Link Search Menu Expand Document
CS 8803

About

Table of contents

  1. About the Class
  2. Prerequisites
  3. Assignments / Grading
    1. Participation and Discussion (50%)
    2. Course project (50%; group of 1~4 students)

About the Class

This is a discussion-based research-oriented class. This is not a lecture-based class (if you prefer lectures, take CS 7650). In each class, students will present and lead discussion of recent research papers on large language model related papers. This class is designed for students who who want to do research, and students will conduct a self-directed research project in this class.

This course will have a heavy workload and will be highly challenging. It assumes that students are already familiar with LLM fundamentals covered in CS 7650 (or an equivalent background gained through self-study or other coursework). The course is intended for PhD students and MS students who have the time, motivation, and technical maturity (e.g., data analysis, creativity, mathematics, programming, technical writing, and critical thinking) to comprehend and conduct research at the level expected by top AI research venues (e.g., ACL, EMNLP, NAACL, ICLR, NeurIPS, COLM).

Research publications: some students may continue on working with the instructor (and TAs) after the class to extend and publish their projects at top ML/NLP conferences! A total of 4 course projects from the Fall 2024 offering of this class have resulted in publications at NAACL 2025, EMNLP 2025, COLM 2025, and one paper currently under submission (listed at the end of this page).

For students on the wait list: The teaching team does not have control over individual students’ enrollment in the class. Permits are issued by the CoC office, prioritizing CS/ML PhD students and MSCS students.)

Prerequisites

This is an advanced-level research-oriented class. The class will require a good understanding of machine learning algorithms (CS 4641/7641), deep learning models (CS 4644/7643), and NLP techniques (CS 4650/7650). For most of the classes, four students will present two assigned research papers in-depth, additional students will be assigned to serve as investigators to lead discussions of their findings about the paper.

Generally speaking, if you want to learn about LLM and want lectures given by an instructor, you should take CS 7650 or other lecture-based AI classes offered at Georgia Tech instead.

Students should have sufficient technical background and feel comfortable reading, presenting, and critically discussing research papers at the level of the following examples before enrolling in this course:

At the same time, if you do have enough background, we do encourage you to take this 8803-LLM class in your first year of your graduate school – especially if you want to complete a paper for publication and/or if you plan to pursue a Ph.D. later.

Assignments / Grading

This course has no midterm or final exams and is primarily based on student-led presentations and discussions, and course projects.

Participation and Discussion (50%)

For most of the classes, we will discuss two research papers, and analyze its strengths and weaknesses.

  • Paper Presentations (2 minutes): 2%
  • Paper Presentations (20-30 minutes): 20%
  • Paper Investigations (1-2 page report, ~10 minutes discussion): 8%
  • Class Attendance: 10%
  • Paper Review: 10%

Each student will be asked to present an assigned paper in pairs on a pre-scheduled date (with occasional groups of 1 or 3, depending on the final enrollment). Presentations will be 20–30 minutes long, with the exact length adjusted based on the paper’s complexity and scope.

In addition, each student will be also assigned as an “investigator” to one other paper (that will be presented by other students) during the semester. The role of the investigator is to look into the datasets used by the authors, challenge the ideas or certain claims in the paper. form questions about the paper and come up with potential answers, run small experiments, etc. The investigators will lead and facilitate the discussion of the papers in the class for the remaining time of the class (10-20 minutes), following the presentations.

Every student will read and write a critique about one of the two papers before the class, and engage in discussions in class.

This class will require in-person attendance. We will take attendance for randomly-selected classes, and deduction will be applied if students miss more than 2 attendance-taking classes.

Course project (50%; group of 1~4 students)

Students will work in group of 1~4 to propose, conduct, give presentations, and write written reports of a self-directed research project. Project topics will be chosen by the students and should relate to large language models (LLMs).

Deliverables:

  • Literature Review: 2%
  • Project Proposal: 2%
  • Midway Report: 5%
  • Final Project Presentation: 10%
  • Final Report: 30%

Rubics:

  • Clarity: For the reasonably well-prepared reader, is it clear what was done and why? Is the report well-written and well structured?
  • Originality / Innovativeness: How original is the approach? Does this project break new ground in topic, methodology, or content? How exciting and innovative is the work that it describes?
  • Soundness / Correctness: First, is the technical approach sound and well-chosen? Second, can one trust the claims of the report – are they supported by proper experiments, proofs, or other argumentation?
  • Meaningful Comparison: Does the author make clear where the problems and methods sit with respect to existing literature? Are any experimental results meaningfully compared with the best prior approaches?
  • Substance: Does this project have enough substance, or would it benefit from more ideas or results?  Note that this question mainly concerns the amount of work; its quality is evaluated in other categories.

Examples (course projects from the Fall 2024 offering of this class):

Given the broader availability of datasets and benchmarks now in 2026, students are encouraged to pursue a wider range of project topics.