1.4 Understand Model Selection in AI Model Hubs and Repositories for Appropriate Use‑cases (e.g., Reasoning, Multimodality) AI-TECHNICAL-PRACTITIONER Practice Questions (Page 3)

Questions 11–15

1. 11expert · hard

   A hospital is building an AI system to assist radiologists. The system must analyze X-ray images and generate a preliminary text report. The hospital also requires the system to explain its findings in a step-by-step manner for the radiologist to review. Which model type should they select?

   • AA multimodal reasoning model that can process images and generate text with a reasoning trace.
   • BA multimodal model that can process images and generate text, but without a reasoning trace.
   • CA vision model that can classify X-ray images.
   • DA text-only reasoning model, because it can generate the report and explain its reasoning.
   Reveal Solution Discussion Select an answer first
2. 12foundation · easy

   Which characteristic is typical of a reasoning model?

   • AIt can only process numerical data
   • BIt is primarily used for image generation
   • CIt is optimized for multi-step logical inference and problem-solving
   • DIt requires no training data
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A healthcare startup is building a tool that lets patients photograph their skin lesions and receive a preliminary description of possible characteristics. The tool must also generate a text summary for the patient. The team is evaluating models from Hugging Face. Which type of model should they choose?

   • AA reasoning-focused model that can think through the image analysis.
   • BA multimodal model that can process both image and text inputs.
   • CA text-only large language model fine-tuned on medical literature.
   • DAn image classification model that outputs a label.
   Reveal Solution Discussion Select an answer first
5. 14foundation · easy

   Which statement correctly describes the primary purpose of Hugging Face as an AI model hub?

   • AIt is a model hub that focuses primarily on computer vision models and does not support text or multimodal models.
   • BIt is a repository for sharing and discovering pretrained models, datasets, and demos, with a strong focus on natural language processing and community collaboration.
   • CIt is a version control system for machine learning code, similar to GitHub, but does not host model weights.
   • DIt is a cloud platform that exclusively hosts proprietary models for enterprise customers, with no community contributions.
   Reveal Solution Discussion Select an answer first
6. 15expert · hard

   A team is selecting a model for a document analysis system that must process both scanned forms (images) and typed text. The system must run on-premises due to data privacy regulations, and the team has a moderate GPU budget. They are evaluating three models from Hugging Face: Model A is a multimodal model with high benchmark scores but requires a large GPU cluster; Model B is a multimodal model with slightly lower scores but runs on a single GPU; Model C is a text-only model with excellent text performance. Which model should they choose?

   • AModel B, because it meets the multimodal requirement and fits the on-premises GPU budget.
   • BModel A, because it has the highest benchmark scores.
   • CModel A, because the team can use cloud GPUs to run it.
   • DModel C, because text-only models are sufficient for document analysis.
   Reveal Solution Discussion Select an answer first