This repository consists the Jupyter Notebooks, and other resources for our Deep Learning From Scratch Working Group, executed at Semagramme, INRIA Nancy in 2022-2023 by yours truly.
Aaaand has subsequently been extended to work in other contexts ...
In 2025, setting up Python envs has never been easier.
All that you have to do is to run the shell file - setup.sh - and you are good to go.
- Step 1: Format your hard drive
- Step 2: Install your preferred Linux distro
- Step 3: ???
- Step 4: Profit?
okbutforrealthough: idk how to run shell scripts on windows. You want to have uv installed in your system.
Then, navigate to the repo and just do a uv sync.
THIS IS REALLY NOT NEEDED but if you want to follow along every step of the way, install Graphviz on your machine.
Run the following command in your terminal (in the project's root directory):
uv run test_setup.pynone of this new-age fancy tools :grumble grumble: tell me what you need in plain text :grumble grumble: !
Have a python env (v3.12+) capable of running notebooks, and have the following libraries installed:
numpy
torch
torchvision
matplotlib
jupyter
scikit-learn
seaborn
pandas
tqdm
My preferred approach would be to have visual studio code running on your machine.
Install the necessary extensions (python, jupyter), and you're good to go.
Open the repository in VS Code.
PS: You could also just code . from within the project's root directory.
Next, open any notebook. Say ./notebooks/1. - Play with tensors.ipynb.
See the "Select Kernel" on the top right:
In some cases, you might be asked to choose what kind of environments. Go for "Python Environments".
Ideally, you should find .venv/bin/python on macOS and Linux, or .venv\Scripts\python on Windows.
Just try executing the first cell (the one with imports) to verify if it works.
That said, you should be good to run the notebook as you prefer. In browser > Jetbrains IDE fwiw.