license: cc-by-4.0
task_categories:
- robotics
tags:
- robot-learning
- imitation-learning
- robot-trajectories
MolmoAct Dataset
This repository contains the MolmoAct Dataset, a mid-training robot dataset comprising over 10,000 high-quality robot trajectories across diverse scenarios and tasks. This dataset was released as part of the paper MolmoAct: Action Reasoning Models that can Reason in Space.
It serves as a crucial component for Action Reasoning Models (ARMs), enabling explainable and steerable robotic behavior. Training with this dataset yields significant improvements in general performance over base models.
All contents in this dataset were collected in-house by Ai2 and are provided in LeRobot format.
Quick Links:
- 📄 Paper
- 💻 GitHub Repository
- 📂 All Models
- 📂 All Data
- 🎬 Blog Post
- 🎬 Video
License and Use
This dataset is licensed under CC BY-4.0. It is intended for research and educational use in accordance with Ai2's Responsible Use Guidelines.
Citation
If you find this dataset useful for your research and applications, please cite the original paper:
@misc{molmoact2025,
title={MolmoAct: Action Reasoning Models that can Reason in Space},
author={Jason Lee and Jiafei Duan and Haoquan Fang and Yuquan Deng and Shuo Liu and Boyang Li and Bohan Fang and Jieyu Zhang and Yi Ru Wang and Sangho Lee and Winson Han and Wilbert Pumacay and Angelica Wu and Rose Hendrix and Karen Farley and Eli VanderBilt and Ali Farhadi and Dieter Fox and Ranjay Krishna},
year={2025},
eprint={2508.07917},
archivePrefix={arXiv},
primaryClass={cs.RO},
url={https://arxiv.org/abs/2508.07917}
}