nielsr HF Staff commited on
Commit
5ed4e29
·
verified ·
1 Parent(s): f3b20e3

Add pipeline tag and paper link

Browse files

This PR adds the `robotics` pipeline tag and a link to the paper page, making the model easier to discover and ensuring proper attribution.

Files changed (1) hide show
  1. README.md +10 -7
README.md CHANGED
@@ -1,12 +1,13 @@
1
  ---
2
  license: apache-2.0
 
3
  tags:
4
- - robotics
5
- - vla
6
- - quantization
7
- - w4a4
8
- - pi0.5
9
- - libero
10
  ---
11
 
12
  # Omega-QVLA — pi0.5 LIBERO W4A4 quantization packs
@@ -14,6 +15,8 @@ tags:
14
  W4A4 quantization packs for the **pi0.5 (openpi) LIBERO** action policies, produced
15
  by [Omega-QVLA](https://github.com/UCMP13753/Omega-QVLA).
16
 
 
 
17
  ## Recipe (per side)
18
 
19
  | side | rotation | quantizer | per-step |
@@ -60,4 +63,4 @@ env CONDA_ROOT=$HOME/miniconda3 METHOD=gptq SUITE=$SUITE WBITS=4 ABITS=4 \
60
  ```
61
 
62
  Both sides load through the GPTQ pack path (`GptqLinear`); the block-form rotation
63
- and per-step scales are applied automatically.
 
1
  ---
2
  license: apache-2.0
3
+ pipeline_tag: robotics
4
  tags:
5
+ - robotics
6
+ - vla
7
+ - quantization
8
+ - w4a4
9
+ - pi0.5
10
+ - libero
11
  ---
12
 
13
  # Omega-QVLA — pi0.5 LIBERO W4A4 quantization packs
 
15
  W4A4 quantization packs for the **pi0.5 (openpi) LIBERO** action policies, produced
16
  by [Omega-QVLA](https://github.com/UCMP13753/Omega-QVLA).
17
 
18
+ This repository accompanies the paper [Ω-QVLA: Robust Quantization for Vision-Language-Action Models via Composite Rotation and Per-step Scaling](https://huggingface.co/papers/2605.28803).
19
+
20
  ## Recipe (per side)
21
 
22
  | side | rotation | quantizer | per-step |
 
63
  ```
64
 
65
  Both sides load through the GPTQ pack path (`GptqLinear`); the block-form rotation
66
+ and per-step scales are applied automatically.