arXiv:2606.03814v1 Announce Type: new Abstract: This paper investigates rubric-aware, multitask fine-tuning of transformer models for automated grading of introductory C++ programming assignments, with the goal of producing grade predictions that better reflect instructor grading behavior than general-purpose LLMs. Using multi-semester CS1 data, student submissions are paired with numeric scores, letter-grade buckets, and assignment rubrics, then preprocessed into unified sequences for transformer input. A BART encoder-decoder with LoRA adaptation is trained to jointly predict numeric grades a
The proliferation of advanced LLMs necessitates research into their specific applications and fine-tuning for tasks like automated grading, addressing limitations of general-purpose models.
Automated, rubric-aware grading could significantly scale and standardize education by reducing instructor workload and providing consistent feedback, impacting workforce development and skill acquisition.
The ability to accurately and automatically assess programming assignments using detailed rubrics represents a step towards more efficient and scalable educational pathways in technical fields.
- · Educational institutions
- · Students
- · AI developers (education)
- · Online learning platforms
- · Traditional manual graders
Automated grading systems become more sophisticated and prevalent in computer science education.
Instructors are freed from mundane grading tasks, allowing them to focus on mentoring and curriculum development, potentially enhancing educational quality.
Massive scaling of technical education becomes more feasible, impacting global skill distribution and economic opportunities.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI