# Min-p Sampling

> Min-p sampling is a truncation sampling method that keeps only tokens whose probability is at least a fixed fraction of the most likely token's probability, so that the cutoff tightens when the model is confident and loosens when it is not.

- Identifier: PTL-0218
- Category: Foundations
- Canonical URL: https://protologue.com/t/min-p-sampling/
- Also known as: min_p, min-p
- Introduced: 2024

## Description

Unlike top-p sampling, which uses a fixed cumulative-probability mass, min-p scales its threshold by the top token's probability. It is offered as a sampling parameter in open-source inference engines such as vLLM and is usually combined with temperature. Its introducing paper reports better quality and diversity at high temperatures, but a later reanalysis argues that the evidence does not show min-p outperforming established sampling methods.

## Related terms

- [Top-p Sampling](https://protologue.com/t/top-p-sampling/)
- [Top-k Sampling](https://protologue.com/t/top-k-sampling/)
- [Temperature](https://protologue.com/t/temperature/)
- [Text Degeneration](https://protologue.com/t/text-degeneration/)

## Sources

- Nguyen et al. (2024). Turning Up the Heat: Min-p Sampling for Creative and Coherent LLM Outputs. https://arxiv.org/abs/2407.01082
- Schaeffer et al. (2025). Min-p, Max Exaggeration: A Critical Analysis of Min-p Sampling in Language Models. https://arxiv.org/abs/2506.13681
- vLLM (2024). Sampling Parameters. https://docs.vllm.ai/en/v0.6.6/dev/sampling_params.html

## Cite this entry

Protologue. (2026). Min-p Sampling. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0218). https://protologue.com/t/min-p-sampling/

License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)