Please use this identifier to cite or link to this item:
https://repository.isls.org/handle/1/12744Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.author | Rahimi, Seyedahmad | |
| dc.contributor.author | Esmaeiligoujar, Salah | |
| dc.contributor.author | Ercan, Deniz | |
| dc.contributor.author | Babaee, Maryam | |
| dc.contributor.author | Gao, Ran | |
| dc.date.accessioned | 2026-09-15T15:41:33Z | - |
| dc.date.available | 2026-09-15T15:33:07Z | |
| dc.date.available | 2026-09-15T15:41:33Z | - |
| dc.date.issued | 2026 | |
| dc.identifier.citation | Rahimi, 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.isbn | 979-8-9906980-7-9 | |
| dc.identifier.issn | 1819-0138 | |
| dc.identifier.uri | https://doi.org/10.22318/icls2026.113185 | |
| dc.identifier.uri | https://repository.isls.org/handle/1/12744 | - |
| dc.description | SHORT | en |
| dc.description.abstract | Large 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.extent | 1223-1227 | en |
| dc.publisher | International Society of the Learning Sciences | |
| dc.relation.ispartof | urn:ISBN:979-8-9906980-7-9 | |
| dc.relation.ispartof | Proceedings of the 20th International Conference of the Learning Sciences - ICLS 2026, pp. 1223-1227 | |
| dc.relation.ispartofseries | Proceedings of the 20th International Conference of the Learning Sciences - ICLS 2026 | en |
| dc.subject | ICLS | en |
| dc.title | Small vs. large language models: Comparing language models in generating productive failure math problems | en |
| dc.title.alternative | Small vs. Large Language Models: Comparing Language Models in Generating Productive Failure Math Problems | en |
| dc.type | conferencePaper | |
| dc.identifier.doi | 10.22318/icls2026.113185 | |
| dc.format.pagerange | 1223-1227 | en |
| dc.title.parent | Proceedings of the 20th International Conference of the Learning Sciences - ICLS 2026 | en |
| dc.bibliographicCitation.conferencename | Proceedings of the 20th International Conference of the Learning Sciences - ICLS 2026 | en |
| dc.format.firstpage | 1223 | en |
| dc.format.lastpage | 1227 | en |
| Appears in Collections: | ISLS Annual Meeting 2026 | |
Files in This Item:
| File | Size | Format | |
|---|---|---|---|
| ICLS2026_1223-1227.pdf | 293.72 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.