First, you must have your Python enviroment for deep learning. You can
install miniconda and create that enviroment. See

https://docs.conda.io/en/latest/miniconda.html

For instance, in Ubuntu 20.04, you may want to download
Miniconda3-py39_4.10.3-Linux-x86_64.sh and execute

sh Miniconda3-py39_4.10.3-Linux-x86_64.sh

to install miniconda with Python 3.9 in your 64-bit machine. For that,
you must seacrh Linux Installers and click on <Miniconda3 Linux
64-bit> for Python 3.9.

A new environment for mo434 in conda can be created by

conda create -n mo434

You may use

conda activate mo434

to activate this environment, or 

conda deactivate

to deactivate it.

Once your environment is created and activated, you may install
packages by typing:

conda install <package-name>

Similarly, you may unsinstall packages by typing

conda uninstall <package-name>

You will need the following packages

numpy, scipy, matplotlib, pandas, imgaug, scikit-learn, umap-learn,
scikit-image, opencv-python, pytorch, torchvision, torchaudio,
tensorflow, keras, jupyter, nltk, transformers, etc. You may install
them as needed to run the notebooks of MO434. The installation of some
packages already installs other packages.

For pytorch, it is better to follow instructions at
https://pytorch.org/

For instance, to install it with cuda toolkit 11.1 for execution in
the GPU of your machine, the command will be

conda install pytorch torchvision torchaudio cudatoolkit=11.1 -c
pytorch -c nvidia

For some packages, such as torch-summary and torch-snippets, you might
have to use

pip install torch-summary
pip install torch-snippets

Have fun!