← Back to Glossary
Perplexity
A statistical metric measuring how surprised an LLM is by a sample of text — lower scores mean better predictions.
Why it exists
Provides an automated mathematical benchmark for how well a language model captures the distribution of test datasets.
How it works
Calculated as the exponential of cross-entropy loss over text. A perplexity of 10 means the model is as uncertain as choosing among 10 equally likely words.
Related terms
Where you'll use this
roadmap