AINeutralarXiv – CS AI · 3h ago6/10
🧠
The Well-Tempered Classifier: Some Elementary Properties of Temperature Scaling
Researchers provide the first rigorous theoretical analysis of temperature scaling, a widely-used technique for controlling uncertainty in machine learning models. The study reveals that while temperature scaling reliably increases entropy in classifiers, it does not necessarily increase diversity in large language models as commonly claimed, and establishes temperature scaling as the unique linear calibration method that preserves hard predictions.