Using Fine-tuned LLMs to Grade Homework
Alternative Title
Using Fine-tuned Large Language Models (LLMs) to Grade Homework
Contributor
University of Central Florida. Faculty Center for Teaching and Learning; University of Central Florida. Division of Digital Learning; Teaching and Learning with AI Conference (2025 : Orlando, Fla.)
Location
Gold Coast I/II
Start Date
28-5-2025 4:00 PM
End Date
18-5-2025 4:25 PM
Publisher
University of Central Florida Libraries
Keywords:
Fine-tuning; LLMs; Homework grading; Prompt engineering; Educational technology
Subjects
Grading and marking (Students)--Computer-assisted instruction; Machine learning--Study and teaching; Grading and marking (Students)--Computer programs; Artificial intelligence--Educational applications; Natural language generation (Computer science)
Description
LLMs have the potential to improve education by automatically grading homework and giving students hints. However, due to hallucinations and general lack of capabilities, using LLM for grading and giving hints has mixed results. The performance of LLMs can be improved by using methods such as prompt engineering, RAG (Retrieval-Augmented Generation), and fine-tuning. In this talk, we explore the possibility of using fine-tuned LLMs to grade and give hints. Fine-tuning an LLM allows one to take a pretrained LLM, such as those built by OpenAI, and adapt the LLM for a highly specific purpose.
Language
eng
Type
Presentation
Format
application/vnd.openxmlformats-officedocument.presentationml.presentation
Rights Statement
All Rights Reserved
Audience
Faculty, Students, Instructional designers
Recommended Citation
Bohacek, Stephan, "Using Fine-tuned LLMs to Grade Homework" (2025). Teaching and Learning with AI Conference Presentations. 61.
https://stars.library.ucf.edu/teachwithai/2025/wednesday/61
Using Fine-tuned LLMs to Grade Homework
Gold Coast I/II
LLMs have the potential to improve education by automatically grading homework and giving students hints. However, due to hallucinations and general lack of capabilities, using LLM for grading and giving hints has mixed results. The performance of LLMs can be improved by using methods such as prompt engineering, RAG (Retrieval-Augmented Generation), and fine-tuning. In this talk, we explore the possibility of using fine-tuned LLMs to grade and give hints. Fine-tuning an LLM allows one to take a pretrained LLM, such as those built by OpenAI, and adapt the LLM for a highly specific purpose.