Typical Sampling Quiz

Answers and explanations appear as you take the quiz.

1. What core quantity does typical sampling compare each candidate token against?
   • The expected information content (entropy) of the distribution
   • The model's overall vocabulary size
   • The token's position in the sentence
   • The temperature parameter alone
2. How does typical sampling treat the single most probable token?
   • It always selects it first
   • It may exclude it if its surprisal is far below the expected value
   • It doubles its probability
   • It ignores probability entirely
3. Typical sampling was motivated by which idea about human language?
   • Humans always choose the rarest possible words
   • Humans never repeat words
   • Humans communicate near a roughly steady rate of information
   • Humans speak in fixed-length sentences
4. What does the tau parameter in typical sampling control?
   • The learning rate during training
   • The number of layers in the model
   • The maximum sentence length
   • The cumulative probability mass collected into the typical set
5. Compared with greedy decoding, what does typical sampling try to change?
   • It samples from tokens whose surprisal is closer to the distribution’s expected surprisal
   • It always selects the most probable token
   • It guarantees every continuation is factually correct
   • It removes probability scores from decoding
6. In which kind of software is typical sampling commonly exposed to developers?
   • Spreadsheet applications
   • Open-source LLM inference libraries like Hugging Face Transformers
   • Image editing tools
   • Database query engines
7. How does typical sampling rank candidate tokens differently from top-p sampling?
   • By alphabetical order of token strings
   • By token position in the prompt
   • By distance between token surprisal and expected surprisal, rather than probability rank alone
   • By the model’s training loss
8. Which value does typical_p control?
   • The model’s vocabulary size
   • The number of transformer layers
   • The probability that the answer is factually correct
   • The target cumulative probability mass collected in the typical set