Thanks for reading the post!

In the notebook we are testing similarities with documents that can have more than just one sentence, so the output will always be the full document.

The input to the fit_transform() must be a list of strings, where each string represents a document (e.g. ['she left her husband . he killed their child . just another day in america .', 'of course it has a song .', ...]) and the output will be one of the documents.

Hope it helps!