azithromycin/LSY-lab_libero_plus_10_noise_all_failures

LIBERO Plus 10 Noise All Failures TsFile

This dataset is an Apache TsFile conversion of LSY-lab/libero_plus_10_noise_all_failures, a LeRobot v3.0 Franka manipulation dataset with successful and failed attempts across ten LIBERO-Plus task instructions. The source repository is published by LSY-lab; the original uploader/author is Maximilian Christof (max-chr).

Modalities: Time-series; Tabular. Numeric robot observations, actions, frame timing, episode/task tags, and mirrored source metadata are included. Camera videos remain in the original Hugging Face dataset.

Source Dataset and Attribution

• Original dataset: LSY-lab/libero_plus_10_noise_all_failures
• Pinned source revision: efc123593f272078361f01611317270ea91f28b4 (files and versions)
• Source organization: LSY-lab
• Original uploader/author: Maximilian Christof (max-chr)
• License: Apache-2.0
• Robot type: franka; LeRobot codebase: v3.0
• Sampling rate: 20 fps; split: train
• The source card does not provide a dataset-specific paper citation or BibTeX entry.
• Source frame layout: data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
• Source video layout: videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4

Tasks and Scale

The train split contains 290 episodes, 84,612 frame rows, and 10 task indexes.

task_index	Instruction	Episodes	Frames
0	put both the cream cheese box and the butter in the basket	2	520
1	put both the alphabet soup and the cream cheese box in the basket	16	4,410
2	put both moka pots on the stove	44	18,886
3	put the white mug on the plate and put the chocolate pudding to the right of the plate	40	10,898
4	pick up the book and place it in the back compartment of the caddy	44	8,340
5	put both the alphabet soup and the tomato sauce in the basket	34	10,092
6	put the white mug on the left plate and put the yellow and white mug on the right plate	26	6,724
7	turn on the stove and put the moka pot on it	12	3,288
8	put the yellow and white mug in the microwave and close it	44	14,152
9	put the black bowl in the bottom drawer of the cabinet and close it	28	7,302

The source numeric Parquet is 13.23 MiB. The original videos/ tree contains 1,160 MP4 files (789.68 MiB), 290 per stream across four 256x256 RGB AV1 streams at 20 fps with no audio:

• observation.images.image - 290 MP4 files
• observation.images.wrist_image - 290 MP4 files
• observation.images.frontview_image - 290 MP4 files
• observation.images.sideview_image - 290 MP4 files

Converted Files

• TsFile: data/lsy_lab_libero_plus_10_noise_all_failures.tsfile
• Table: lsy_lab_libero_plus_10_noise_all_failures
• Source shards: 1 Parquet -> one merged TsFile
• Rows: 84,612; episodes/devices: 290
• TsFile size: 9.29 MiB (9,737,649 bytes)
• meta/ is mirrored from the source; meta/info.json documents the converted schema, source revision, and video policy.

TsFile Schema

TIME columns:

• Time

TAG columns (TsFile table-model device dimensions):

• episode_index
• task_index

FIELD columns:

• frame_index
• sample_index
• observation_state_0
• observation_state_1
• observation_state_2
• observation_state_3
• observation_state_4
• observation_state_5
• observation_state_6
• observation_state_7
• observation_states_ee_state_0
• observation_states_ee_state_1
• observation_states_ee_state_2
• observation_states_ee_state_3
• observation_states_ee_state_4
• observation_states_ee_state_5
• observation_states_joint_state_0
• observation_states_joint_state_1
• observation_states_joint_state_2
• observation_states_joint_state_3
• observation_states_joint_state_4
• observation_states_joint_state_5
• observation_states_joint_state_6
• observation_states_gripper_state_0
• observation_states_gripper_state_1
• action_0
• action_1
• action_2
• action_3
• action_4
• action_5
• action_6

Flattened vector groups (FLOAT32 scalar fields; source . becomes _):

• observation.state (8 values) -> observation_state_0 ... observation_state_7
• observation.states.ee_state (6 values) -> observation_states_ee_state_0 ... observation_states_ee_state_5
• observation.states.joint_state (7 values) -> observation_states_joint_state_0 ... observation_states_joint_state_6
• observation.states.gripper_state (2 values) -> observation_states_gripper_state_0 ... observation_states_gripper_state_1
• action (7 values) -> action_0 ... action_6

Scalar renames and drops:

• index -> sample_index
• timestamp -> used for Time and dropped because it equals Time / 1000 seconds
• Video columns observation.images.image, observation.images.wrist_image, observation.images.frontview_image, observation.images.sideview_image are omitted from TsFile; they remain in the source repository.

Time, Encoding, and Compression

• Time is round(timestamp * 1000) as INT64 milliseconds and restarts at zero for each episode.
• Rows are sorted by episode_index, task_index, then Time; all 84,612 (TAG, Time) keys are unique.
• FLOAT/DOUBLE -> GORILLA + LZ4.
• INT32/INT64 -> TS_2DIFF + LZ4.
• TIME -> TS_2DIFF + LZ4.
• BOOLEAN -> RLE + LZ4 (no BOOLEAN source field is present in this revision).
• TAG values use TsFile table-model device/TAG segments.
• The source action vectors contain 610 IEEE-754 infinite values (510 positive, 100 negative); they are preserved as source values and are not filtered.

Videos and Frame Alignment

Videos are not copied to this TsFile repository. Retrieve them from the original videos/ tree. Align numeric rows to MP4 frames using episode_index and frame_index; meta/episodes/chunk-000/file-000.parquet contains each stream's chunk/file and timestamp ranges.

Validation

Local JSON and Markdown reports record source/staged/TsFile row counts, schema, physical codecs, TAG/device counts, Time monotonicity, Inf preservation, and output size. The converted TsFile is smaller than the source numeric Parquet.

Minimal Read Example

from tsfile import TsFileReader

reader = TsFileReader("data/lsy_lab_libero_plus_10_noise_all_failures.tsfile")
table_name = "lsy_lab_libero_plus_10_noise_all_failures"
columns = [
    "episode_index", "task_index", "frame_index", "sample_index",
    "action_0", "observation_state_0",
]
with reader.query_table(table_name, columns, batch_size=65536) as result:
    batch = result.read_arrow_batch()
    print(batch.to_pandas().head())
reader.close()

The dataset-specific YAML and converter remain available locally under D:/code/config/ and D:/code/scripts/; validation reports are local artifacts and are not part of the TsFile upload set.