Commit
·
ec3cb26
1
Parent(s):
790eb33
update the bibtex for neurips
Browse files
README.md
CHANGED
|
@@ -65,10 +65,10 @@ For integration with continuous generative models:
|
|
| 65 |
|
| 66 |
```python
|
| 67 |
# Encode to normalized continuous parameters [-1, 1]
|
| 68 |
-
params =
|
| 69 |
|
| 70 |
# Decode back
|
| 71 |
-
reconstructed =
|
| 72 |
```
|
| 73 |
|
| 74 |
## Parameters
|
|
@@ -82,9 +82,7 @@ reconstructed = tokenizer.decode_continuous(params)
|
|
| 82 |
| `degree_p` | B-spline polynomial degree. Higher degrees produce smoother curves (3=cubic, 4=quartic) | 4 |
|
| 83 |
| `device` | Torch device (`"cuda"` or `"cpu"`) | `"cuda"` |
|
| 84 |
| `gripper_zero_order` | Use piecewise-constant (degree 0) splines for gripper DOFs. Useful for binary gripper states | `False` |
|
| 85 |
-
| `gripper_dof` | Number of gripper DOFs. Only used when `gripper_zero_order=True` | 1 |
|
| 86 |
-
| `init_cond_order` | Initial boundary condition order: 0=none, 1=position only, 2=position+velocity | 0 |
|
| 87 |
-
| `end_cond_order` | End boundary condition order (same options as `init_cond_order`) | 0 |
|
| 88 |
| `enforce_init_pos` | Enforce initial position constraint during decoding | `False` |
|
| 89 |
|
| 90 |
### Token Count
|
|
@@ -144,11 +142,12 @@ If you use BEAST in your research, please cite:
|
|
| 144 |
**BibTeX:**
|
| 145 |
|
| 146 |
```bibtex
|
| 147 |
-
@
|
| 148 |
-
|
| 149 |
-
|
| 150 |
-
|
| 151 |
-
|
|
|
|
|
|
|
| 152 |
}
|
| 153 |
-
|
| 154 |
```
|
|
|
|
| 65 |
|
| 66 |
```python
|
| 67 |
# Encode to normalized continuous parameters [-1, 1]
|
| 68 |
+
params = beast.encode_continuous(trajectories, update_bounds=True)
|
| 69 |
|
| 70 |
# Decode back
|
| 71 |
+
reconstructed = beast.decode_continuous(params)
|
| 72 |
```
|
| 73 |
|
| 74 |
## Parameters
|
|
|
|
| 82 |
| `degree_p` | B-spline polynomial degree. Higher degrees produce smoother curves (3=cubic, 4=quartic) | 4 |
|
| 83 |
| `device` | Torch device (`"cuda"` or `"cpu"`) | `"cuda"` |
|
| 84 |
| `gripper_zero_order` | Use piecewise-constant (degree 0) splines for gripper DOFs. Useful for binary gripper states | `False` |
|
| 85 |
+
| `gripper_dof` | Number of gripper DOFs, assumed to be in the end. Only used when `gripper_zero_order=True` | 1 |
|
|
|
|
|
|
|
| 86 |
| `enforce_init_pos` | Enforce initial position constraint during decoding | `False` |
|
| 87 |
|
| 88 |
### Token Count
|
|
|
|
| 142 |
**BibTeX:**
|
| 143 |
|
| 144 |
```bibtex
|
| 145 |
+
@inproceedings{
|
| 146 |
+
zhou2025beast,
|
| 147 |
+
title={{BEAST}: Efficient Tokenization of B-Splines Encoded Action Sequences for Imitation Learning},
|
| 148 |
+
author={Hongyi Zhou and Weiran Liao and Xi Huang and Yucheng Tang and Fabian Otto and Xiaogang Jia and Xinkai Jiang and Simon Hilber and Ge Li and Qian Wang and {\"O}mer Erdin{\c{c}} Ya{\u{g}}murlu and Nils Blank and Moritz Reuss and Rudolf Lioutikov},
|
| 149 |
+
booktitle={The Thirty-ninth Annual Conference on Neural Information Processing Systems},
|
| 150 |
+
year={2025},
|
| 151 |
+
url={https://openreview.net/forum?id=rQCl1sf62w}
|
| 152 |
}
|
|
|
|
| 153 |
```
|