easy5 pts · first answer countsllm-decoding

What sampling temperature does

When generating text from a language model, what happens as you lower the sampling temperature towards 0?

A.Output gets more random, spreading probability evenlyB.Output gets more deterministic, favouring the top tokensC.The model writes longer replies on averageD.The context window shrinks to save GPU memory

Discussion 0

Join free to post in the discussion.

What sampling temperature does

Discussion 0

Related questions