AI-102 Question 464
Select 3You are building a customer support chatbot using Azure OpenAI's large multimodal model capabilities. The chatbot must handle both text input from customers and image-based queries (e.g., customers uploading product images for assistance). Which steps should you take to implement this solution effectively?
- A
Use the GPT-4 model from Azure OpenAI for processing both text and image inputs.
- B
Enable the vision capabilities of the multimodal model by providing the appropriate API key and endpoint during setup.
- C
Use the DALL-E model for processing the image inputs before passing them to the GPT-4 model.
- D
Preprocess the image inputs to extract relevant features before passing them to the multimodal model.
- E
Enable image analysis by integrating Azure Cognitive Services Vision API with the Azure OpenAI multimodal model.
Show answer and explanation
Correct answers: A, B, D
Explanation
Azure OpenAI's large multimodal models, such as GPT-4, are capable of processing both text and image inputs natively. To leverage these capabilities, ensure the correct API configuration is in place, preprocess image inputs as needed for compatibility, and utilize the model's built-in features without relying on external services like the Vision API or unrelated models like DALL-E.
- A. Correct.
Correct: The GPT-4 model in Azure OpenAI supports multimodal capabilities, such as processing both text and image inputs.
- B. Correct.
Correct: To use the multimodal capabilities of Azure OpenAI, you must configure the API key and endpoint correctly during setup.
- C. Incorrect.
Incorrect: DALL-E is specifically for generating images from text prompts, not for processing image inputs within a multimodal context.
- D. Correct.
Correct: Preprocessing images (e.g., resizing, format conversion) is often necessary to ensure compatibility with the multimodal model's input requirements.
- E. Incorrect.
Incorrect: While Azure Cognitive Services Vision API is useful for image analysis, the multimodal model in Azure OpenAI can directly handle image and text inputs without requiring an external integration.