HomeData Science ToolsLLM temperature Data Science ToolsBy RiyaSeptember 27, 20261 min read LLM temperature Temperature rescales an LLM's logits before softmax: P(token) = softmax(logits / T). Low temperature makes the distribution sharper; high temperature makes it flatter. Please turn on JavaScript to use the interactive simulator. You might also like Maximum water in container problem Blind 75 Problem List Python installation in Windows ShareCopied