Decoding and inference behavior · MLD-1215
Temperature
A scale applied to logits before normalization to sharpen or flatten a distribution.
Original reference page · detailed expansion pending. Original definitions, examples, notes and citations are preserved below. New examples are illustrative, not benchmark results.
Original example
Lower temperature to favor high-scoring tokens more strongly.
Original caveat
Temperature zero is usually an implementation convention for greedy selection.
Context and aliases
Level: Intermediate. Stages: Inference · Evaluation · Deployment.
Tags: SLM · LLM · NLP.
Aliases: Sampling temperature.
Related concepts
Concepts linking here
Original references
- Hugging Face: generation strategies ↗ · Reference link
- Fast Inference from Transformers via Speculative Decoding ↗ · Consulted
- vLLM documentation ↗ · Consulted
- Original reference URL ↗