Which term is defined as 'a parameter in the LLM that influences the level of randomness or creativity in generated outputs'?

Which term is defined as 'a parameter in the LLM that influences the level of randomness or creativity in generated outputs'?

• A. Specificity
• B. Temperature
• C. False Positive
• D. True Negative

Answer: (A)

Temperature controls randomness in the outputs of a language model. It works by shaping the probability distribution the model uses to pick the next word. When you increase the temperature, the distribution becomes flatter, so the model is more willing to select less likely words, which boosts creativity and variety but can reduce coherence. Lowering the temperature makes the distribution peakier, so the model sticks to high-probability words and produces more predictable, focused responses. Specificity isn’t a standard term for this kind of behavior in language models, and false positive or true negative are terms from binary classification and evaluation, not generation settings. So the parameter described—the one that modulates randomness or creativity—is temperature.

Temperature controls randomness in the outputs of a language model. It works by shaping the probability distribution the model uses to pick the next word. When you increase the temperature, the distribution becomes flatter, so the model is more willing to select less likely words, which boosts creativity and variety but can reduce coherence. Lowering the temperature makes the distribution peakier, so the model sticks to high-probability words and produces more predictable, focused responses.

Specificity isn’t a standard term for this kind of behavior in language models, and false positive or true negative are terms from binary classification and evaluation, not generation settings. So the parameter described—the one that modulates randomness or creativity—is temperature.