Evaluation metric for language models based on how surprised the model is by a held-out token sequence.
AI Glossary
Perplexity
Coverage in development
Evaluation metric for language models based on how surprised the model is by a held-out token sequence.
Definition
Plain English explanation
Perplexity scores how well a LM predicts text.
Related glossary terms
Last reviewed
Sources
- NIST AI Risk Management Framework — risk vocabulary context for AI systems
- Brel Digital company and technology profiles — applied usage evidence where tagged
Correction request
If a technology assignment or hub description is inaccurate, submit a correction via the Corrections Policy.
All technologies → · AI Models → · APIs & SDKs → · Integrations → · Compliance → · Browse all companies → · Explore industries → · Compare →