Datasets:
SO100 Four Tasks TsFile
Apache TsFile edition of Mwuqiu/So100_Four_Tasks, a LeRobot v2.1 SO100 manipulation dataset. The numeric trajectories are stored in one table-model TsFile.
Source and attribution
- Original author, repository owner, and uploader: WuQiu (Mwuqiu).
- License: Apache-2.0.
- The source card does not provide a homepage, paper, or completed BibTeX citation.
- Split: train; 100 episodes; 44,599 frame rows; five tasks; 30 FPS; 100 source Parquet shards.
- The repository name says
Four_Tasks, while its currentmeta/info.jsonandmeta/tasks.jsonlcontain five tasks. The current metadata and Parquet data are used here.
| Task index | Task | Episodes | Rows |
|---|---|---|---|
| 0 | Put the red square sponge into the box | 20 | 8,641 |
| 1 | Put the green square sponge into the box | 20 | 8,984 |
| 2 | Put the yellow duck into the box | 20 | 9,286 |
| 3 | Put the black duck into the box | 20 | 9,230 |
| 4 | Put the blue square sponge into the box | 20 | 8,458 |
Data layout
The table is mwuqiu_so100_four_tasks and contains 44,599 rows across 100 TAG devices. The source Parquet shards total 2,460,918 bytes; the TsFile is 864,985 bytes (35.1% of the source Parquet size).
| Column | TsFile role | Type | Meaning |
|---|---|---|---|
Time |
TIME | INT64 milliseconds | round(timestamp * 1000), restarting at zero per episode |
episode_index |
TAG | STRING from source INT64 | Source episode identity |
task_index |
TAG | STRING from source INT64 | Source task identity |
frame_index |
FIELD | INT64 | Frame position within the episode |
sample_index |
FIELD | INT64 | Source index, renamed for clarity |
action_0 ... action_5 |
FIELD | FLOAT | Flattened action[6] |
observation_state_0 ... observation_state_5 |
FIELD | FLOAT | Flattened observation.state[6] |
timestamp is not retained as a separate FIELD because it is represented by Time / 1000 seconds. The vector column names preserve their source prefixes, with dots changed to underscores. No trajectory row, episode, task, action dimension, or state dimension is removed.
The data/ directory contains only mwuqiu_so100_four_tasks.tsfile. Source metadata remains under meta/; source trajectory Parquet files are not placed in meta/.
Videos and alignment
The 200 source H.264 videos are not included here. They remain under videos/chunk-000 in the original repository:
observation.images.topview: 100 files, 315,226,127 bytesobservation.images.wristview: 100 files, 491,561,693 bytes
The source template is videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4. Use episode_index and frame_index to align each numeric row with both 30 FPS video streams.
Read example
from tsfile import TsFileReader
reader = TsFileReader("data/mwuqiu_so100_four_tasks.tsfile")
with reader.query_table(
"mwuqiu_so100_four_tasks",
["episode_index", "task_index", "frame_index", "sample_index", "action_0", "observation_state_0"],
batch_size=1024,
) as result:
batch = result.read_arrow_batch()
print(batch.to_pandas().head())
reader.close()
- Downloads last month
- 36