-
Thanks for clarifiying!
I have another issue considering the following question:
Category: AI-102 – Plan and manage an Azure AI solution
You are working with an Azure subscription that includes the following resources: OpenAIResource1, an Azure OpenAI resource used to generate generative answers for user queries in a chatbot, ContentFilter1, an Azure AI Content Safety resource designed to filter out objectionable content from both the chatbot’s input and output, and SpeechToTextResource1, an Azure Cognitive Services Speech-to-Text resource for converting spoken language into text.
Your goal is to improve the accuracy of the content filter configurations by conducting tests with sample questions to ensure the content filter is appropriately fine-tuned.
Solution: You plan to use the Moderate text content feature from Content Safety Studio to perform the tests.
Which solution will meet this requirement?
Correct answer: “Yes”What I don’t understand here:
We are asking to “fine-tune” a model which requires retraining in Machine-Learning-Lingo. Simply applying the “moderate text” feature improves our solution but does not incorporate retraining a model, hence fine-tuning is not applied. Therefore, compared to our previous case where we should have considered the complete process including retraining, not just labelling, we have exactly the same case as before – as retraining is not part of the solution and no fine-tuning will be applied. The answer should therefore be “No”. Your correct answer is “Yes”. This again is misleading and contradicting your previous answer.
