CSE142 Project

Setting up your dev env

Prerequisites:

• Install python 3.7.7 (tensorflow supports python 3.5-3.7) 64 bits to prevent strange errors.
• Install Microsoft Visual C++ Redistributable packages for x64 (2015, 2017, 2019).
• Install virtualenv: python -m pip install virtualenv

Dependencies

For linux or macOs, go to the official doc

• python --version needs to be Python 3.5-3.7
• pip3 --version needs to be >= 19
• virtualenv --version should be installed (version used for this project: 20.0.18)

Virtual environment

Open a terminal in this folder.

• virtualenv --system-site-packages -p python ./venv will create a new folder "/venv"
• .\venv\Scripts\activate activate the env. Now when you'll use python commands, they will have effect on that environment.
• You can upgrade pip if needed: pip install --upgrade pip
• pip list shwos the packages installed in the environment
• deactivate to exit the virtual environment

Tensorflow

Make sure that you are in the virtual environment.

• Run pip install --upgrade tensorflow to install tensorflow. This project works with tensorflow 2.1
• Verify the installation: python -c "import tensorflow as tf;print(tf.reduce_sum(tf.random.normal([1000, 1000])))"

Required packages

Make sure that you are in the virtual environment.

• Install requirements: pip install -r requirements.txt

Run the code using your GPU (if you've got a good one)

• Follow these instructions and install Cuda 10.1 and not 10.0 or 10.2!!
• Instructions on how to install cuDNN because it's not written in the given website: doc nvidia

Mess around with the program

Make sure that you are in the virtual environment.

Jupyter notebook:

• ipython kernel install --user --name=.venv install a kernel inside the environment, to use to run in the Jupyter notebook there
• Run jupyter notebook: jupyter notebook (and not python -m jupyter notebook!)
• Open dcgan.ipynb. Make sure that the kernel is selected when you open it (.venv). To change it, go to "Kernel", and then "Change kernel".

Google collab

• Simply import the .ipynb file

Data

the data comes from this source