This repo contains Jupyter Notebooks and presentations presented during the meetings of the AI section of Bit Scientific Group @ Cracow University of Science and Technology, where we dive deep into Machine Learning awesomeness.
The notebooks are intended to compose an introductory Machine Learning course, tackling topics ranging from linear regression to various forms of Deep Learning. Their aim is to introduce the reader to the math and techniques behind ML, as well as the most popular programming tools designed with ML in mind, such as Scikit-Learn and PyTorch.
All notebooks are written and tested in Python 3.6.6 and an Unix OS. Their usage is not guaranteed (though probably possible) with other versions of the language and other OS's.
We recommend Anaconda for managing packages and Python environments.
-
Download, install Anaconda and add it to your
$PATH
environment variable. -
Create a new Python environment for your BIT AI excercises
conda create -n bit_ai -- python=3.6.6 jupyter
- Clone the repository and enter the cloned directory
git clone https://github.com/aghbit/BIT_AI.git
cd BIT_AI
- Activate the environment and install required packages
source activate bit_ai
pip install -r requrements.txt
That's it!
Before each meeting of BIT AI, it is always advised to update the repo - you never know what last-minute additions may have been made!
cd BIT_AI
git pull
Remember to acivate your Conda Environment before running the notebooks!
source activate bit_ai
jupyter-notebook
Jupyter Notebooks can be converted into various formats, such as
For that purpose, Makefile
has been provided. To build PDFs from notebooks, you will also need to install LaTeX.
To build pdfs of all notebooks:
make
To build PDFs of only selected notebooks:
make KEY=<an element of the notebooks' path to be matched>
So, for example
make KEY=1_regression
will build PDFs of only the notebooks which contain 1_regression
in their path.
jupyter nbconvert <notebook_path>.ipynb --execute --to latex
pdflatex <notebook_path>.tex