Datasets:
Download examples/reading_library__open_hall__00/README.md from KurtDu/SpaceCode-Bench: direct link, hf CLI and curl.
- Browser
- Download file 2.4 kB
-
https://huggingface.co/datasets/KurtDu/SpaceCode-Bench/resolve/main/examples/reading_library__open_hall__00/README.md
- Command line
-
hf download hf://datasets/KurtDu/SpaceCode-Bench/examples/reading_library__open_hall__00/README.md
-
curl -L -o README.md https://huggingface.co/datasets/KurtDu/SpaceCode-Bench/resolve/main/examples/reading_library__open_hall__00/README.md
Worked example: edit a reading library
Task: reading_library__open_hall__00__edit. This is a repository-owned author control, not a model submission.
The complete public request is in task.json, prompt.txt and contract.json. The solver receives starter.blend and the published edit region. edited.blend is the authored positive control; base.blend supplies the original reference environment. These native files have the same SHA256 bindings as the website's corresponding controls.
| Control | A | G | F | P | Native score | Critical pass |
|---|---|---|---|---|---|---|
| Requested edit | 1 | 1 | 1 | 1 | 100 | true |
| Unchanged starter | 1 | 0.9787234043 | 1 | 1 | 49 | false |
The unchanged starter preserves the protected environment but fails the requested intervention. A high mean geometric score cannot override that critical failure. R is inapplicable to editing; C, V and full_score are unmeasured.
evidence/ and unchanged-starter-evidence/ contain actual independently extracted geometry, inspection, fixed renders, bound preservation evidence and clause-level results. starter-evidence/ is the preservation reference. scoring-requested_edit/ and scoring-unchanged_starter/ record evaluator-7 numerical recomputation, its process limits and actual dependency identities. provenance.json identifies the native artifacts and distinguishes this new numerical run from the preserved extraction. No model API was called.
To rebuild the authored native scenes with Blender 4.2.23:
blender --background --factory-startup --python build_scene.py -- \
--repository /absolute/path/to/SpaceCode-Bench --out /absolute/path/to/rebuilt
The builder uses spec.json and the repository's procedural mesher. A rebuilt .blend may differ byte-for-byte because of save metadata; compare extracted geometry under the same documented toolchain. Author generation is a reproducibility aid and is not an allowed solver input.
For model evaluation, download the complete dataset and use the repository's spacecode batch or spacecode evaluate entry. They validate public/private resources and isolate candidate execution before extraction and numerical scoring. The standalone legacy spacecode score command does not implement this native protocol.
Data: CC BY 4.0, Yuhao Du. Builder code: MIT. See the licenses at the repository root.