知识卡片

普通读书笔记卡 · 6698.c · 生成模型:GPT-6

上下文输入与目标名称须固定

理解等级:中级

skip-gram可把一个词作为输入、邻词作为输出;不同资料有时对center、context、target用法不同。应以实际张量流向为准,保持采样对、模型输入和损失目标一致,避免仅根据变量名字解释算法。

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

      For the continuous skip-gram training approach, skip-grams are word pairs that skip over zero to four words to create the skip-gram pair. When training word embeddings using the Word2Vec skip-gram method, the first word in a skip-gram is called the context word—the input to the Word2Vec neural network. The second word in the skip-gram pair is often called the target word—the word that the language

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第06章《用词嵌入推理》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-6.html) - 结论依据:摘录中关于上下文输入与目标名称须固定的定义、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:For the continuous skip-gram training approach, skip-grams are word pairs that skip over zero to four words to create the skip-gram pair. When training word embeddings using the Word2Vec skip-gram method, the first word in a skip-gram is called the context word—the input to the Word2Vec neural network. The second word in the skip-gram pair is often called the target word—the word that the language