問題文
A team asks the same model the same question about test effort on separate occasions and receives noticeably different answers each time. The team wants the answers to vary less. Which pair of measures does the syllabus recall for this?
選択肢
- Change the output format and adjust the given model role, because these measures reshape the produced answer form during repeated interactions.
- Shorten the prompt length and reduce the supplied example count, because these measures shrink the parsed token burden during repeated interactions.
- Lower the temperature setting and fix the random seed value, because these measures narrow the sampled probability spread during repeated interactions.
- Raise the context window and widen the training data volume, because these measures enlarge the stored factual coverage during repeated interactions.