Gumbel Machine: Counterfactual Student Writing Generation via Gumbel Noise Steering
The Gumbel Machine is a new approach to generating counterfactual student writing that aims to improve educational outcomes. It utilizes a controlled decoding algorithm called $eta$-Hindsight control to ensure generated texts are similar to reference works while still being improved versions of students' original writings. Experiments show that this method effectively produces counterfactuals that align with rubric criteria and maintain similarity to the original texts.
- ▪The Gumbel Machine leverages large language models to generate counterfactuals for student writing.
- ▪A novel decoding algorithm, $eta$-Hindsight control, is central to the Gumbel Machine's approach.
- ▪Experiments demonstrate the effectiveness of the Gumbel Machine in producing high-quality counterfactuals.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.27249 (cs) [Submitted on 26 May 2026] Title:Gumbel Machine: Counterfactual Student Writing Generation via Gumbel Noise Steering Authors:Hunter McNichols, Alexander Scarlatos, Mihai Dascalu, Danielle McNamara, Andrew Lan View a PDF of the paper titled Gumbel Machine: Counterfactual Student Writing Generation via Gumbel Noise Steering, by Hunter McNichols and 4 other authors View PDF HTML (experimental) Abstract:An effective method of teaching across disciplines is to provide examples of high-quality work. However, an example may be significantly different from a student's current work, making it challenging for them to emulate.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
