Natural Language Processing Perplexity

You are working a new project where you are building a language model (LM) for a very specific sub-domain. All the sentences are of maximum length of 20 and the size of the vocabulary is 432. After you finish implementing your first LM based on bigrams and you evaluate, you find you have a perplexity of 450. Based only on this information, you find the perplexity value:

  1. Too low but possible
  2. It is impossibly too low
  3. Too high but possible
  4. It is impossibly too high
  5. can’t tell

Sample Solution