Concept

Adjusting Sampling Temperature for Output Diversity

In processes that generate multiple candidate solutions, such as parallel scaling, the sampling temperature can be adjusted as a mechanism to control the diversity of the outputs. A higher temperature increases randomness and variety, while a lower temperature leads to more deterministic and similar results.

0

1

Updated 2026-01-15

Contributors are:

Who are from:

Tags

Ch.5 Inference - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences