update README.md
Browse files
README.md
CHANGED
|
@@ -1,16 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
# TBAC-VLR1-3B-preview
|
| 2 |
|
| 3 |
-
Trained by TencentBAC on Qwen2.5-VL-3B-Instruct through GRPO
|
| 4 |
-
|
| 5 |
## Performance
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
|
| 7 |
-
|
| 8 |
-
| :-------------------: | :---------: | :-------: | :------------: | :-----------: | :-----------: | :-------: | :----------: |
|
| 9 |
-
| Taichu-VLR-3B | 33.6 | 64.9 | 23.1 | 32.1 | 12.6 | 30.4 | 38.7 |
|
| 10 |
-
| VLAA-Thinker-Qwen2.5VL-3B | 35.4 | 61 | 24.4 | 36.4 | 18.2 | 33.8 | 38.5 |
|
| 11 |
-
| **TBAC-VLR1-3B-preview** | **35.7** | 64.8 | 25.0 | 33.2 | 17.7 | 32.4 | 40.8 |
|
| 12 |
|
| 13 |
-
|
| 14 |
|
| 15 |
## Usage
|
| 16 |
```python
|
|
@@ -67,9 +81,10 @@ print(output_text)
|
|
| 67 |
## Citation
|
| 68 |
If you find our model useful in your research, please consider giving ❤️ and citations. Thanks!
|
| 69 |
```
|
| 70 |
-
@misc{
|
| 71 |
title={TBAC-VLR1-3B-preview},
|
| 72 |
author={Junzhe Xu and Yuyang yin},
|
|
|
|
| 73 |
year={2025},
|
| 74 |
}
|
| 75 |
```
|
|
@@ -78,8 +93,4 @@ If you find our model useful in your research, please consider giving ❤️ and
|
|
| 78 |
|
| 79 |
**About**
|
| 80 |
|
| 81 |
-
Created by the Tencent BAC Group. All rights reserved.
|
| 82 |
-
|
| 83 |
-
**License**
|
| 84 |
-
|
| 85 |
-
This work is licensed under a [Creative Commons Attribution-NonCommercial 4.0 International License](https://creativecommons.org/licenses/by-nc/4.0/).
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
base_model:
|
| 6 |
+
- Qwen/Qwen2.5-VL-3B-Instruct
|
| 7 |
+
pipeline_tag: visual-question-answering
|
| 8 |
+
---
|
| 9 |
# TBAC-VLR1-3B-preview
|
| 10 |
|
|
|
|
|
|
|
| 11 |
## Performance
|
| 12 |
+
| Model | **Average** | **MathVista**| **MathVision** | **MathVerse** | **DynaMath** | **WeMath**| **LogicVista** |
|
| 13 |
+
| :-------------------: | :---------: | :-----------:| :------------: | :-----------: | :-----------: | :-------: | :----------: |
|
| 14 |
+
| Qwen2-VL-2B | 20.5 | 48.0 | 16.1 | 17.5 | 3.8 | 10.8 | 26.6 |
|
| 15 |
+
| InternVL2.5-2B | 21.2 | 51.1 | 14.0 | 22.3 | 4.4 | 8.0 | 27.3 |
|
| 16 |
+
| InternVL3-2B | 29.1 | 57.6 | 20.2 | 24.5 | 14.8 | 22.9 | 40.3 |
|
| 17 |
+
| Qwen2.5-VL-3B | 31.8 | 61.2 | 21.9 | 31.2 | 13.2 | 22.9 | 40.3 |
|
| 18 |
+
| VLM-R1-3B-Math-0305 | 33.4 | 62.7 | 21.9 | 32.2 | 13.0 | 30.0 | 40.5 |
|
| 19 |
+
| Taichu-VLR-3B | 33.6 | 64.9 | 23.1 | 32.1 | 12.6 | 30.4 | 38.7 |
|
| 20 |
+
| VLAA-Thinker-Qwen2.5VL-3B | 35.4 | 61.0 | 24.4 | 36.4 | 18.2 | 33.8 | 38.5 |
|
| 21 |
+
| **TBAC-VLR1-3B-preview** | **35.7** | 64.8 | 25.0 | 33.2 | 17.7 | 32.4 | 40.8 |
|
| 22 |
+
|
| 23 |
+
|
| 24 |
|
| 25 |
+
The compared results are sourced from https://opencompass.org.cn.
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
|
| 27 |
+
The results of our model are self-reported, obtained by running evaluations offline on each benchmark.
|
| 28 |
|
| 29 |
## Usage
|
| 30 |
```python
|
|
|
|
| 81 |
## Citation
|
| 82 |
If you find our model useful in your research, please consider giving ❤️ and citations. Thanks!
|
| 83 |
```
|
| 84 |
+
@misc{Xu2025tbacvlr1,
|
| 85 |
title={TBAC-VLR1-3B-preview},
|
| 86 |
author={Junzhe Xu and Yuyang yin},
|
| 87 |
+
url={https://huggingface.co/TencentBAC/TBAC-VLR1-3B-preview},
|
| 88 |
year={2025},
|
| 89 |
}
|
| 90 |
```
|
|
|
|
| 93 |
|
| 94 |
**About**
|
| 95 |
|
| 96 |
+
Created by the Tencent BAC Group. All rights reserved.
|
|
|
|
|
|
|
|
|
|
|
|