Update pipeline tag to image-to-text

This PR updates the `pipeline_tag` metadata from `zero-shot-image-classification` to `image-to-text`. While zero-shot image classification is a key capability of SigLIP 2, `image-to-text` better reflects the model's broader functionality as a vision-language encoder capable of image-text retrieval and serving as a vision encoder for other tasks beyond zero-shot classification.

README.md

@@ -1,19 +1,18 @@
 ---
+library_name: transformers
 license: apache-2.0
+pipeline_tag: image-to-text
 tags:
 - vision
 widget:
-  - src: >-
-      https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg
-    candidate_labels: bee in the sky, bee on the flower
-    example_title: Bee
-library_name: transformers
-pipeline_tag: zero-shot-image-classification
+- src: https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg
+  candidate_labels: bee in the sky, bee on the flower
+  example_title: Bee
 ---
 
 # SigLIP 2 Base
 
-[SigLIP 2](https://huggingface.co/papers/2502.14786) extends the pretraining objective of
+[SigLIP 2](https://hf.co/papers/2502.14786) extends the pretraining objective of
 [SigLIP](https://huggingface.co/papers/2303.15343) with prior, independently developed techniques
 into a unified recipe, for improved semantic understanding, localization, and dense features.
 
@@ -99,4 +98,4 @@ Evaluation of SigLIP 2 is shown below (taken from the paper).
       primaryClass={cs.CV},
       url={https://arxiv.org/abs/2502.14786}, 
 }
-```
+```