Troubleshooting and Optimization

• Topic 01

  Diagnosing Poor or Off-Target Outputs

  45 questions
• Topic 02

  Refining Prompts to Fix Common Failure Modes

  45 questions
• Topic 03

  Managing Long Contexts and Truncation Issues

  45 questions
• Topic 04

  Improving Speed, Cost, and Efficiency

  45 questions
• Topic 05

  Iterating Toward Reliable, Repeatable Results

  45 questions

A few questions from this unit, with the answer and a full explanation. The complete bank is available when you start practising.

1. An output was accurate but far too long. Which refinement most directly fixes this?

   • A
     Specify the required length or ask for a concise response
     Correct answer
   • B
     Switch to a more capable model
   • C
     Ask it to answer only from provided sources
   • D
     Add several examples of similar answers
   Explanation

   Excess length is a verbosity failure, so the matching lever is an explicit length or conciseness constraint. Other levers address other modes. Key takeaway: fix over-long output by constraining length directly.

2. A manager insists an output is 'low quality' and wants a more capable model. On review, the content is accurate and useful but addresses a broader question than was asked. What is the best-supported conclusion?

   • A
     It is a formatting failure, so only the layout needs changing
   • B
     It is off-target from an under-specified prompt, so tightening the question is the correct fix, not changing the model
     Correct answer
   • C
     It is a capability failure, so a more powerful model is required
   • D
     It is a hallucination, so every claim must be re-verified from scratch
   Explanation

   Accurate but broader-than-asked content is the signature of off-target output caused by an under-specified prompt. Escalating to a bigger model misattributes the cause and does not address the ambiguity. The remedy is to narrow and clarify the question. Key takeaway: off-target from vagueness is fixed by clarifying the ask, not upgrading the model.

3. You must chunk a long technical specification but several requirements interlock across sections. How do you chunk while protecting those relationships?

   • A
     Chunk along natural boundaries and carry forward a running summary of cross-section requirements into each chunk
     Correct answer
   • B
     Chunk at arbitrary fixed lengths and ignore any overlap
   • C
     Chunk each sentence separately to be thorough
   • D
     Rely on the tool to remember earlier chunks unaided
   Explanation

   Chunking on natural boundaries and passing a running summary of interlocking requirements into each chunk preserves the cross-section relationships that naive splitting would lose, while still fitting the window. Key takeaway: carry a running summary of shared requirements across chunks to preserve interlocking meaning.