Datasetpublic

xiaojiahao/libero-10-baseline-subsets

LIBERO-10 dataset selection experiments The training datasets are organized into two folders: Folder Contents Ratios baselines/ 12 datasets selected by three baseline objectives 0.2, 0.4, 0.6, 0.8 led_selected/ 9 LED control/candidate datasets 0.2, 0.4, 0.6 Each dataset directory contains complete images, observations and actions in 1.0.0/ (prepared RLDS/TFDS format). Dataset records and selection membership are unchanged by this folder reorganization.… See the full description on the dataset page: https://huggingface.co/datasets/xiaojiahao/libero-10-baseline-subsets.

sourceHugging Facemitupdated 22h agoView on Hugging Face

0likes134downloads

Dataset Card

LIBERO-10 dataset selection experiments

The training datasets are organized into two folders:

Folder	Contents	Ratios
baselines/	12 datasets selected by three baseline objectives	0.2, 0.4, 0.6, 0.8
led_selected/	9 LED control/candidate datasets	0.2, 0.4, 0.6

Each dataset directory contains complete images, observations and actions in 1.0.0/ (prepared RLDS/TFDS format). Dataset records and selection membership are unchanged by this folder reorganization.

—Baseline descriptions and loading instructions
—LED versions and loading instructions

Within led_selected/, suffixes distinguish legacy_control (original LED control), led_refined (stronger search under the original LED objective), and led_coverage (experimental coverage-regularized variant).

Use tfds.builder_from_directory; the internal dataset name remains liber_o10. Include the baselines/ or led_selected/ prefix when downloading and loading. Nominal ratios retain 80, 157, 231 or 308 episodes per dataset from the same 379-episode source.

Source: OpenVLA modified LIBERO RLDS. Please credit LIBERO and OpenVLA.