Fix task tag

Hello, we have CLIP-like models (multimodal image retrieval/embeddings) under zero-shot-image-classification pipeline tag, so fixing it

README.md

@@ -14,7 +14,7 @@ language:
 - fr
 library_name: transformers
 license: mit
-pipeline_tag: image-feature-extraction
+pipeline_tag: zero-shot-image-classification
 ---
 
 ## mmE5-mllama-11b-instruct
@@ -157,4 +157,4 @@ print(similarity)
   journal={arXiv preprint arXiv:2502.08468},
   year={2025}
 }
-```
+```