I am preparing for interviews for machine learning engineer positions.
So I am brushing up my skills in natural language and pytorch as well. It's been a while since I worked on intensive machine learning model development.
The past few years have been mostly work on my own startup, doing fullstack engineering and platform development.
My plan is to implement a series of classical NLP models in pytorch from scratch.
To make things more appealing, I would also find common datasets on kaggle and see how these models perform on them.
To be more specific, here is a list of models I am considering to implement:
- MLP & logistic regression
- text classification tasks -- RNN, LSTM, biLSTM
- NER task -- biLSTM + CRF
- machine translation -- transformer based architectures