PyTorch version of my Deep Learning class

Slides

• Convolutional Neural Networks for Object Detection
• Convolutional Neural Networks for Semantic and Instance Segmentation
• Recommender Systems
• NLP - Word Embeddings
• NLP - Language Modeling with Recurrent Neural Networks
• NLP - Attention Mechanisms and Transformers

Notebooks

Either:

git clone https://github.com/ogrisel/dlclass-pytorch
cd notebooks
jupyter lab

or via google colab:

• https://colab.research.google.com/github/ogrisel/dlclass-pytorch

You can also import individual notebooks from the https://github.com/ogrisel/dlclass-pytorch repository into other notebook services such has Kaggle Code for instance.

Credits

The original version written in collaboration with Charles Ollion with the tensorflow / Keras library is still available at:

https://m2dsupsdlclass.github.io/lectures-labs/

Compared to the original repo, this repo has:

• a subset of the slides and notebooks, slightly updated to use PyTorch
• a new notebook about GPT models.

Contrary to the original repo, all the notebooks should download their data files from online locations instead of expecting files in the same folder. Similarly the solutions to the exercises are now included inline. This should make it easier to use GPU-enabled notebook services such as Google Colab or Kaggle Notebooks.