👉 Temperature computing is a method used in machine learning, particularly in generative models like language models, to control the randomness or creativity of generated outputs. It adjusts the probability distribution of possible next steps or words by scaling the logits (raw model outputs) before applying a softmax function. A higher temperature increases the randomness, making the model more likely to explore less probable but potentially more creative or diverse outputs, while a lower temperature makes the model more deterministic and focused on generating outputs that closely match the training data. This mechanism allows for dynamic control over the generation process, enabling the model to balance between consistency and novelty in its outputs.
