A contrastive mixed-context fine-tuning method lets a 60M-parameter T5 model generate topic-controlled educational questions, with best topical alignment from augmented data and a Jaccard Wikipedia metric.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CY 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A Novel Approach to Scalable and Automatic Topic-Controlled Question Generation in Education
A contrastive mixed-context fine-tuning method lets a 60M-parameter T5 model generate topic-controlled educational questions, with best topical alignment from augmented data and a Jaccard Wikipedia metric.