Introducing Background Temperature to Characterise Hidden Randomness in Large Language Models — arXiv2