Please use this identifier to cite or link to this item: https://repository.isls.org/handle/1/12744
Full metadata record
DC FieldValueLanguage
dc.contributor.authorRahimi, Seyedahmad
dc.contributor.authorEsmaeiligoujar, Salah
dc.contributor.authorErcan, Deniz
dc.contributor.authorBabaee, Maryam
dc.contributor.authorGao, Ran
dc.date.accessioned2026-09-15T15:41:33Z-
dc.date.available2026-09-15T15:33:07Z
dc.date.available2026-09-15T15:41:33Z-
dc.date.issued2026
dc.identifier.citationRahimi, S., Esmaeiligoujar, S., Ercan, D., Babaee, M., & Gao, R. (2026). Small vs. large language models: Comparing language models in generating productive failure math problems. In Litts, B. K., DeLiema, D., Lee, C., Krist, S. Mawasi, A., Martin, L., & Kumpulainen, K. (Eds.), Proceedings of the 20th International Conference of the Learning Sciences - ICLS 2026 (pp. 1223-1227). International Society of the Learning Sciences.en
dc.identifier.isbn979-8-9906980-7-9
dc.identifier.issn1819-0138
dc.identifier.urihttps://doi.org/10.22318/icls2026.113185
dc.identifier.urihttps://repository.isls.org/handle/1/12744-
dc.descriptionSHORTen
dc.description.abstractLarge language models (LLMs) such as GPT-4o can generate high-quality math problems but are resource-intensive and difficult to fine-tune. Small language models (SLMs) like Phi-3 may provide a more practical alternative, as they can run locally and be adapted for specific tasks. In this study, we generated 60 Productive Failure (PF) math problems, with half produced by GPT-4o and half by Phi-3. To ensure comparability, we controlled for grade level, Common Core standard, theme, interest, and prior knowledge. Trained raters evaluated the problems across five PF dimensions. Results showed that GPT-4o (M = 13.38; SD = 1.88) outperformed Phi-3 (M = 10.67; SD = 2.73), with a large effect size (d = 1.16). These findings indicate that although GPT-4o currently generates more effective PF problems, Phi-3 shows promising potential. With further fine-tuning and task-specific adaptation, SLMs like Phi-3 may offer a sustainable, cost-effective solution for educational problem generation.en
dc.format.extent1223-1227en
dc.publisherInternational Society of the Learning Sciences
dc.relation.ispartofurn:ISBN:979-8-9906980-7-9
dc.relation.ispartofProceedings of the 20th International Conference of the Learning Sciences - ICLS 2026, pp. 1223-1227
dc.relation.ispartofseriesProceedings of the 20th International Conference of the Learning Sciences - ICLS 2026en
dc.subjectICLSen
dc.titleSmall vs. large language models: Comparing language models in generating productive failure math problemsen
dc.title.alternativeSmall vs. Large Language Models: Comparing Language Models in Generating Productive Failure Math Problemsen
dc.typeconferencePaper
dc.identifier.doi10.22318/icls2026.113185
dc.format.pagerange1223-1227en
dc.title.parentProceedings of the 20th International Conference of the Learning Sciences - ICLS 2026en
dc.bibliographicCitation.conferencenameProceedings of the 20th International Conference of the Learning Sciences - ICLS 2026en
dc.format.firstpage1223en
dc.format.lastpage1227en
Appears in Collections:ISLS Annual Meeting 2026

Files in This Item:
File SizeFormat 
ICLS2026_1223-1227.pdf293.72 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.