Juelg/RPD-maniskill

Maniskill Sub-Dataset in RLDS Format used in RPD

This repository contains the maniskill subset in RLDS format used to train Octo and OpenVLA in the paper Refined Policy Distillation: From VLA Generalists to RL Experts, which distilled these models using RL.

Project page: https://refined-policy-distillation.github.io Code: https://github.com/Refined-Policy-Distillation/RPD

Checkout octo-base-1.5-finetuned-maniskill and openvla-7b-finetuned-maniskill which have been fine-tuned with that dataset.

Sample Usage

You can download the dataset using git lfs:

git lfs install
git clone git@Juelg/RPD-maniskill

The dataset is in RLDS format and can be loaded and visualized using tools such as dlimp.

Citation

If you find RPD useful for your work, please consider citing it:

@inproceedings{juelg2025refinedpolicydistillationvla,
    title={{Refined Policy Distillation}: {F}rom {VLA} Generalists to {RL} Experts}, 
    author={Tobias Jülg and Wolfram Burgard and Florian Walter},
    year={2025},
    booktitle={Proc.~of the IEEE/RSJ Int.~Conf.~on Intelligent Robots and Systems (IROS)},
    note={Accepted for publication.}
}