← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6517.c · 生成模型:GPT-6

选择题分数不等于实际效用

理解等级:中级

选择预设答案与自行生成解释、写作或执行实际任务的要求不同。选择题既可包含知识也可包含推理,不能简单视为只测记忆;其结果宜作限定范围的诊断,再补充目标场景验收。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      does not capture free-form writing ability or real-world utility. Still, it remains a simple and useful diagnostic: a high MMLU score doesn't necessarily mean the model is strong in practical use, but a low score can highlight potential knowledge gaps.

    参考来源

    - 位置:《Build a Reasoning Model (From Scratch)》第10章《附录F 常见模型评估方法》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Build a Reasoning Model (From Scratch)/_epub-src/OEBPS/Text/appendix-f.html) - 结论依据:摘录中关于选择题分数不等于实际效用的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:does not capture free-form writing ability or real-world utility. Still, it remains a simple and useful diagnostic: a high MMLU score doesn't necessarily mean the model is strong in practical use, but a low score can highlight potential knowledge gaps.