Marking Justification Quality Evaluation on NSW marking justification (test)
80.4Win RateCurriculum-Grounded Marking Pipeline
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Curriculum-Grounded Marking PipelineComparison Baseline=human2026.06 | 80.4 | 26 | 11 | 0 | |
| Curriculum-Grounded Marking PipelineComparison Baseline=gpt-52026.06 | 69.6 | 19 | 17 | 1 | |
| Curriculum-Grounded Marking PipelineComparison Baseline=gpt-5-nano2026.06 | 69.6 | 17 | 19 | 1 |