In this course we will be running Python code in Jupyter notebooks. We will be using the Keras package for all of our work with neural networks. Keras runs on top of Tensorflow - to get the best performance from Tensorflow we will want to use the GPU version of Tensorflow to leverage the computational power provided by GPUs.
The big problem with this set up is that your laptop may not have necessary GPU hardware to run Tensorflow on a GPU. If your laptop is not sufficiently powerful to run Tensorflow I have provided a number of alternative resources that you can use.
There are two specific cases where you may be able to install Tensorflow with GPU support on your own laptop.
The first case is a Mac laptop with an M1 or M2 chip. Here are instructions on how to install Tensorflow on a Mac with an Apple Silicon chip.
The second case is a Windows laptop or desktop with a discrete NVidia graphics card installed. Here are instructions on how to install Tensorflow on a Windows laptop with NVidia hardware.
I have set up a couple of computers in Briggs 419 with the necessary software installed. I have created separate user accounts for everyone in the class on these machines. To log in to one of these machines you will use your Lawrence ID as your password to log in.
Later in the course when we need access to more powerful hardware I will also provide access on a couple of more powerful machines that you will be able to log into over the network. I will provide instructions on how to use those machines at a later date.
Google Colab is a service provided by Google that allows you to easily run Jupyter notebooks on a server provided by Google. Google offers a free service tier that provides access to Colab on a less powerful server. The free tier should be sufficient for you to complete most of the assignments in this class.
To access this service go to colab.google in your browser. To open a notebook that I have provided, you can download the notebook to your computer from the course web site. Then click the New Notebook button on the Colab landing page and select the Upload Notebook command from the File menu in the notebook to upload the notebook to Colab.