CCAR-F Context and Reliability Practice Question
A financial analyst uses Claude to generate quarterly earnings summaries from raw financial data. The summaries must include precise numerical figures. The analyst notices that Claude occasionally transposes digits when copying numbers from the input. Which architectural change is most effective for reducing this error?
⚠ Common exam trap
The trap here is thinking that fine-tuning or temperature adjustments can fix a specific attention-related error, when prompt restructuring to separate extraction and summarization is more effective.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Use a chain-of-thought prompt that asks Claude to first extract all numbers into a structured list, then generate the summary using that list.
Digit transposition often occurs when the model must simultaneously extract and synthesize information. By using a chain-of-thought prompt to first extract numbers into a structured list, you isolate the extraction task, reducing errors. This also enables programmatic verification. Other options either increase randomness, require costly fine-tuning, or reduce context without targeting the error.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Increase the temperature parameter to encourage more creative output.
Why it's wrong here
Increasing temperature increases randomness and creativity, which would likely worsen numerical accuracy. The goal is to reduce digit transposition errors, so higher randomness is counterproductive. This change would not address the root cause and could introduce more variability in the output, making the summaries less reliable.
- ✓
Use a chain-of-thought prompt that asks Claude to first extract all numbers into a structured list, then generate the summary using that list.
Why this is correct
By separating extraction from summarization, you reduce the cognitive load on the model and create an intermediate artifact that can be verified. This structured approach minimizes the chance of digit transposition because the model focuses on one task at a time. It also allows for automated validation of the extracted numbers against the source data.
- ✗
Fine-tune Claude on a dataset of financial summaries with correct numbers.
Why it's wrong here
Fine-tuning may improve overall performance but is not a targeted solution for digit transposition, which often stems from attention lapses in long contexts. It requires a large, high-quality dataset and may not generalize to new data formats. This approach is expensive and less direct than restructuring the prompt to enforce extraction and verification.
- ✗
Reduce the context window by providing only the most recent quarter's data.
Why it's wrong here
Reducing context might help if the model is overwhelmed, but it does not address the specific error of transposing digits. The analyst likely needs multiple quarters for comparison. This change could omit critical information and does not enforce accuracy in copying numbers. It is a blunt approach that may harm the summary's completeness.
About these practice questions
One of 271 original CCAR-F practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Anthropic exam blueprint
This CCAR-F practice question is part of Courseiva's free Anthropic certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the CCAR-F exam.