You can find in this repo some projects I worked on in 2017/early 2018 somehow related to ML and particularly NLP. Most of them are not yet user-friendly and would require some cleaning.

You can find :

• char-RNN : Code for a char-RNN, inspired by the article http://karpathy.github.io/2015/05/21/rnn-effectiveness/ but recoded from scratch.

• from_LSTM_to_automata : project to speed-up a LSTM trained on text by transforming it into a finite-state automaton.

• training_projects : smaller projects I did for training (predicting presidents from wikipedia hyperlinks, predicting the shots of Kobe Bryant, generating names from a certain country etc)