In late 2017 I got enough of a feel that NLP was moving fast to figure I had better get acquainted with it. I had a technical background, but nothing in NLP.
About two years later, I wrote an article with a practical roadmap to learning NLP [1].
A few months later I write a no-maths explainer to transformer models (BERT and GPT-2.) [2] GPT-3 is basically the same but bigger. ChatGPT is a little different).
Recently, I wondered what, if anything, I'd change about these, given where we are with NLP today. The only thing I'd say is getting acquainted with the HuggingFace ecosystem.
In late 2017 I got enough of a feel that NLP was moving fast to figure I had better get acquainted with it. I had a technical background, but nothing in NLP. About two years later, I wrote an article with a practical roadmap to learning NLP [1].
A few months later I write a no-maths explainer to transformer models (BERT and GPT-2.) [2] GPT-3 is basically the same but bigger. ChatGPT is a little different).
Recently, I wondered what, if anything, I'd change about these, given where we are with NLP today. The only thing I'd say is getting acquainted with the HuggingFace ecosystem.
[1] https://towardsdatascience.com/learn-nlp-the-practical-way-b...
[2] https://towardsdatascience.com/the-abba-explainer-to-bert-an...