---
license: apache-2.0
authors:
- Karim Ziadi
task_categories:
- robotics
tags:
- tsfile
- timeseries
- tabular
- robotics
- lerobot
- manipulation
modality:
- timeseries
- tabular
pretty_name: Record LexxPluss PickNDrop Bottle 02 TsFile
configs:
- config_name: default
  data_files:
  - split: train
    path: data/biamterdex_record_lexxpluss_pickndrop_bottle_02.tsfile
size_categories:
- 10K<n<100K
---

# Record LexxPluss PickNDrop Bottle 02 TsFile

Apache TsFile conversion of [`Biamterdex/record-lexxpluss_pickNdrop_bottle-02`](https://huggingface.co/datasets/Biamterdex/record-lexxpluss_pickNdrop_bottle-02), a LeRobot v2.1 `so101_follower` robot dataset.

## Source and provenance

- Original dataset: [https://huggingface.co/datasets/Biamterdex/record-lexxpluss_pickNdrop_bottle-02](https://huggingface.co/datasets/Biamterdex/record-lexxpluss_pickNdrop_bottle-02)
- Inspected revision: `3b90eceab9b474337399e1310d149f17118c28f5`
- Original author/uploader: [Karim Ziadi (`Biamterdex`)](https://huggingface.co/Biamterdex)
- License: Apache-2.0; task: `Pick and Drop dressing bottle`; split: `train` (`0:52`)
- Sampling rate: 30 fps; 52 episodes; 34,896 frame rows; 1 task; 52 Parquet shards
- Source frame layout: `data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet`

The source card provides no paper or BibTeX citation.

## Converted files

- TsFile: `data/biamterdex_record_lexxpluss_pickndrop_bottle_02.tsfile` (583,676 bytes; 0.56 MiB)
- Table: `biamterdex_record_lexxpluss_pickndrop_bottle_02`; rows: 34,896; episodes/devices: 52
- Time precision: milliseconds; `meta/` is mirrored from the source.

## TsFile schema and conversion

`Time = round(timestamp * 1000)` as INT64 milliseconds, restarting at zero per episode. The source `timestamp` is dropped as redundant (`Time / 1000` seconds). `index` is renamed to `sample_index`; `frame_index` is retained.

TAG/device columns: `episode_index`, `task_index`.

FIELD columns: `action[6]` -> `action_0` ... `action_5`; `observation.state[6]` -> `observation_state_0` ... `observation_state_5` (all FLOAT), plus `frame_index` and `sample_index` (INT64). No rows or numeric state/action dimensions are dropped.

Storage profile: Time `TS_2DIFF + LZ4`; FLOAT/DOUBLE `GORILLA + LZ4`; INT32/INT64 `TS_2DIFF + LZ4`; BOOLEAN `RLE + LZ4` (no BOOLEAN source columns); TAGs use TsFile table-model device/tag storage.

## Videos (not included)

The source contains 156 MP4 files (52 per stream, about 955,558,019 bytes) under [`videos/`](https://huggingface.co/datasets/Biamterdex/record-lexxpluss_pickNdrop_bottle-02/tree/3b90eceab9b474337399e1310d149f17118c28f5/videos): `videos/chunk-000/{observation.images.front,observation.images.top,observation.images.wrist}/episode_XXXXXX.mp4`. Videos are not copied to or uploaded with this TsFile repository. Numeric rows remain frame-aligned via `episode_index` and `frame_index`.

## Validation

Local validation: **PASS**. Source/staged/TsFile metadata rows: 34,896 / 34,896 / 34,896; query sample: 704 rows; TsFile/source Parquet ratio: 37.35%.

## Minimal read example

```python
from tsfile import TsFileReader
reader = TsFileReader("data/biamterdex_record_lexxpluss_pickndrop_bottle_02.tsfile")
with reader.query_table("biamterdex_record_lexxpluss_pickndrop_bottle_02", ["episode_index", "task_index", "Time", "frame_index", "action_0"], batch_size=4096) as result:
    print(result.read_arrow_batch().to_pandas().head())
reader.close()
```

Please cite the original Hugging Face dataset and Karim Ziadi (`Biamterdex`).