General
Image2Struct
Original LaTeX and music-sheet image-to-code predictions with their recorded multimodal requests and per-metric grades. Each response is one published grading dimension of an attempt, not an independent model generation. Webpage exports redact their stimuli and outputs and remain raw-only.
30,636items
14subjects
Metadata: CC-BY-4.0 (paper Appendix L.6). Code: Apache-2.0. Original images and text remain subject to their respective owners; the authors do not claim ownership of these assets.license
vision_languagedomain
code_generationdomain
imagemodality
textmodality
item-level responses released
Response matrix
Loading response matrix…
lowhighUnobserved
Scale: Item-specific grading scales