September 11th
What is Data Science from a technical perspective? Cutting through the BS and marketing to what is actually useful in Social Sciences.
Dr. Kenton Murray will talk about the fundamentals of modern AI systems. The readings are particularly technical and it is not expected that students without an engineering or math background will understand everything. However, it is expected that you will have at least tried to read through it and come to class with questions about what you did not understand.
- We will start by discussing the Transformer paper (Vaswani et al., 2017). This describes a Machine Learning Model (specifically, it is a neural network architecture). It is the “T” in “GPT”.
- This paper was very important, but also dense, so many people learn the details of it through The Annotated Transformer blog post (Rush, 2018).
- Finally, we will read the GPT-1 paper (which was actually never pubished aside from the researchers’ peronal websited (Radford et al., 2018).
This class and its readings are also mandatory for one of your four relections. It is due before class in two weeks (no class next week - enjoy!).