Skip to content
Quantoros

What are common failure modes in prompting and how do you debug them?

LLM & AILevel: IntermediateProduction Prompt Design

What are common failure modes in prompting and how do you debug them?

Hints
  • Identify the stage: pretraining, adaptation, generation or evaluation.
  • Name the inputs, constraints and a case where the approach fails.

Answer

FailureDiagnosisFix
Wrong formatModel ignored format instructionsAdd few-shot format example; use JSON mode
Too verboseNo length constraint"Answer in 2–3 sentences maximum"
Refuses valid requestsOver-cautious safety trainingReframe request; clarify legitimate use case in system prompt
Inconsistent across inputsPrompt ambiguous or too shortAdd examples that cover edge cases
HallucinatingRelying on parametric knowledgeAdd "only use provided context; say 'I don't know' if unsure"
Language wrongDefault language not specifiedSpecify target language explicitly

Source and licence: JustInCache & contributors — Awesome AI Interviews · Apache-2.0

Related questions