Is Temperature the Creativity Parameter of Large Language Models?

arXiv:2405.00492v1 Announce Type: new
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 …

abstract arxiv creative creativity cs.ai cs.cl diverse language language models large language large language models llm llms plagiarism randomness tasks type