AI/ML fundamentals
⏱ ~3-min readAceMark Guide
What this topic is really about
Fine-tuning adapts an existing pre-trained model to a specific task by updating its weights with a smaller, targeted dataset, saving significant time and compute. In contrast, building a model from scratch (Option A) is pre-training, which requires massive datasets and resources to initialize weights from zero.
Supervised learning relies on a training dataset containing labeled examples, where each input is paired with its correct target output to guide the model's learning process. Unlabeled data is instead used in unsupervised learning, where the model must find patterns without explicit guidance.
See the mechanism
Inference is the phase where a fully trained model is deployed to evaluate new, unseen data and generate predictions. A diagram for this topic isn't available yet — the worked example below walks the same reasoning step by step.
An exam-style question, fully explained
Which is an example of inference?
- Identify what the question tests: Which is an example of inference.
- Inference is the phase where a fully trained model is deployed to evaluate new, unseen data and generate predictions.
- In contrast, adjusting hyperparameters or preparing data are tasks performed during the model training and development phase, not during active inference.
Traps the examiner sets
- Read each option carefully — distractors on AI/ML fundamentals are designed to look plausible.
- Re-check the exact wording of the question stem before committing to an answer.
- Watch the qualifiers ("always", "only", "except") that flip a correct-looking option.
Test your recall
Answer each from memory — you'll see instantly whether you're right and why.
Run a focused 10-question mini-mock on AI/ML fundamentals and see it stick.
Practice more of this topic →