azunre/transfer-learning-for-nlp

Companion repository to Paul Azunre's "Transfer Learning for Natural Language Processing" book

transfer-learning-for-nlp's Issues

Reference Frameworks out of dat

One of the frustrations with the exampes in the book is that they require a particular version of the various frameworks used. I normally allow for this when I buy a book and try to convert the examples so that they work with the latest versions of the frameworks. The learning from the process is normally very useful. I have really struggled to do this with the examples in Chapter 3 for Elmo and Bert. I know kaggle references are provided using the relevant versions of the frameworks, but I have even struggled to get those to run both in Kaggle and in virtual enviroments on my machine. It would be nice to have a migration guide that provides some examples of how these projects could be migrated to current versions of Tensorflow and Keras. I know this is a tall order, but without it the book appears rooted in the past and has limited value.

Which framework will the book use?

I see tensorflow, pytorch, spacy, allennlp and thinc, etc. In the requirements.txt will you be focusing on one framework or will you use multiple frameworks?

Examples with text/datasets in other languages

Hi,

I would like to see the book with examples of QA, Text Extraction, Semantic and etc, using other languages besides english too.

Tks