Q11 — AWS AIF-C01 Ch.2

Question 11 of 100 | ← Chapter 2

When working with foundation models (FMs), which option is beneficial for performing continual pretraining?

• A. Helps reduce model complexity
• B. Model performance improves over time ✓
• C. Reduces training time required
• D. Optimizes model inference time

Correct Answer: B. Model performance improves over time

Explanation

Continual pretraining enables foundation models to absorb new domain-specific terminology, patterns, and knowledge over time without altering architecture. By exposing the model to fresh, relevant data streams, continual pretraining enhances its understanding and generation capabilities within evolving domains—leading to measurable improvements in downstream task performance (e.g., accuracy, coherence, relevance). While it does not inherently reduce model complexity, training time, or inference latency, sustained performance gains are the primary benefit. Thus, option B correctly identifies the core advantage of continual pretraining.