Add dataset card (inputs)

README.md

@@ -0,0 +1,143 @@
+---
+license: odc-by
+language:
+- en
+pretty_name: CADGenBench (Inputs)
+task_categories:
+- image-to-3d
+- text-to-3d
+tags:
+- cad
+- 3d
+- step
+- mechanical-engineering
+- benchmark
+- cad-generation
+- cad-editing
+size_categories:
+- n<1K
+annotations_creators:
+- expert-generated
+language_creators:
+- expert-generated
+multilinguality:
+- monolingual
+source_datasets:
+- original
+---
+
+# CADGenBench (Inputs)
+
+Public inputs for the **CADGenBench** benchmark, which measures how well AI
+systems produce correct 3D mechanical parts as STEP files. This repository
+holds the task inputs only; the ground truth is withheld in a separate private
+repository so that the leaderboard's evaluation is the single source of truth.
+
+- **Leaderboard Space**: [`HuggingAI4Engineering/cadgenbench-leaderboard`](https://huggingface.co/spaces/HuggingAI4Engineering/cadgenbench-leaderboard)
+- **Browse the tasks**: the [Tasks tab](https://huggingface.co/spaces/HuggingAI4Engineering/cadgenbench-leaderboard) on the Space (thumbnails, search, generation/editing filter, per-task detail)
+- **Benchmark code**: [`github.com/huggingface/cadgenbench`](https://github.com/huggingface/cadgenbench)
+- **Submissions + results**: [`HuggingAI4Engineering/cadgenbench-submissions`](https://huggingface.co/datasets/HuggingAI4Engineering/cadgenbench-submissions)
+- **Ground truth**: [`HuggingAI4Engineering/cadgenbench-data-gt`](https://huggingface.co/datasets/HuggingAI4Engineering/cadgenbench-data-gt) (private)
+
+## Dataset Summary
+
+CADGenBench contains **81 fixtures** of mechanical parts split across two
+tasks:
+
+- **Generation (49)**: reproduce the part as an accurate 3D solid from an
+  engineering drawing.
+- **Editing (32)**: apply a described change to an existing STEP solid.
+
+Each submission is one `output.step` per fixture. Outputs are scored against
+the private ground truth by the CAD Score pipeline: a hard validity gate
+followed by a weighted mean of three orthogonal metrics (shape similarity,
+interface match, topology match). See the
+[metric definitions](https://github.com/huggingface/cadgenbench/blob/main/docs/metrics.md).
+
+## Supported Tasks and Leaderboards
+
+The benchmark is tool-agnostic: a submission is one STEP file per fixture,
+produced by any system (one LLM, several, a script, or by hand). Submit and
+view results on the
+[leaderboard Space](https://huggingface.co/spaces/HuggingAI4Engineering/cadgenbench-leaderboard);
+the full submission contract (zip layout, `meta.json`, validity gate) is in
+[`docs/benchmark/submission.md`](https://github.com/huggingface/cadgenbench/blob/main/docs/benchmark/submission.md).
+
+## Languages
+
+Task prompts and descriptions are in English.
+
+## Dataset Structure
+
+### Data Instances
+
+One directory per fixture, named by its numeric id. There are two shapes:
+
+```
+# Generation fixture
+<id>/
+├── description.yaml      # prompt + metadata
+└── input.png            # the engineering drawing (input2.png, ... when multi-image)
+
+# Editing fixture
+<id>/
+├── description.yaml      # prompt + metadata
+├── edit_description.txt  # the requested change, as an instruction
+├── input.step           # the starting solid to edit
+├── input.mesh.npz       # trusted watertight mesh sidecar for input.step
+└── renders/             # iso / front / top / right PNGs of the starting solid
+```
+
+### Data Fields
+
+`description.yaml` carries:
+
+| Field | Type | Description |
+|---|---|---|
+| `description` | string | The task prompt. |
+| `task_type` | `"generation"` \| `"editing"` | The task family. |
+| `input_files` | list of strings | The input files the task declares (e.g. `input.png`, `input2.png`, or `input.step`). |
+| `input_type` | `"text+image"` \| `"text+step"` | Modality of the inputs. |
+
+### Data Splits
+
+There is no train/test split — all 81 fixtures are an evaluation set
+(49 generation, 32 editing). The fixture set may grow over time; consumers
+should discover fixtures at runtime rather than hardcoding the list.
+
+## Dataset Creation
+
+### Source Data
+
+The underlying CAD geometry is sourced from [Mecado](https://www.mecado.com).
+Fixtures are real mechanical parts with mating interfaces (locating jigs, bolt
+patterns, slots). For each part, an engineering drawing (generation) or a
+starting solid plus an edit instruction (editing) forms the public input; the
+solved solid and its interface sub-volumes are held privately as ground truth.
+
+### Curation Rationale
+
+The inputs are released publicly so contestants see exactly what they are
+solving, while the ground truth stays private so the leaderboard's server-side
+evaluation is the only path to a score.
+
+## Considerations for Using the Data
+
+This is a benchmark **input** set, not a training corpus. Ground truth is not
+included here. To participate, generate one `output.step` per fixture and
+submit through the leaderboard Space.
+
+## Additional Information
+
+### Dataset Curators
+
+The CADGenBench team ([HuggingAI4Engineering](https://huggingface.co/HuggingAI4Engineering)).
+
+### Licensing Information
+
+Released under the [Open Data Commons Attribution License (ODC-BY)](https://opendatacommons.org/licenses/by/1-0/).
+
+### Acknowledgements
+
+CAD geometry sourced from [Mecado](https://www.mecado.com). Thanks to the
+Mecado team.