Download sources/robotics/metadata/README.md from modelomics/modelome: direct link, hf CLI and curl.
- Browser
- Download file 2.69 kB
-
https://huggingface.co/datasets/modelomics/modelome/resolve/main/sources/robotics/metadata/README.md
- Command line
-
hf download hf://datasets/modelomics/modelome/sources/robotics/metadata/README.md
-
curl -L -o README.md https://huggingface.co/datasets/modelomics/modelome/resolve/main/sources/robotics/metadata/README.md
pretty_name: MODELOME public metadata
tags:
- machine-learning
- model-registry
- provenance
- research
MODELOME public metadata
This is a provenance-preserving, metadata-only export of the MODELOME registry. It records model identities, source-attributed artifact links, and trained-release identifiers. The source registry commit is 00000000000000000018-bd2b210ce6463571-41e46451164d (generation 18).
Included files
- artifact_identifiers.parquet: 21 rows.
- artifact_url_links.parquet: 32 rows.
- artifacts.parquet: 21 rows.
- model_aliases.parquet: 41 rows.
- model_artifacts.parquet: 21 rows.
- model_identifiers.parquet: 21 rows.
- model_releases.parquet: 21 rows.
- models.parquet: 21 rows.
- release_identifiers.parquet: 21 rows.
model_artifacts.parquet joins model IDs to their current source artifacts. artifact_url_links.parquet joins a current artifact to a discovered external URL and, when present in this export, the matching target artifact ID. Together with artifacts.parquet and identifiers, those tables support paper/code/model and model/release reconstruction without claiming that a URL is a binary payload. model_releases.parquet may include weight_files_json, a JSON array of safe relative filenames declared by a Hugging Face model record. weight_files_complete says whether the revision's file listing finished; false means the array contains only observed files. Other release metadata is omitted.
Deliberately excluded
This release does not include source-record raw payloads, paper abstracts or full text, model-card bodies, README bodies, source HTML, or unreviewed evidence values. Follow the external URLs and identifiers for those materials under their own terms.
Source attribution and rights
source-manifest.json records configured sources and observed source checkpoints, including configured sources with no observed checkpoint. The checkpoint's complete flag and counts are reported only when that checkpoint exists; they do not claim unrun sources were ingested. A true complete flag means the adapter finished its checkpointed scan; it does not claim exhaustive historical or global coverage. Source-level terms remain in force. In particular, derivative users must preserve attribution and applicable share-alike obligations for records derived from the Papers with Code archival datasets. The arXiv snapshot contributes factual metadata only; paper-specific licenses remain attached to the original records and paper content is not included.
Contributing source checkpoints:
- octo-checkpoints (octo_checkpoints)
- openpi-checkpoint-manifest (openpi_checkpoint_manifest)
- openvla-checkpoints (openvla_checkpoints)