Improve model card: add robotics pipeline tag and project links
#2
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,11 +1,13 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
| 4 |
# PhysBrain-VLA
|
| 5 |
|
| 6 |
PhysBrain 1.0 — Physical Intelligence for Embodied General AI
|
| 7 |
|
| 8 |
-
[
|
| 9 |
|
| 10 |
---
|
| 11 |
|
|
@@ -35,6 +37,33 @@ A principled training methodology that breaks the **visual shortcut dilemma** in
|
|
| 35 |
|
| 36 |
Together, these three technologies form the PhysBrain 1.0 system: **PhysBrain** (base model) × **TwinBrainVLA** (architecture) × **LangForce** (training strategy) — achieving SOTA across multiple embodied intelligence benchmarks with exceptional data efficiency.
|
| 37 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 38 |
## Citation
|
| 39 |
|
| 40 |
If you find PhysBrain useful in your research, please consider citing our work:
|
|
@@ -63,4 +92,4 @@ If you find PhysBrain useful in your research, please consider citing our work:
|
|
| 63 |
|
| 64 |
## License
|
| 65 |
|
| 66 |
-
This project is released under the [Apache 2.0 License](LICENSE).
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
pipeline_tag: robotics
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
# PhysBrain-VLA
|
| 7 |
|
| 8 |
PhysBrain 1.0 — Physical Intelligence for Embodied General AI
|
| 9 |
|
| 10 |
+
[[📄 Paper](https://huggingface.co/papers/2605.15298)] [[🏠 Project Page](https://phys-brain.github.io/)] [[💻 Code](https://github.com/Phys-Brain/PhysBrain-VLA)]
|
| 11 |
|
| 12 |
---
|
| 13 |
|
|
|
|
| 37 |
|
| 38 |
Together, these three technologies form the PhysBrain 1.0 system: **PhysBrain** (base model) × **TwinBrainVLA** (architecture) × **LangForce** (training strategy) — achieving SOTA across multiple embodied intelligence benchmarks with exceptional data efficiency.
|
| 39 |
|
| 40 |
+
## Open Source Plan
|
| 41 |
+
|
| 42 |
+
All PhysBrain 1.0 VLA model checkpoints are now available. You can find the full collection at [🤗 Hugging Face](https://huggingface.co/collections/Phys-Brain/physbrain-10-vla).
|
| 43 |
+
|
| 44 |
+
The current release status is as follows:
|
| 45 |
+
|
| 46 |
+
| Component | Status |
|
| 47 |
+
| ------------------------------------------------------ | ------------ |
|
| 48 |
+
| PhysBrain 1.0 VLA (RoboCasa Fine-Tuned) | ✅ Available |
|
| 49 |
+
| PhysBrain 1.0 VLA (LIBERO Fine-Tuned) | ✅ Available |
|
| 50 |
+
| PhysBrain 1.0 VLA (SIMPLER WidowX Robot Fine-Tuned) | ✅ Available |
|
| 51 |
+
| PhysBrain 1.0 VLA (SIMPLER Google Robot Fine-Tuned) | ✅ Available |
|
| 52 |
+
| Inference Code | ✅ Available |
|
| 53 |
+
|
| 54 |
+
## Getting Started
|
| 55 |
+
|
| 56 |
+
PhysBrain-VLA is built on top of the **starVLA** scaffold. To use it, follow these two steps:
|
| 57 |
+
|
| 58 |
+
1. **Copy the framework file** into your starVLA codebase:
|
| 59 |
+
```powershell
|
| 60 |
+
cp physbrain_vla/PhysBrainVLA.py <path-to-starVLA>/starVLA/model/framework/
|
| 61 |
+
```
|
| 62 |
+
|
| 63 |
+
2. **Load and deploy** the model following the standard starVLA checkpoint loading workflow.
|
| 64 |
+
|
| 65 |
+
For detailed starVLA setup and inference instructions, please refer to the starVLA repository.
|
| 66 |
+
|
| 67 |
## Citation
|
| 68 |
|
| 69 |
If you find PhysBrain useful in your research, please consider citing our work:
|
|
|
|
| 92 |
|
| 93 |
## License
|
| 94 |
|
| 95 |
+
This project is released under the [Apache 2.0 License](LICENSE).
|