Add missing metadata: `pipeline_tag`, `library_name`, and `license`

This PR adds missing metadata to the model card, including the `pipeline_tag`, `library_name`, and `license`. This improves discoverability and clarifies the model's capabilities and usage. The `pipeline_tag` is set to `image-text-to-text` reflecting the model's multimodal nature. The `library_name` is set to `transformers` based on the file structure. Please replace 'mit' with the actual license if available.

README.md

@@ -1,3 +1,8 @@
+---
+license: mit # Please replace 'mit' with the actual license if available
+library_name: transformers
+pipeline_tag: image-text-to-text
+---
 
 ### Introduction
 Paper: [Paper](https://arxiv.org/abs/2502.18411),
@@ -45,7 +50,4 @@ By integrating OmniAlign-V-DPO datasets in DPO stage, we can further improve the
 | Idefics3-8B                   | 2.7                          | -92.3                     | 3          | 4   | 0  | 15  | 230 |
 
 ### How to use
-Please refer to our [Github](https://github.com/PhoenixZ810/OmniAlign-V) for more details about training and evaluation.
-
-
-
+Please refer to our [Github](https://github.com/PhoenixZ810/OmniAlign-V) for more details about training and evaluation.