← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8704.b · 生成模型:GPT-6

降维图只帮助观察局部模式

理解等级:中级

PCA或t-SNE将高维向量投到低维以展示聚集现象。二维距离和簇形受算法及参数影响,不能据图认定高维语义有天然边界或模型理解了现实类别。

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

      We can visualize the embedding space by using dimensionality reduction techniques like PCA or t-SNE. Figure 11-1 depicts the visualization of embeddings of posts on X (formerly Twitter) by members of the US Congress created by Nomic AI using its Atlas tool. You can view a detailed version of the visualization at Nomic’s blog. Let’s explore how we can use embeddings for semantic search. For

    参考来源

    - 位置:《Designing Large Language Model Applications》第11章《表示学习与嵌入》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch11.html) - 结论依据:摘录中关于降维图只帮助观察局部模式的定义、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:We can visualize the embedding space by using dimensionality reduction techniques like PCA or t-SNE. Figure 11-1 depicts the visualization of embeddings of posts on X (formerly Twitter) by members of the US Congress created by Nomic AI using its Atlas tool. You can view a detailed version of the visualization at Nomic’s blog. Let’s explore how we can use embeddings for semantic search. For