Is Temperature the Creativity Parameter of Large Language Models?
This work addresses the problem of understanding and controlling creativity in LLMs for users and researchers, showing that temperature's role is more nuanced and weak than commonly claimed, which is incremental in clarifying a specific aspect of LLM behavior.
The paper investigated whether the temperature parameter in large language models (LLMs) acts as a 'creativity parameter' by analyzing narrative generation tasks, finding that temperature is only weakly correlated with novelty and moderately with incoherence, with no significant effects on cohesion or typicality, and overall leads to slightly more novel outputs at higher temperatures.
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.