Update README.md
Browse files
README.md
CHANGED
|
@@ -30,12 +30,12 @@ tags:
|
|
| 30 |
</p>
|
| 31 |
|
| 32 |
|
| 33 |
-
|
| 34 |
## π Introduction
|
| 35 |
|
| 36 |
**UniPic2-Metaquery-9B** is an unified multimodal model built on Qwen2.5-VL-Instruct and SD3.5-Medium. It delivers end-to-end image understanding, text-to-image (T2I) generation, and image editing, and runs smoothly on a single 16 GB consumer GPU.
|
| 37 |
<div align="center"> <img src="teaser.png" alt="Model Teaser" width="720"> </div>
|
| 38 |
|
|
|
|
| 39 |
## π Benchmarks
|
| 40 |
|
| 41 |
**UniPic2-Metaquery-9B** w/o GRPO achieves competitive results across a variety of vision-language tasks:
|
|
@@ -47,7 +47,7 @@ tags:
|
|
| 47 |
| βοΈ **GEditBench-EN** | 6.90 |
|
| 48 |
| π§ͺ **ImgEdit-Bench** | 4.10 |
|
| 49 |
|
| 50 |
-
|
| 51 |
|
| 52 |
## π§ Usage
|
| 53 |
|
|
@@ -203,10 +203,11 @@ edited_image = pipeline(
|
|
| 203 |
edited_image.save("image_editing.png")
|
| 204 |
```
|
| 205 |
|
| 206 |
-
## π License
|
| 207 |
|
|
|
|
| 208 |
This model is released under the MIT License.
|
| 209 |
|
|
|
|
| 210 |
## Citation
|
| 211 |
If you use Skywork-UniPic in your research, please cite:
|
| 212 |
```
|
|
|
|
| 30 |
</p>
|
| 31 |
|
| 32 |
|
|
|
|
| 33 |
## π Introduction
|
| 34 |
|
| 35 |
**UniPic2-Metaquery-9B** is an unified multimodal model built on Qwen2.5-VL-Instruct and SD3.5-Medium. It delivers end-to-end image understanding, text-to-image (T2I) generation, and image editing, and runs smoothly on a single 16 GB consumer GPU.
|
| 36 |
<div align="center"> <img src="teaser.png" alt="Model Teaser" width="720"> </div>
|
| 37 |
|
| 38 |
+
|
| 39 |
## π Benchmarks
|
| 40 |
|
| 41 |
**UniPic2-Metaquery-9B** w/o GRPO achieves competitive results across a variety of vision-language tasks:
|
|
|
|
| 47 |
| βοΈ **GEditBench-EN** | 6.90 |
|
| 48 |
| π§ͺ **ImgEdit-Bench** | 4.10 |
|
| 49 |
|
| 50 |
+
---
|
| 51 |
|
| 52 |
## π§ Usage
|
| 53 |
|
|
|
|
| 203 |
edited_image.save("image_editing.png")
|
| 204 |
```
|
| 205 |
|
|
|
|
| 206 |
|
| 207 |
+
## π License
|
| 208 |
This model is released under the MIT License.
|
| 209 |
|
| 210 |
+
|
| 211 |
## Citation
|
| 212 |
If you use Skywork-UniPic in your research, please cite:
|
| 213 |
```
|