This course is part of a series of modules for data science. This course assumes you have a decent basis in Python and experience with something similar to this Data Analyses & Visualisation course https://github.com/raoulg/MADS-DAV
The lessons can be found inside the notebooksfolder.
The source code for the lessons can be found in the src folder.
The book we will be using is Understanding Deep Learning. It is available as pdf here: https://udlbook.github.io/udlbook/ but it is highly recommended to buy the book.
├── README.md <- This file
├── .gitignore <- Stuff not to add to git
├── .lefthook.yml <- Config file for lefthook
├── pyproject.toml <- Human readable file. This specifies the libraries I installed to
| let the code run, and their versions.
├── data
│ ├── external <- Data from third party sources.
│ ├── processed <- The processed datasets
│ └── raw <- The original, raw data
│
├── models <- Trained models
│
├── notebooks <- Jupyter notebooks. Naming convention is xx_name_of_module.ipynb where
│ xx is the number of the lesson
├── presentations <- Contains all powerpoint presentations in .pdf format
├── references <- background information
| └── codestyle <- Some code Code style standards
| └── leerdoelen <- Learning goals per lesson, including pages to read and videos to watch
│
├── reports <- Generated analysis like PDF, LaTeX, etc.
└── figures <- Generated graphics and figures to be used in reporting
For this project you will need some dependencies.
The project uses python 3.11 or 3.12, and all dependencies are defined within the pyproject.toml file.
The .lefthook.yml file is used by lefthook, and automatically lints & cleans the code before I commit it. Because as a student you probably dont commit things, you can ignore it, but you might want to use lefthook in your own repositories.
I have separated the management of datasets and the trainingloop code. You will find them as dependencies in the project:
- https://github.com/raoulg/mads_datasets all code for datasets and dataloaders
- https://github.com/raoulg/mltrainer all code for the training
Both of these dependencies will be used a lot in the notebooks; by separating them it is easier to use the code in your own repositories. In addition to that, you can consider the packages as "extra material"; the way the pacakges are set up is something you can study if you are already more experienced in programming.
In addition to that, there is a codestyle repo that covers most of the codestyle guidelines for the course. You will be expected to follow the codestyle guidelines for your own coding.
The installation guide assumes a UNIX system (os x or linux). If you have the option to use a VM, see the references folder for lab setups (both for azure and surf, pick the one you are provided with). For the people that are stuck on a windows machine, please use git bash whenever I refer to a terminal or cli (command line interface).
please note that uv might already be installed on your VM.
- check if uv is already installed with
which uv. If it returns a location, eg/Users/user/.cargo/bin/uv, uv is installed. - else, install uv with
curl -LsSf https://astral.sh/uv/install.sh | shfor macOS and Linux. For windows, see the uv docs. - start a new terminal and test with
which uvif the installation was succesful.
run in the cli:
git clone https://github.com/raoulg/MADS-MachineLearning-course.git
git config --global user.name "Mona Lisa"git config --global user.email "[email protected]"
cd MADS-MachineLearning-course/uv sync
If you want easy access to the VM, add your ssh key to the VM:
- copy your local ssh key to the VM, see github docs and/or the azure manual.
cd ~/.sshnano authorized_keys(or, if you know how to use it, install and usenvim) copy paste your key to the end of the file (leave existing keys) and save the file, then exit- check with
cat authorized_keysthat your key is added.
Get yourself familiar with the toolbox, if you arent already
- uv: either read the docs uv projects docs or watch a video on youtube, eg uv for python
- git: do a tutorial or watch a video
- terminal/unix: get familiar with the terminal and the unix commands, again video or some tutorial