presented on 21 Aug 2019 at the MIT-Harvard Theoretical and Computational Neuroscience Journal Club.
Author: Martin Schrimpf [github] [scholar]
Link to notebook: https://nbviewer.jupyter.org/github/mschrimpf/pytorch_tutorial/blob/master/tutorial.ipynb
- You have been exposed to PyTorch, ideally even run it
- When someone sends modeling code to you, you're comfortable going through it
- You have a reference to look things up or start building your own models
If you have conda (suggested), run
conda install numpy matplotlib seaborn jupyter tqdm
conda install pytorch torchvision -c pytorch
You can also pip install "pytorch_tutorial @ git+https://github.com/mschrimpf/pytorch_tutorial",
or pip install by hand.
In general, I also recommend using the PyCharm IDE, but you can use whatever you are comfortable with.
For this tutorial, please have a jupyter notebook server running.
- PyTorch examples, a collection of sample code for vision, text, RL etc.
- fast.ai, a practical deep learning course built with PyTorch
- Andrew Ng's coursera course, the first and still very relevant Machine Learning course with theory and practices