artifact_type string | generated_at_utc string | source_root string | file_count int64 | total_bytes int64 | files list |
|---|---|---|---|---|---|
raw_hdf5_mirror | 2026-03-19T05:40:20.899721+00:00 | /home/youngjoon/github/latent_action/datasets | 1 | 4,610,179,960 | [
{
"relative_path": "can/ph/image_v15.hdf5",
"bytes": 4610179960,
"sha256": "aa45de6d36a3b03ee0df9207a600ebb42c39d94bbebfd63b9ab77944887328f8",
"task": "can",
"variant": "ph",
"modality": "image",
"version": "v15",
"demo_count": 200,
"action_dim": 7,
"obs_keys": [
"agent... |
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
joon-stack/latent-action-can-ph-image-v15
Raw HDF5 artifact mirror for latent_action training.
This repository is not a native load_dataset(...) dataset.
It stores robomimic-style HDF5 files for download and local path overrides.
Contents
- Files: 1
- Total bytes: 4610179960
Usage
Download the needed HDF5 file locally and pass it to Hydra:
python run.py data.paths=[/absolute/path/to/can/ph/image_v15.hdf5]
Files
can/ph/image_v15.hdf5demos=200, modality=image, variant=ph, action_dim=7, obs_keys=agentview_image, object, robot0_eef_pos, robot0_eef_quat, robot0_eef_quat_site, robot0_eye_in_hand_image, robot0_gripper_qpos, robot0_gripper_qvel, robot0_joint_pos, robot0_joint_pos_cos, robot0_joint_pos_sin, robot0_joint_vel
- Downloads last month
- 29