Plan-and-Solve Prompting Quiz

Answers and explanations appear as you take the quiz.

1. What core change does Plan-and-Solve prompting make to standard zero-shot chain-of-thought?
   • It first asks the model to devise a plan, then execute it step by step
   • It fine-tunes the model on planning data
   • It removes step-by-step reasoning entirely
   • It requires dozens of worked examples in the prompt
2. Which specific weaknesses of plain chain-of-thought was Plan-and-Solve designed to address?
   • Slow inference speed and high memory use
   • Missing reasoning steps and calculation errors
   • Toxic or biased outputs
   • Inability to produce any reasoning at all
3. What does the enhanced PS+ variant add to the basic Plan-and-Solve prompt?
   • A second language model to grade the answer
   • Access to a calculator API
   • Instructions to extract variables, compute intermediate results, and attend to numbers
   • A requirement to answer in JSON
4. A key advantage of Plan-and-Solve over few-shot chain-of-thought is that it:
   • Always achieves perfect accuracy
   • Requires retraining the model
   • Only works on vision tasks
   • Needs no hand-crafted worked examples in the prompt
5. On which type of benchmark was Plan-and-Solve prominently evaluated?
   • Arithmetic and math word-problem benchmarks like GSM8K
   • Image classification datasets
   • Speech recognition corpora
   • Protein-folding tasks
6. Why does generating a plan before the detailed steps tend to help an autoregressive model?
   • It deletes the original question from context
   • The plan's subgoals condition the later reasoning tokens the model produces
   • It doubles the model's parameter count
   • It switches the model into a non-text mode
7. A prompt asks a model to list subgoals before solving a word problem. Which Plan-and-Solve stage does that implement?
   • A few-shot worked example
   • The final answer verification stage
   • The planning stage that decomposes the task before execution
   • A decoding-time beam search
8. A PS+ prompt reminds the model to extract quantities and calculate intermediate results. What is the intent of those reminders?
   • Replace the problem statement with demonstrations
   • Change the model’s parameters during inference
   • Guarantee every answer is correct
   • Reduce calculation and step omissions by directing attention to variables and intermediate quantities