Optimizing Temperature for Language Models with Multi-Sample Inference
Abstract
Multi-sample aggregation strategies, such as majority voting and best-of-N sampling, are widely used in contemporary large language models (LLMs) to enhance predictive accuracy across various tasks. A key challenge in this process is temperature selection, which significantly impacts model performance. Existing approaches either rely on a fixed default temperature or require labeled validation data for tuning, which are often scarce and difficult to obtain. This paper addresses the challenge of automatically identifying the (near)-optimal temperature for different LLMs using multi-sample aggregation strategies, without relying on task-specific validation data. We provide a comprehensive analysis of temperature's role in performance optimization, considering variations in model architectures, datasets, task types, model sizes, and predictive accuracy. Furthermore, we propose a novel entropy-based metric for automated temperature optimization, which consistently outperforms fixed-temperature baselines. Additionally, we incorporate a stochastic process model to enhance interpretability, offering deeper insights into the relationship between temperature and model performance.
Cite
@article{arxiv.2502.05234,
title = {Optimizing Temperature for Language Models with Multi-Sample Inference},
author = {Weihua Du and Yiming Yang and Sean Welleck},
journal= {arXiv preprint arXiv:2502.05234},
year = {2025}
}
Comments
ICML2025, 21 pages. Code available at https://github.com/StigLidu/TURN