Speculative Sampling Verification Quiz

Answers and explanations appear as you take the quiz.

1. What is the core idea behind speculative sampling?
   • Use a small draft model to propose multiple tokens that a large model verifies in parallel
   • Train one giant model to skip layers dynamically
   • Cache every possible sentence in advance
   • Replace sampling with greedy decoding to save time
2. Why does speculative sampling not degrade output quality?
   • Because the draft model is always correct
   • Because the verification rule makes the final distribution identical to pure target sampling
   • Because it only works on short prompts
   • Because it averages the two models' outputs
3. In exact speculative sampling, what happens when a drafted token is rejected?
   • It outputs the rejected draft token unchanged
   • It discards the entire conversation and restarts
   • The algorithm samples from a corrected residual distribution at that position
   • It permanently changes the target model
4. Under exact speculative sampling, what is guaranteed about the output distribution?
   • It is always deterministic
   • It matches the draft model’s distribution instead
   • It is more diverse than every target-model output
   • It matches the target model’s sampling distribution
5. What information does the target verification pass provide in exact speculative sampling?
   • Target probabilities for proposed positions, used for acceptance tests and residual resampling
   • A guarantee that no target-model computation is needed
   • Only the draft model’s next-token distribution
   • A corrected token sequence without any sampling
6. Which description correctly distinguishes two self-speculative approaches?
   • Both use only an unrelated small draft model
   • Medusa adds prediction heads, while EAGLE uses feature-space proposals
   • Medusa is a tokenizer and EAGLE is a reward model
   • Both remove the target model from decoding
7. The draft assigns probability 0.8 to a token and the target assigns 0.4. What is the speculative acceptance probability?
   • 0.2
   • 0.4
   • 0.5
   • 1.0
8. What most limits speculative sampling’s realized speedup?
   • The color of the model’s interface
   • The target model’s name length
   • Whether the prompt contains punctuation
   • The draft-model cost and how much proposed work the target accepts