The Vision-QA layer of D14 (BenchCAD), packaged separately with its intended usage made
explicit: this is an evaluation benchmark / agentic-setting seed, NOT direct VLM training data.
2,400 records: a 4-view orthographic render of a CAD part + a numeric question +
the gold answer. The underlying CadQuery program that generated the part (and the answers) is
preserved in metadata.gt_code.
Why this is… See the full description on the dataset page: https://huggingface.co/datasets/AI4Manufacturing/D14-vision-qa.
