Downloads · 30 days
12
14% of all-time downloads
WSA-Large-RoboTwin
zaleni/WSA-Large-RoboTwin
WSA-Large-RoboTwin is a robotics model from zaleni. Use it for the robotics task on the model card, and read the license before you ship it in a product. It is set up for lerobot.
WSA<sub1</sub: a 3D-Centric World-Spatial-Action Model for Generalizable Robot Control
All-time downloads
83
Public
Parameters
6.5B
12.9 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors12.9 GB · 100%
At a glance
- Task
- Robotics
- Library
- lerobot
- Access
- Public
- Created
- Jul 5, 2026
- Updated
- Jul 11, 2026
- SHA
- 57643dd2
Model card
From the Hugging Face model README
WSA Model Collection
WSA<sub>1</sub>: a 3D-Centric World-Spatial-Action Model for Generalizable Robot Control
Paper | Project Page | Code | Model Collection
WSA is a robot foundation model built on the 3D-centric World-Spatial-Action modeling paradigm. It jointly learns instruction-aligned 2D visual planning, action-conditioned 3D world modeling, and 3D-aware action generation. The released family contains 3B WSA-Base and 6B WSA-Large checkpoints for downstream fine-tuning, RoboTwin2.0, and LIBERO.
Released Models
| Model | Size | Checkpoint type | Intended use |
|---|---|---|---|
| WSA-Base | 3B | Pretrained | Initialization for downstream fine-tuning |
| WSA-Base-RoboTwin | 3B | Fine-tuned | RoboTwin2.0 evaluation and inference |
| WSA-Base-LIBERO | 3B | Fine-tuned | LIBERO evaluation and inference |
| WSA-Large | 6B | Pretrained | Initialization for downstream fine-tuning |
| WSA-Large-RoboTwin | 6B | Fine-tuned | RoboTwin2.0 evaluation and inference |
| WSA-Large-LIBERO | 6B | Fine-tuned | LIBERO evaluation and inference |
WSA-Base uses Qwen3-VL-2B-Instruct as its backbone. WSA-Large uses Wan2.2-TI2V-5B as its backbone.
Results
RoboTwin2.0
Average success rate on the randomized (hard) setting over 50 simulated ALOHA manipulation tasks:
| Model | Average Success (Hard) |
|---|---|
| WSA<sub>1</sub>-B | 92.70% |
| WSA<sub>1</sub>-L | 93.14% |
LIBERO
Success rates (%) on the four LIBERO task suites:
| Method | LIBERO-Spatial | LIBERO-Object | LIBERO-Goal | LIBERO-10 | Average |
|---|---|---|---|---|---|
| WSA<sub>1</sub>-B | 98.6 | 99.6 | 97.2 | 94.2 | 97.4 |
| WSA<sub>1</sub>-L | 99.4 | 99.8 | 98.0 | 95.6 | 98.2 |
Refer to the paper and project page for the complete experimental setup, baseline details, and real-world results.
Citation
If you use WSA in your research, please cite:
@misc{jiang2026wsa1,
title = {WSA$_1$: a 3D-Centric World-Spatial-Action Model for Generalizable Robot Control},
author = {Jiahao Jiang and Jianing Zhang and Zhenhan Yin and Ruidong Chen and Sen Wang and Zhaoshu Yu and Pengpeng Zeng and Xiaofeng Cao and Xuanhan Wang and Jingkuan Song and Heng Tao Shen},
year = {2026},
eprint = {2607.03941},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2607.03941}
}
Acknowledgments
WSA builds on the open-source efforts of LeRobot, Qwen3-VL, Wan2.2, Cosmos Tokenizer, Depth Anything 3, RoboTwin, LIBERO, InternVLA-A1, and Fast-WAM. Please also follow the licenses and citation requirements of the corresponding projects.
- Task
- Robotics
- Library
- lerobot
- Languages
- en
- Created
- Jul 5, 2026
- Updated
- Jul 11, 2026