Closed BryanCutler closed 4 years ago
cc @frreiss to please take a look, also @kmh4321 and @MLnick if interested.
@frreiss I had a problem in the notebook trying to align a span column with the token_id of the BERT DataFrame. The span column gets dropped with the DataFrame is saved, but I wanted to show something at the end, but I seem to be missing something to add back in the span column.
@BryanCutler where is the notebook itself? Doesn't seem part of the PR currently.
Oops, sorry @MLnick , it would help if I added the file. Doing it now..
Ok, now the commit with the notebook is showing up. Strange. Will review momentarily.
Thanks for reviewing @frreiss , I'll work on an update
Just a couple minor things picked up so far - still working through the technical details of the notebook. Though agreed with @frreiss that some more description of the problem & goal etc would be useful.
Thanks @MLnick , I will reword things to better state the problem & goal.
@frreiss and @MLnick , I updated a lot of the wording and added more descriptions and links. Hopefully, this better states the problem and goal of the notebook. Please take another look when you can, thanks!
Not sure why the ReviewNB link wasn't added, but here it is
https://app.reviewnb.com/CODAIT/text-extensions-for-pandas/pull/116/
I removed the "Further Analysis" section now, will add it back if I can make it worth while.
Will merge this when tests pass
This notebook uses parts of the CoNLL-2003 tutorial to showcase the use of TensorArray to represent BERT embeddings for model training and prediction results.