lectures

Why we included this project

Oxford's Deep NLP course ran in Hilary Term 2017, organized by Phil Blunsom with the DeepMind Natural Language Research Group, and this repo holds the complete set of lecture slides and materials from it. The teaching staff included researchers who did much of the early work in neural NLP, and the slides reflect how they approached the subject: recurrent networks for language modeling first, then conditional models for transduction tasks like machine translation and speech-to-text, and finally systems that combine those pieces for question answering. Because the topics build in that order, the collection works as a coherent curriculum rather than a random stack of decks. The linked practicals walk you through implementing word2vec and text classification yourself. Treat it as a study path and reference for the foundations of neural NLP, not as software you can run.