Closed djukicn closed 3 years ago
This is a really nice notebook, but it is a bit long imho. I would suggest we break it into three parts (embeddings, word to doc similarity, maps). Perhaps there could also be a file with precomputed embeddings to avoid running the entire embeddings workflow just to get maps?
Just an idea. :)
I like this notebook too and I also agree with @ajdapretnar to split this notebook into three notebooks.
I also added requirements.txt file in the examples directory so please just add requirements required in this notebook.
Added a notebook showing how to represent documents as vectors and a few use cases of the obtained vectors.