datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
UCS-Bench
UCS-Bench
ICML 2026 Keep It in Mind: User Centric Continual Spatial Intelligence Reasoning in Egocentric Video Streams
UCS-Bench is a benchmark for evaluating user-centric continual spatial intelligence in streaming egocentric videos.
The dataset contains 170+ hours of egocentric visual observations and 8.1K+ timestamped questions. It is designed to test whether models can understand dynamic spatial environments from a user's point of view, maintain long-term… See the full description on the dataset page: https://huggingface.co/datasets/cocowy1/UCS-Bench.SpatialMemoryagilex_cocktail_sunset_coconutThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.0",
"robot_type": "arx5_bimanual",
"total_episodes": 29,
"total_frames": 23327,
"total_tasks": 1,
"total_videos": 87,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 25,
"splits": {
"train": "0:29"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/villekuosmanen/agilex_cocktail_sunset_coconut.assemble_battery_long_lerobot_v3
assemble_battery_long — two-stage SARM annotations
LeRobot v3 conversion and two-stage SARM annotation of m1ku2/assemble_battery_long.
Contents
160 trajectories: 144 perfect and 16 retry
153,624 frames at 30 FPS
Four video representations: high, left wrist, right wrist, and a vertical synchronized composite
perfect/anchor.hdf5 is a real-robot state anchor and is intentionally excluded
SARM vocabulary
Position the mouse and battery for insertion… See the full description on the dataset page: https://huggingface.co/datasets/Cocoyawn32/assemble_battery_long_lerobot_v3.vlanext-kairos-robolab-eval
VLANeXt-Kairos · RoboLab Evaluation Rollouts
Rollout videos and per-episode metrics from evaluating VLANeXt-Kairos (QWorld
world-model VLA) checkpoints in the RoboLab / Isaac-Sim simulator on a DROID
(Franka Panda, joint-position) manipulation task.
This repository is a long-lived evaluation archive. Each top-level folder is one
evaluated checkpoint; new checkpoints / tasks are added as sibling folders over time
(see Naming convention below).
What's inside… See the full description on the dataset page: https://huggingface.co/datasets/Cocoyawn32/vlanext-kairos-robolab-eval.assemble_battery_long_sarm_two_stage
assemble_battery_long — two-stage SARM annotations
LeRobot v3 conversion and two-stage SARM annotation of m1ku2/assemble_battery_long.
Contents
160 trajectories: 144 perfect and 16 retry
153,624 frames at 30 FPS
Four video representations: high, left wrist, right wrist, and a vertical synchronized composite
perfect/anchor.hdf5 is a real-robot state anchor and is intentionally excluded
SARM vocabulary
Position the mouse and battery for insertion… See the full description on the dataset page: https://huggingface.co/datasets/Cocoyawn32/assemble_battery_long_sarm_two_stage.record-test-aThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "so101_follower",
"total_episodes": 2,
"total_frames": 1579,
"total_tasks": 1,
"total_videos": 2,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:2"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/CoCosul/record-test-a.Record_Parallelogram_SI2025This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "so101_follower",
"total_episodes": 1,
"total_frames": 244,
"total_tasks": 1,
"total_videos": 2,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:1"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/CoCosul/Record_Parallelogram_SI2025.cocoa2This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "so101_follower",
"total_episodes": 10,
"total_frames": 10600,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 30,
"splits": {
"train": "0:10"
},
"data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/yoshikokulala/cocoa2.cocoa_move2This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "bi_so100_follower",
"total_episodes": 10,
"total_frames": 3458,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 30,
"splits": {
"train": "0:10"
},
"data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/yoshikokulala/cocoa_move2.cocoa_move3This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "bi_so100_follower",
"total_episodes": 10,
"total_frames": 3457,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 30,
"splits": {
"train": "0:10"
},
"data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/yoshikokulala/cocoa_move3.cocoa_moveThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "bi_so100_follower",
"total_episodes": 10,
"total_frames": 3454,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 30,
"splits": {
"train": "0:10"
},
"data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/yoshikokulala/cocoa_move.v6v5