You pass a photo to a deployed multimodal model and ask what is in it. What makes this possible?
Why B is the answer
The model accepts images as prompt input
Multimodal models take images directly in the prompt alongside text.
In detail
- Multimodal models accept images as prompt input.
- Check that the deployed model supports images.
Use case
An illustrative scenario, not a real incident.
A mechanic uploads a photo of a part and asks what it is. A multimodal model takes the image as input and replies in text.
See it in different ways
mindmap
root((AI workloads))
Text
Sentiment
Key phrases
Entities
Speech
Recognition
Synthesis
Vision
Classification
Object detection
OCR
Documents
Information extraction
Generative
Text
Imagesflowchart LR S0["Photo"] --> S1["Question"] --> S2["Multimodal model"] --> S3["Text answer"]
Key point: Multimodal models accept images as input.
Why the other options are wrong
- A. Azure Monitor describes the image for the model
- Azure Monitor is for observability.
- C. The image was first converted into a video
- There is no conversion to video.
- D. The model was fine-tuned on that exact photo
- No fine-tuning on that photo is needed.