Hugging Face's logo

Datasets:
Berkeley-ICON-Lab
/
ALTER-data

Modalities:
Tabular
Formats:
json
Size:
< 1K
ArXiv:
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
ALTER-data / USAGE.md
dayiethan's picture
dayiethan
Document verified ALTER hardware v1 and pinned public code
dc19166 verified
preview code
|
Raw History Blame Contribute Delete
2.3 kB

Downloading ALTER simulation artifacts

Models: https://huggingface.co/Berkeley-ICON-Lab/ALTER-models Data and provenance: https://huggingface.co/datasets/Berkeley-ICON-Lab/ALTER-data Code: https://github.com/labicon/ALTER/tree/e180a013e30d7b7c5478e029229ff9cd3154028e

The simulation code is merged into the public repository’s main branch. The commands below pin the tested release revision for reproducible downloads. Both artifact uploads are complete. release-manifest.json records immutable artifact revisions and checksums. bundle-index.json and SHA256SUMS list this repository’s artifact contents. The manifest pins the commits containing the artifacts, which precede this documentation commit.

Check out the tested code revision and follow its installation guide:

git clone https://github.com/labicon/ALTER.git
cd ALTER
git checkout e180a013e30d7b7c5478e029229ff9cd3154028e

Then, from the repository root run:

python scripts/download_release.py --manifest release/manifest.json \
  --bundles simulation-models simulation-pretraining-data \
  simulation-adaptation-data simulation-provenance \
  --destination /path/to/ALTER-artifacts
python scripts/materialize_release.py --bundle-root /path/to/ALTER-artifacts \
  --output "$PWD/public-validation/inputs"
python scripts/validate_release_models.py --inputs "$PWD/public-validation/inputs" \
  --output "$PWD/public-validation/model-check.json" --training-smoke

Materialization requires all four bundles. It verifies hashes, relocates portable artifact references and strictly validates derived checkpoint contracts. Direct use of portable statistics before materialization is not supported. Only load trusted pickle/PyTorch files. Short tests check functionality; they do not rerun full training or establish paper success rates. Hardware v1 is documented separately in hardware/v1/USAGE.md.

Hardware release code

Use hardware v1 instructions and the tested code revision. Hardware and simulation use separate download directories and manifests.