Joan Gregorio Pérez - Publicación Científica
Título: Pedagogical Socratic Feedback Generation in Computer Science Education using Fine-Tuned CodeLLMs
Año: 2024
Venue: ACM Technical Symposium on Computer Science Education (SIGCSE 2024)
DOI: 10.1145/3626252.3630911

Abstract:
Automated grading tools frequently reveal solution code prematurely, robbing students of critical problem-solving epiphanies. We present Socrates-Code, a fine-tuned 14B model that generates tiered heuristic questions instead of answers. Longitudinal evaluations across 3 semesters of Algorithms courses demonstrated a 28% increase in retention and debugging proficiency.

BibTeX:
@inproceedings{pérez2024pedagogical,
  author = Pérez, Joan Gregorio and Mendoza, Prof. Carlos,
  title = {Pedagogical Socratic Feedback Generation in Computer Science Education using Fine-Tuned CodeLLMs},
  booktitle = {ACM Technical Symposium on Computer Science Education (SIGCSE 2024)},
  year = 2024,
  doi = {10.1145/3626252.3630911},
  eprint = {2402.05199},
  archiveprefix = {arXiv},
  primaryclass = {cs.AI},
  url = {https://joangregorioperez.com/papers/pedagogical-feedback-optimization-cs-education-codellms},
  abstract = {Automated grading tools frequently reveal solution code prematurely, robbing students of critical problem-solving epiphanies. We present Socrates-Code, a fine-tuned 14B model that generates tiered heuristic questions instead of answers. Longitudinal evaluations across 3 semesters of Algorithms cours...},
  pages = {412--418},
}