aicaffeinelife/cuhackit_demo

Repository containing the Jupyter Notebook for CUHackit.

CUHackit_Demo

This repository contains the jupyter notebook and slides exported to PDF for reference. The code is there for you to tinker with and make the baseline network better.

Setting up a PyTorch Development Environment

For the purposes of this talk, we'll take advantage of the Palmetto Supercomputing Cluster. Convieniently, we have JupyterHub available to us for exploring jupyter notebooks. However, before we use that, we need to get our environment up and running.

Palmetto Setup:

Login to your palmetto login node and load the following modules:

1) anaconda/5.1.0 

2) cuda-toolkit/9.0.176   

3) cuDNN/9.0v7

You also want to create (or edit) a .jhubrc file and write these exact same lines to that file as well. Now our dev environment is ready to be created. Create a new conda environment like so:

conda create -n <my-name> python=3.5 

The <my-name> is to be replaced by a name of your choice. The name will be important in how you refer to your environment. Finally, we want to install ipython and then a kernelspec (this tells Jupyter to refer our anaconda environment). The kernel spec can now be installed like so:

python -m ipykernel install --user --name <my-name> --display-name <some-display-name> 

Once all of this is ready, we are all set to install PyTorch and then run our notebook. Simply clone this repo in your home directory and we're all set to go.