Back to Rerun

README

examples/python/lerobot_loader/README.md

0.31.41.5 KB
Original Source
<!--[metadata] title = "LeRobot loader" tags = ["2D", "Video", "Loader", "Hugging Face", "LeRobot"] thumbnail = "https://static.rerun.io/LeRobot/ec638243c38d01c0d9e27ef4e52e62c43c6e4ba4/480w.png" thumbnail_dimensions = [480, 275] --> <picture> <source media="(max-width: 480px)" srcset="https://static.rerun.io/LeRobot/ec638243c38d01c0d9e27ef4e52e62c43c6e4ba4/480w.png"> <source media="(max-width: 768px)" srcset="https://static.rerun.io/LeRobot/ec638243c38d01c0d9e27ef4e52e62c43c6e4ba4/768w.png"> <source media="(max-width: 1024px)" srcset="https://static.rerun.io/LeRobot/ec638243c38d01c0d9e27ef4e52e62c43c6e4ba4/1024w.png"> <source media="(max-width: 1200px)" srcset="https://static.rerun.io/LeRobot/ec638243c38d01c0d9e27ef4e52e62c43c6e4ba4/1200w.png"> </picture>

Overview

Rerun has a built in data-loader to visualize LeRobot datasets.

Try it out

Here is a sample dataset used in SmolVLA

bash
git lfs install # If not already installed
git clone https://huggingface.co/datasets/satvikahuja/mixer_on_off_new_1

Then you can open the Viewer and select the directory to load the dataset, or you can open it directly from the terminal:

bash
rerun mixer_on_off_new_1

SDK support

Since this dataloader is included other SDK functionalities should work similar to loading a rerun file but pointing at the directory instead.