Update README.md
Browse files
README.md
CHANGED
|
@@ -1,5 +1,27 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
language:
|
| 4 |
-
- en
|
| 5 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
datasets:
|
| 6 |
+
- nvidia/PhysicalAI-Robotics-GR00T-Teleop-Sim
|
| 7 |
+
---
|
| 8 |
+
|
| 9 |
+
# DIAL Checkpoints
|
| 10 |
+
|
| 11 |
+
<p align="left">
|
| 12 |
+
<a href="https://chenyi99.github.io/dial/">Project Page</a> |
|
| 13 |
+
<a href="https://chenyi99.github.io/dial/DIAL.pdf">Paper</a> |
|
| 14 |
+
<a href="https://github.com/xpeng-robotics/DIAL">Code</a>
|
| 15 |
+
</p>
|
| 16 |
+
|
| 17 |
+
Model weights for **DIAL** (**D**ecoupling **I**ntent and **A**ction via **L**atent World Modeling), an end-to-end Vision-Language-Action (VLA) framework built on [NVIDIA Isaac GR00T N1.5](https://github.com/NVIDIA/Isaac-GR00T/tree/n1.5-release) with a [Qwen2.5-VL-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct) backbone.
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
## Available Checkpoints
|
| 21 |
+
|
| 22 |
+
| Checkpoint | Training Data | Steps | Description |
|
| 23 |
+
|---|---|---|---|
|
| 24 |
+
| `DIAL-3B-fewshot` | EgoDex human data + 10% GR1 simulation data | 20K per stage (3-stage) | Co-trained with heterogeneous human demonstrations |
|
| 25 |
+
| `DIAL-3B-fulldata` | All GR1 simulation data (~24,000 demos) | 40K per stage (2-stage) | Trained on full teleoperation trajectories in simulation |
|
| 26 |
+
|
| 27 |
+
For installation, training, and evaluation instructions, please refer to the [GitHub repository](https://github.com/xpeng-robotics/DIAL).
|