CS 7931, University of Utah

This seminar is designed for anyone interested in expanding their knowledge of LLMs and other frontier models.

Instructor:

• Professor ****Ana Marasović ([email protected]) leads the seminar
• Professors Ziad Al-Halah, Daniel Brown, & Vivek Srikumar will (try to) participate
• 7 graduate student volunteers give lectures every other week

When: Wednesdays 11:50-12:50 (remote not available)

Where: Marriott Library 1725 (recordings not available)

[Jump to the seminar schedule.]

Seminar format:

• 1 topic over 2 weeks
• 1st week: A grad student gives a deep dive in the core technical parts of 1-2 recent advances relating to the topic.
• 2nd week: A role-playing paper discussion of some paper on the same topic. The paper will be nominated by the same student and the student will lead the discussion. Students enrolled in the seminar will be assigned roles and must show to the seminar and participate. Please see this for an example of a role-playing seminar: https://utah-explainability.github.io/assignments/paper_discussions/. More details will be shared later.

What we expect you already know: Deep learning, transformer architecture, pretraining-finetuning, neural conditional generation, in-context learning, chain-of-thought prompting.