The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
StarBenchServer offline benchmark assets
This dataset contains the validated, reproducible runtime inputs for four StarBenchServer benchmarks:
robodojo/— RoboDojo source and official Assets, plus the qualified runtime receipt.behavior/— BEHAVIOR-1K source/data and the Isaac Sim overlay.robocasa365/— RoboCasa and RoboSuite source trees plus the extra Python dependencies used by the harness.robotwin/— pinned RoboTwin deployment and its required asset archive.common/— StarBenchServer/Harness integration files, shared runtime support, evidence receipts and verification script.
The package is intended for an offline target that can reach Hugging Face but cannot reach GitHub, PyPI, Harbor or other registries. The two OCI image layouts under the benchmark folders contain the validated runtime images. A host still needs a compatible NVIDIA driver, GPU, container runtime and (for Isaac Sim) acceptance of NVIDIA's license. Policy model checkpoints, evaluation results, videos, logs, credentials and caches are deliberately excluded.
Download
python - <<'PY'
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="JackAILab/assests", repo_type="dataset",
local_dir="/opt/starbench-assets",
allow_patterns=["common/**", "robodojo/**", "behavior/**", "robocasa365/**", "robotwin/**"],
)
PY
Run python /opt/starbench-assets/common/verify_bundle.py /opt/starbench-assets/robotwin (and the corresponding robodojo, behavior and robocasa365 paths) after download. Each benchmark README gives the image import, environment variables and smoke-test entrypoint.
This export is an asset/runtime distribution, not a replacement for NVIDIA's host driver or license terms.
- Downloads last month
- 8