# CapVector: Learning Transferable Capability Vectors in Parametric Space for Vision-Language-Action Models

<div align="center">

[![Paper](https://img.shields.io/badge/Paper-A42C25?style=for-the-badge&logo=arxiv&logoColor=white)](http://arxiv.org/abs/) [![Page](https://img.shields.io/badge/Project--Page-blue?style=for-the-badge&logo=homepage&logoColor=white)](https://capvector.github.io/) [![Hugging Face Collection](https://img.shields.io/badge/Models-fcd022?style=for-the-badge&logo=huggingface&logoColor=white)](https://huggingface.co/haofuly/capvector_models_collection)

</div>

CapVector is a training recipe for vision-language-action (VLA) models that extracts a transferable capability vector from the parameter difference between auxiliary-objective SFT methods and standard SFT methods. This vector is merged into a pretrained VLA to form a stronger initialization, and downstream adaptation uses standard SFT with a lightweight orthogonal regularization loss to preserve the injected capability.


## 🌟 Key Features
- **Efficient downstream adaptation**: CapVector recovers much of the benefit of auxiliary-objective SFT methods, while keeping the downstream overhead close to standard SFT.
- **Versatility**: CapVector fits for OpenVLA-based, OpenPi-based, and StarVLA-based backbones.
- **Generalization**: CapVector is designed to transfer across tasks, environments, and robot embodiments.


## 🚀 Get Started

This repository provides two implementation paths: 
- [`capvector-oft/`](./capvector-oft) based implementation
- [`capvector-pi05/`](./capvector-pi05) based implementation. 

Choose the subdirectory that matches your base model and training stack. Follow the subproject README for environment setup, data preparation, training, and inference.

[`capvector-pi05/`](./capvector-pi05) provides the capability vector extraction and merging scripts.


## 🌏 Contact
For further discussion and collaboration, please feel free to contact us via Email and WeChat:

| Author | Email | WeChat |
|:---:|:---:|:---:|
| Wenxuan Song | songwenxuan0115@gmail.com | swx0757 |


## ❤️ Acknowledgments

CapVector builds on and interfaces with several excellent open-source projects, including:

- [OpenVLA-OFT](https://github.com/moojink/openvla-oft)
- [OpenPI](https://github.com/Physical-Intelligence/openpi)


## 🖊 Citation

If you find this work useful, please cite:

```bibtex
@article{song2026capvector,
  title   = {CapVector: Learning Transferable Capability Vectors in Parametric Space for Vision-Language-Action Models},
  author  = {Song, Wenxuan and Zhao, Han and Li, Fuhao and Zhou, Ziyang and Wang, Xi and Lyu, Jing and Ding, Pengxiang and Wang, Yan and Wang, Donglin and Li, Haoang},
  journal = {Preprint},
  year    = {2026}
}
```