name on the Utility page in the AIStoryBuilders application.

Once your model is fine-tuned and ready to go, you can start using it to generate responses by selecting it from the AI Model dropdown.

Export To Excel

You can also click the Export Excel button to export the training data to Excel. This will allow you to edit the data and then use the JsonL Utility page to convert it from Excel to JsonL to upload manually to OpenAI. See "Step-by-Step Guide to Manually Fine-Tuning Your AI Model with OpenAI" below.

Editing Text

Before submitting the training data to OpenAI, you can click on any section in the left-hand window and edit the content of the text in the (Training Data) and (Instruction) tab on the right.

Step-by-Step Guide to Manually Fine-Tuning Your AI Model with OpenAI

1. Create at Least 10 Samples
   
   
   
   Fine-tuning an AI model with OpenAI starts with creating training data. You'll need at least 10 samples, however the more, the better. Use the Download Template button to retrieve the Excel file to use for the data. A sample of the type of content you would want to create can be found in the file at the following link: FineTunedSample.xls
   
   
2. Create the JSONL File
   
   
   
   Once you have your training data entered into the Excel template, you need to format it in the JSONL file format to upload to OpenAI. JSONL stands for JSON Lines, which means each line must be a valid JSON object. This is difficult to do manually. Click the Upload Excel file and create JsonL file button to open the dialog to allow you to select the Excel file with your training data.
   
   
   
   A JSONL file will be returned. Save the file.
   
   
3. Upload to OpenAI
   
   
   
   After preparing your JSONL file, navigate to: https://platform.openai.com/finetune and upload it to OpenAI through their user interface.
   
   
4. Receive the Email on progress from OpenAI
   
   
   
   Once you trigger the fine-tuning process, OpenAI will use your training data to fine-tune the model. You'll receive an email notification once your model starts training and another when the training is complete. This email will contain important information about your model's training status and any issues encountered during the process.
   
   
5. (Optional) Rename the Model
   
   
   
   OpenAI assigns a unique identifier to each fine-tuned model. If desired, you can rename your model to something more memorable or descriptive. This can be done by selecting the Edit button next to the model's name on the Utility page in the AIStoryBuilders application.
   
   
6. Consume the Model When Making an AI Prompt
   
   
   
   Once your model is fine-tuned and ready to go, you can start using it to generate responses by selecting it from the AI Model dropdown.