datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
patch-policy-datasetspatch_policy_libero_object_tipsv2_evalpatch_policy_libero10_tipsv2_lang_evalpatch_policy_libero10_tipsv2_evalpatch_policy_libero10_tipsv2_token_evallibero-goal-points-patchpolicy
LIBERO-Goal point clouds for patch_policy_cleanup
The generated-not-downloaded files needed to train the point-cloud and point-caption conditioners in
llm-robotics/patch_policy_cleanup, plus the two
HDF5 trees they were derived from.
The base LIBERO demonstrations (images, actions, states) are not here — they come from
gaoyuezhou/patch-policy-datasets.
This dataset only supplies the per-demo files that repo's scripts generate, so a new machine does not have
to re-run them.… See the full description on the dataset page: https://huggingface.co/datasets/ParsaSharifi/libero-goal-points-patchpolicy.patch_policy_libero_object_smoke_evalpatch_policy_libero_object_evalpatch-policy-libero-spatial
LIBERO Spatial, re-rendered at 224x224 (Patch Policy layout)
The 500 demonstrations of LIBERO Spatial (10 tasks x 50) from
yifengzhu-hf/LIBERO-datasets (Apache-2.0), converted to
the per-demo layout of the Patch Policy LIBERO Goal release, so that
datasets.libero.LiberoGoalDataset(data_directory, suite="libero_spatial") reads it.
What changed from the source
Each demo's saved MuJoCo states are replayed in the LIBERO simulator and both cameras are rendered again… See the full description on the dataset page: https://huggingface.co/datasets/ParsaSharifi/patch-policy-libero-spatial.