Evaluation metrics and uncertainty · MLD-0890
Perplexity
The exponential of average token negative log-likelihood under a specified tokenization and log convention.
Original reference page · detailed expansion pending. Original definitions, examples, notes and citations are preserved below. New examples are illustrative, not benchmark results.
Original example
Measure language-model predictive fit.
Original caveat
Values are not directly comparable across incompatible tokenizers or datasets.
Context and aliases
Level: Intermediate. Stages: Evaluation · Research.
Tags: ML · DL · SLM · LLM.
Aliases: PPL.
Related concepts
Concepts linking here
Original references
- scikit-learn: cross-validation ↗ · Reference link
- Google Machine Learning Glossary ↗ · Consulted
- scikit-learn: evaluation metrics ↗ · Reference link
- Original reference URL ↗