Learn PyTorch with project-based tutorials. These tutorials demonstrate modern techniques with readable code and use regular data from the internet.
Applying recurrent neural networks to natural language tasks, from classification to generation.
- Classifying Names with a Character-Level RNN
- Generating Names with a Character-Level RNN
- 🔥 Translation with a Sequence to Sequence Network and Attention
- WIP Sentiment Analysis with a Word-Level RNN and GloVe Embeddings
- WIP Sentence Similarity with a Word-Level Autoencoder
- WIP Intent Parsing with Recursive Application of Recurrent Neural Networks
I assume you have at least installed PyTorch, know Python, and understand Tensors:
- http://pytorch.org/ For installation instructions
- Deep Learning with PyTorch: A 60-minute Blitz to get started with PyTorch in general
- jcjohnson's PyTorch examples for a wide and deep overview
- Introduction to PyTorch for former Torchies if you are a former Lua Torch user
You should know about Recurrent Neural Networks and how they work:
- The Unreasonable Effectiveness of Recurrent Neural Networks shows a bunch of real life examples
- Deep Learning, NLP, and Representations for an overview on word embeddings and RNNs for NLP
- Understanding LSTM Networks is about LSTMs specifically but also informative about RNNs in general
And for more, read the papers that introduced these topics:
- Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation
- Sequence to Sequence Learning with Neural Networks
- Neural Machine Translation by Jointly Learning to Align and Translate
- A Neural Conversational Model
The quickest way to run these on a fresh Linux or Mac machine is to install Anaconda:
curl -LO https://repo.continuum.io/archive/Anaconda3-4.3.0-Linux-x86_64.sh
bash Anaconda3-4.3.0-Linux-x86_64.sh
Then install PyTorch:
conda install pytorch -c soumith
Then clone this repo and start Jupyter Notebook:
git clone http://github.com/spro/practical-pytorch
cd practical-pytorch
jupyter notebook
If you have ideas or find mistakes please leave a note.