Question 117
CCA-F is a newly released exam. These practice questions reflect our best current understanding — the exact wording and question formats may change in the live exam until the certification matures. Use them to learn the concepts, not to memorize.
Scenario: Context Management and Reliability - A document-extraction pipeline reports 97% overall accuracy and the team wants to fully automate it. A spot check reveals one document type (handwritten forms) has roughly 40% errors, hidden inside the aggregate. What practice would have surfaced this risk?
- AAnalyze accuracy by document type and field, and use stratified random sampling rather than relying on the aggregate metric.
- BTrust the 97% aggregate and automate everything.
- CIncrease `max_tokens` for all extractions.
- DAsk the model to self-rate its confidence and trust it.