It depends where you fall between 'wanting to do something with models like GPT-2' and 'wanting to understand where we are at with NLP'. If it's the first, you

It depends where you fall between 'wanting to do something with models like GPT-2' and 'wanting to understand where we are at with NLP'. If it's the first, you should look at HuggingFace, which package transfer learning in a very accessible way to use. But I'd say that's probably too a big leap from no experience at all to that. I wrote a detailed article on how to get up to speed with NLP - tracing the path I have gone through. My main thing was getting quickly to the point where I could actually use this stuff: https://towardsdatascience.com/learn-nlp-the-practical-way-b854ce1035c4 https://towardsdatascience.com/learn-nlp-the-practical-way-b...