Is Temperature the Creativity Parameter of Large Language Models?

TL;DR

This paper empirically investigates whether the temperature parameter in large language models truly controls creativity, finding only weak correlations with novelty and coherence, and suggesting more nuanced approaches for controlling LLM creativity.

Contribution

The study provides a systematic analysis of the relationship between temperature and creativity in LLMs, challenging the notion that temperature is a direct creativity parameter.

Findings

Temperature is weakly correlated with novelty.

Temperature is moderately correlated with incoherence.

No significant relationship between temperature and cohesion or typicality.

Abstract

Large language models (LLMs) are applied to all sorts of creative tasks, and their outputs vary from beautiful, to peculiar, to pastiche, into plain plagiarism. The temperature parameter of an LLM regulates the amount of randomness, leading to more diverse outputs; therefore, it is often claimed to be the creativity parameter. Here, we investigate this claim using a narrative generation task with a predetermined fixed context, model and prompt. Specifically, we present an empirical analysis of the LLM output for different temperature values using four necessary conditions for creativity in narrative generation: novelty, typicality, cohesion, and coherence. We find that temperature is weakly correlated with novelty, and unsurprisingly, moderately correlated with incoherence, but there is no relationship with either cohesion or typicality. However, the influence of temperature on creativity is far more nuanced and weak than suggested by the "creativity parameter" claim; overall results suggest that the LLM generates slightly more novel outputs as temperatures get higher. Finally, we discuss ideas to allow more controlled LLM creativity, rather than relying on chance via changing the temperature parameter.