easy5 pts · first answer countstransformerstrainingevaluation

Reading a perplexity number

Model A has a perplexity of 8 on a held-out test set and model B has 15 on the same set with the same tokenizer. What does this tell you?

A.B predicts the test text better because higher perplexity means more knowledgeB.A writes text that is about twice as long as B'sC.A predicts the test text better; it is less surprised by each next tokenD.Nothing, because perplexity only measures training speed

Discussion 0

Join free to post in the discussion.

Reading a perplexity number

Discussion 0

Related questions