ECE498BH: LLM Reasoning for Engineering (Spring 2025)

Course Information

• Instructor: Bin Hu (binhu7@illinois.edu)

• Term: Spring 2025

• Office Hours: T 2-3pm, CSL 145

• Lectures: Tu/Th 9:30-10:50am, 2100 Sidney Lu Mech Engr Bldg

• TA: Xingang Guo (xingang2@illinois.edu), Aaron Havens (ahavens2@illinois.edu)

• For a complete syllabus, see here.

Course Description

This course explores the cutting-edge intersection of large language models (LLMs) and machine reasoning, with a specific emphasis on their transformative potential in engineering disciplines. Modern LLMs, such as GPT, Claude, Gemini, and Llama, are foundation models with vast knowledge bases. These models have demonstrated significant potential in solving complex reasoning and coding tasks. But what do they offer engineers? This course addresses that question by examining LLM reasoning and its application to a range of engineering fields, including control systems, circuit design, power systems, signal processing, aerospace, and transportation engineering. Key topics include: How do LLMs function? How can they be leveraged for reasoning? What is the quality of LLM-generated reasoning for various engineering tasks? How much can we trust the engineering design solutions from LLMs? What are the fundamental limitations of LLM reasoning for engineering? What engineering benchmarks exist for evaluating LLM capabilities? How can LLM reasoning be integrated with domain-specific tools to create LLM agents? Finally, what are the future directions for building even more powerful reasoning machines for engineering applications? In addition to lectures, students will present the latest research papers, and team up to work on course projects.

Required Materials

There is no required textbook for the class. All course material will be presented in class and/or provided online. Links for relevant papers will be listed in the resourse section of the course website.

Prerequisites

ECE313 (or equivalent); Math 257 (or equivalent); the students are also recommended to have some background in one of the engineering domains (such as electrical, computer, mechanical, aerospace, civil, chemical, material,etc) and Python programming.

Grading

We will handle assignments and project reports via Gradescope. Use entry code YR376G to add the course on Gradescope.

• Grades for the students in Section P3 (3 credits) will be weighted as follows: Class Participation (5%), Benchmarking Assignment (35%), and Final Project (60%).

• Grades for the students in Section P4 (4 credits) will be weighted as follows: Class Participation (5%), Benchmarking Assignment (20%), Paper Review/Presentation (15%), and Final Project (60%).

• Benchmarking Assignment: All the students will be asked to work on choosing a specific engineering topic and benchmarking the capabilities of LLMs on that chosen topic. Detailed guidelines will be given in class.

• Final Project: All the students will be asked to work on one final project that combines LLM reasoning with their own engineering background. The grade of the final project is weighted as: Identifying Team Member (2%)+ Identifying Topic (8%)+ Midway Milestone Report (10%) + Midway Milestone Presentation (5%)+Final Presentation (15%)+ Final Report (20%). The NeurIPS template will be used for final report. Detailed guidelines will be posted on 02/27/2025.

• Paper Review and Presentation: The students in Section P4 (4 credits) are required to work on a paper review/presentation assignment. Detailed guidelines will be given soon.