Add pipeline tag and clean metadata

This PR adds the missing `pipeline_tag: image-classification` to the model card metadata and removes the arXiv ID from the metadata tags, since the arXiv ID should only appear in the markdown content. The model card content is otherwise preserved.

README.md

@@ -1,16 +1,17 @@
 ---
-license: mit
 base_model: openai/clip-vit-base-patch32
+license: mit
+pipeline_tag: image-classification
 tags:
 - smat
 - model-merging
 - vision
-- arxiv:2609.33437
 ---
+
 # CLIP ViT base-patch32 · FT · SUN397
 
 A FT expert from Table 2 of **SMAT: Simple and Efficient Merge-Aware Training** (seed 42). SMAT trains experts with model merging in mind.
 
 [Paper](https://arxiv.org/abs/2609.33437) · [GitHub & usage](https://github.com/egangu/smat/blob/main/docs/HUGGINGFACE.md)
 
-`encoder.pt` is the original FP32 vision-encoder state dictionary; load it with the SMAT code and the matching OpenAI CLIP base model. Dataset terms apply separately.
+`encoder.pt` is the original FP32 vision-encoder state dictionary; load it with the SMAT code and the matching OpenAI CLIP base model. Dataset terms apply separately.