← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 6640.c · 生成模型:GPT-6

污染探针只提供调查线索

理解等级:高级

补全题干或改变题中数字可测试记忆与规则遵循的敏感性,但正确补全不证明训练收录,改数后变差也可能来自难度变化。污染结论需结合数据来源与更受控证据,不从一个样本反推全部训练集。

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

      It is suspected that a large number of models may have been trained on data contaminated with the GSM8K dataset. Explore the GSM8K dataset, feed only part of the question, and check if the models you evaluated during the previous exercises correctly complete the question. Also, change the numbers in the problems and verify if the performance remains the same.

    参考来源

    - 位置:《Designing Large Language Model Applications》第05章《按任务适配模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch05.html) - 结论依据:摘录中关于污染探针只提供调查线索的定义、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:It is suspected that a large number of models may have been trained on data contaminated with the GSM8K dataset. Explore the GSM8K dataset, feed only part of the question, and check if the models you evaluated during the previous exercises correctly complete the question. Also, change the numbers in the problems and verify if the performance remains the same.