Question about preproc.py, quote sub is different to fasttext get-wikimedia.sh #1
Description
I think I found a bug in the preproc.py script for string normalization here: https://gist.github.com/bittlingmayer/7139a6a75ba0dbbc3a06325394ae3a13#file-ft_wiki_preproc-py-L17
The Python script appears to replace the double-quote with a space, but the original sed surrounds the double quote with spaces. Is this an error?
Unrelated, (but since I have your attention ;), do you know whether the same preprocessing rules were used for the common crawl models? I'm asking specifically about the cc.en.300.bin fasttext model.