知识卡片

普通读书笔记卡 · 6634.a · 生成模型:GPT-6

模型家族选择仍需任务实测

理解等级:中级

编码器可用于分类或嵌入,编码器解码器可用于条件生成,仅解码器可统一多种提示任务。这些是用途线索,不是某骨架永久过时或某任务必选某类型的定理;已有满足要求的方案不必只因潮流迁移。

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

      The creators of the UL2 language model claim that encoder-only models should be considered obsolete. I personally wouldn’t go that far; encoder-only models are still great choices for classification tasks. Moreover, if you already have a satisfactory pipeline for your use case built around encoder-only models, I would say if it ain’t broke, why fix it?

    参考来源

    - 位置:《Designing Large Language Model Applications》第04章《架构与学习目标》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch04.html) - 结论依据:摘录中关于模型家族选择仍需任务实测的定义、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:The creators of the UL2 language model claim that encoder-only models should be considered obsolete. I personally wouldn’t go that far; encoder-only models are still great choices for classification tasks. Moreover, if you already have a satisfactory pipeline for your use case built around encoder-only models, I would say if it ain’t broke, why fix it?