Education
SketchJudge
SketchJudge (arXiv 2601.06944): 16 multimodal LLMs act as graders of 1,015 hand-drawn student STEM answers (geometry, physics, chart, flowchart). response = binary grading accuracy, 1 iff the model's correctness verdict matches the expert gold label, under WithRef/NoRef settings and baseline/CoT/rubric prompt styles.
4,052items
38subjects
Apache-2.0 for the GitHub project. The external diagram archive has no separate license declaration; captured data remain local pending redistribution clarification.license
educationdomain
reasoningdomain
textmodality
imagemodality
item-level responses released
Saturation status: No
Response matrix
Loading response matrix…
Correct (1)Incorrect (0)Unobserved
Scale: 1 = correct · 0 = incorrect