Yes, it is publicly available on the Hugging Face Hub.

How do I access and load the data?

Use the Hugging Face datasets library with load_dataset('IPEC-COMMUNITY/language_table_lerobot').

What license applies?

Check the repository card on Hugging Face for the specific license terms.

language_table_lerobot

Large-scale xArm dataset with language instructions for table tasks.

DatasetText & NLP↓ 266K/moFree

Open dataset

Updated 2026-06-18

What is language_table_lerobot?

language_table_lerobot is a robotics dataset of xArm trajectories paired with language instructions for table-based manipulation tasks. It was created with the LeRobot library and supplies episode metadata, video paths, and frame-level records across 443 chunks.

It supports training of imitation learning, vision-language-action, and language-conditioned control models in research and development settings.

What you can build with language_table_lerobot

Train language-conditioned imitation policies

Use the 127k task instructions paired with xArm trajectories to train models that map natural language commands to robot actions via behavioral cloning.

Benchmark vision-language-action models

Evaluate multimodal architectures on the 7M+ frames of synchronized video, state, and action data stored in parquet files.

Develop task generalization methods

Leverage the 442k episodes across diverse manipulation scenarios to test zero-shot or few-shot transfer of robot skills to new instructions.

Load language_table_lerobot

Python

from datasets import load_dataset

ds = load_dataset("IPEC-COMMUNITY/language_table_lerobot")

1pip install datasets lerobot
2from datasets import load_dataset
3ds = load_dataset('IPEC-COMMUNITY/language_table_lerobot', split='train')
4Load parquet files and linked MP4 videos per episode
5Feed trajectories into LeRobot training scripts for policy learning