Summer Session: Kick Off Meeting

Deep Learning Community of Practice

For Next Meeting

• We’re starting off from zero and implementing the core engine that powers deep learning: back propagation.

• Please watch at least the first 1h22m of Andrej Karpathy’s lecture here and try to follow along in your Jupyter Notebook.

• Please note what you find confusing – someone else probably finds it confusing too!

• Reminder: For this you’ll need to install numpy, matplotlib, graphviz, and pytorch.

  • If you’re using conda you may need to install graphviz using pip. You can do this from your notebook by typing the command !pip install graphviz. The leading ! will cause this to be treated as a shell command rather than a python command.

  • numpy conda pip

  • matplotlib conda pip

  • graphviz conda pip

  • pytorch Install Instructions

Materials

• Here is a tiny neural network in Excel ( File). Download it and try changing the weights and biases of the layers! In practice we’ll change all the weights and biases at once, but this is nice for building intuition.

• Here are the Meeting Slides.