Get Started
Home
Topics
Search
Library
Research questionHow can LLM graders assign accurate marks while grounding each judgment in rubrics and student-answer evidence?Accurate score prediction does not ensure that a grader followed the mark scheme or relied on relevant evidence in the student answer. Grading errors can also be difficult to locate when the model’s reasoning changes its final mark.
AI
Evaluation & Benchmarks
LLM Pretraining & Post-training
Natural Language Processing
Reasoning
Latest papersRecent research connected to this question, newest first.EDIT: Evidence-Diagnosed Intervention Training for Rule-Faithful LLM GradingThe source evaluates LLM grading on two real-world, multi-subject benchmarks using in-domain and out-of-domain splits. It also examines deterministic rubric-edit interventions and internal signals for final-mark beliefs and input grounding; the reported evidence concerns the EDIT training framework and its evaluated baselines.research paper · Sep 2, 2026
Related questions
How can rubric-based LLM grading resist prompt injection without compromising fair judgments?How can we tell whether agreement among LLM judges reflects human alignment or shared blind spots?How can LLMs assess academic proposals and peer feedback with pedagogically grounded scores?How can we evaluate LLM reasoning quality beyond final-answer accuracy across deployment contexts?