Essentials of prompt engineeringFull notesSummaryRingkasanStoriesPracticeModifying Prompts10 exam-style questions on this lesson.Question 1 of 10A bank's assistant must give consistent, repeatable answers to policy questions. How should it set the temperature?ASet it above the maximum allowedBLow, for more deterministic and focused outputCHigh, for more creative outputDTemperature doesn't affect consistencyCheck answerQuestion 2 of 10A marketing team wants varied, creative slogan ideas. Which change helps?AAdd more stop sequencesBLower the maximum length to 5 tokensCRaise the temperatureDLower the temperatureCheck answerQuestion 3 of 10Which parameter samples only from the smallest set of tokens whose probabilities add up to a chosen value?AMaximum lengthBTop KCTemperatureDTop PCheck answerQuestion 4 of 10A team sets Top K to 10. What happens?AIt samples only from the 10 most likely next tokensBIt caps the output at 10 tokens in totalCIt generates 10 answers and returns the bestDIt sets the temperature to 10 for samplingCheck answerQuestion 5 of 10Responses sometimes run on far too long and cost too many tokens. Which parameter caps this?ATop KBMaximum lengthCTop PDTemperatureCheck answerQuestion 6 of 10A developer wants generation to stop the moment the model writes "END_OF_ANSWER", even if the maximum length isn't reached. Which setting fits?ATemperatureBTop KCStop sequencesDTop PCheck answerQuestion 7 of 10Where should the requested output be placed in a prompt, according to best practices?AIn a separate API callBNowhere; it should be impliedCAt the very beginning, before any contextDAt the end of the promptCheck answerQuestion 8 of 10. Select two.Which TWO are prompting best practices from the course? (Select TWO.)ABreak complex tasks into subtasks or ask the model to think step by stepBAvoid telling the model what the output is forCUse prompt templates for consistent, reliable inputsDNever include examplesEKeep prompts vague so the model is creativeCheck answer0 of 2 selectedQuestion 9 of 10A prompt says "Summarize this in three bullet points under 50 words." What is this an example of?AUsing directives for the response type and lengthBNegative prompting to avoid long outputCA stop sequence that ends the outputDPrompt leaking of the output formatCheck answerQuestion 10 of 10Which parameters are available for a given foundation model?AOnly temperature, on every modelBIt depends on the modelCNone; parameters can't be changedDThe same five parameters on every modelCheck answerUnderstanding Prompts7 exam-style questions on this lesson.Prompt Engineering Techniques8 exam-style questions on this lesson.