DeepNLP-models-Pytorch

Why we included this project

This set of Jupyter notebooks rebuilds the classic deep learning models behind modern NLP by hand, using plain PyTorch instead of wrapping everything in a high-level library. Each notebook pairs a model with the paper it follows, working through word embeddings like skip-gram and GloVe, RNN language models, neural machine translation with attention, and text CNNs plus recursive networks for sentiment. The sequence mirrors the Stanford CS224n syllabus, so the order of topics tracks how the course builds understanding. One honest caveat: the code targets PyTorch 0.2-era APIs and Python 3.5, so it is a study reference rather than production-ready software, and you will need to adapt older calls to run it on current PyTorch. Students and researchers who want to read concrete implementations of core NLP ideas and compare how each model was first formulated will find it most useful.