AI-102 Question 489
Single answerYou are tasked with fine-tuning an Azure OpenAI model to classify customer feedback into three categories: 'Positive', 'Negative', and 'Neutral'. Which of the following steps is NOT required during the fine-tuning process?
- A
Prepare and clean the dataset, ensuring it is in JSONL format.
- B
Deploy the fine-tuned model to an Azure Kubernetes Service (AKS) cluster.
- C
Upload the dataset to an Azure Storage Blob or directly to the Azure OpenAI service.
- D
Specify the training parameters such as the number of epochs during the fine-tuning process.
Show answer and explanation
Correct answer: B
Explanation
Fine-tuning an Azure OpenAI model involves preparing a well-formatted dataset, uploading it to the appropriate location, and specifying training parameters. However, deploying the model to an AKS cluster is not necessary, as Azure OpenAI manages the deployment of fine-tuned models automatically within its ecosystem. This makes option 2 the correct answer.
- A. Incorrect.
Preparing and cleaning the dataset and ensuring it is in JSONL format is a critical step for fine-tuning an Azure OpenAI model. Properly formatted data is required for training.
- B. Correct.
Deploying the fine-tuned model to an Azure Kubernetes Service (AKS) cluster is not a required step for fine-tuning. Azure OpenAI manages the deployment of the fine-tuned model within its environment, and AKS is not necessary for this process.
- C. Incorrect.
Uploading the dataset to an Azure Storage Blob or directly to the Azure OpenAI service is a required step, as fine-tuning requires access to the training dataset.
- D. Incorrect.
Specifying training parameters, such as the number of epochs, is an essential part of the fine-tuning process to control how the model learns from the dataset.