---
annotations_creators: [manual]
language: [en]
license: apache-2.0
multilinguality: [monolingual]
pretty_name: ARM4R Dataset
size_categories: [100K<examples<1M]
source_datasets: [original]
task_categories: [robotics]
task_ids: [grasping]
---

# Pre-training Auto-regressive Robotic Models with 4D Representations

by [Dantong Niu*](https://scholar.google.com/citations?user=AzlUrvUAAAAJ&hl=en), [Yuvan Sharma*](https://scholar.google.com/citations?hl=en&user=1_IIcds8es4C), [Haoru Xue](https://haoruxue.github.io/), [Giscard Biamby](https://scholar.google.com/citations?user=s0Fof5IAAAAJ&hl=en), [Junyi Zhang](https://www.junyi42.com/), [Ziteng Ji](https://scholar.google.com/citations?user=VIskd7AAAAAJ&hl=zh-CN), [Trevor Darrell†](https://people.eecs.berkeley.edu/~trevor/), and [Roei Herzig†](https://roeiherz.github.io/)  

<sup>*Equal contribution, †Equal advising</sup>  
**Berkeley AI Research, UC Berkeley**  
**ICML 2025**

[**Paper**](https://arxiv.org/pdf/2502.13142) • [**Code**](https://github.com/Dantong88/arm4r) • [**Models**](https://huggingface.co/datasets/yuvansharma/arm4r-ckpts) • [**Dataset**](https://huggingface.co/datasets/yuvansharma/arm4r-data)


The structure for the data is as follows:

```
.
├── .gitattributes
├── README.md
├── epic_clips.json  # contains mapping for episode id --> language instruction (76,014 episodes)
├── epic_tasks_final.zip  # contains extracted 3D point data for Epic-Kitchens (76,014 episodes)
└── real_kinova_release_data.zip # contains collected data for real world Kinova Gen3 setup (2,550 episodes)

```
## Citation

If you find our work helpful, please consider citing:

```bibtex
@article{niu2025pre,
  title={Pre-training auto-regressive robotic models with 4d representations},
  author={Niu, Dantong and Sharma, Yuvan and Xue, Haoru and Biamby, Giscard and Zhang, Junyi and Ji, Ziteng and Darrell, Trevor and Herzig, Roei},
  journal={arXiv preprint arXiv:2502.13142},
  year={2025}
}