Improve model card: add library\_name, pipeline\_tag, and links

This PR improves the model card by:

* Adding the `library_name` to the metadata.
* Adding the `pipeline_tag` to the metadata, ensuring the model can be found at [https://huggingface.co/models?pipeline\_tag=image-text-to-text](https://huggingface.co/models?pipeline_tag=image-text-to-text).
* Linking to the paper.
* Adding a brief description of the model.

README.md

@@ -2,9 +2,13 @@
 language:
 - en
 license: apache-2.0
+library_name: transformers
+pipeline_tag: image-text-to-text
 ---
+
 This is BLIP3o-8B checkpoint trained on the **open source** data.
 
+This model is a visual language model based on the paper [BLIP3-o: A Family of Fully Open Unified Multimodal Models-Architecture, Training and Dataset](https://huggingface.co/papers/2505.09568). The code is available at https://github.com/JiuhaiChen/BLIP3o.
 
 | Model               | Pretrain Data                                             | GenEval | DBP    | WISE |
 |---------------------|-----------------------------------------------------------|---------|--------|------|
@@ -12,8 +16,6 @@ This is BLIP3o-8B checkpoint trained on the **open source** data.
 | 8B (open source)    | 30 million open-source data                           | 0.83    | 80.73  | 0.52 |
 | 8B (paper reported) | 30 million open-source + 30 million proprietary data  | 0.84    | 81.60  | 0.62 |
 
-
-
 Here is the category results for WISE.
 
 |       Model      |        Pretrain Data        | Cultural | Time | Space | Biology | Physics | Chemistry | Overall |
@@ -21,16 +23,6 @@ Here is the category results for WISE.
 | 8B (open source) | 30 million open-source data |   0.49   | 0.51 |  0.63 |   0.54  |   0.63  |    0.37   |   0.52  |
 | 8B (paper reported)  | 30 million open-source + 30 million proprietary data|  0.63  | 0.57 | 0.70 |  0.62 |  0.66 |   0.51  |  0.62   |
 
-
-
-
-
-
-
-
-
-
-
 ### Download
 
 ```
@@ -57,4 +49,4 @@ Launch with your model path:
 
 ```
 python app.py /path/to/your/model
-```
+```