We're making a few videos that highlight word embeddings. Before training word embeddings we figured it might help the intuition if we first trained some letter embeddings. It might suprise you but the idea with an embedding can also be demonstrated with letters as opposed to words. We're demonstrating the results of an experiment here.
If you want to reproduce these steps, you can find the resources (including links to live colab notebooks) on our github page:
[ Ссылка ].
![](https://i.ytimg.com/vi/mWvnlVw_LiY/maxresdefault.jpg)