---
library_name: araclip
tags:
- clip
- model_hub_mixin
- pytorch_model_hub_mixin
---

This model has been pushed to the Hub using the [PytorchModelHubMixin](https://huggingface.co/docs/huggingface_hub/package_reference/mixins#huggingface_hub.PyTorchModelHubMixin) integration:
- Library: https://github.com/Arabic-Clip/Araclip.git

## How to use 
```
pip install git+https://github.com/Arabic-Clip/Araclip.git
```
```python
# load model
import numpy as np
from PIL import Image
from araclip import AraClip
model = AraClip.from_pretrained("Arabic-Clip/araclip")

# data
labels = ["قطة جالسة", "قطة تقفز" ,"كلب", "حصان"]
image = Image.open("cat.png")

# embed data 
image_features = model.embed(image=image)
text_features = np.stack([model.embed(text=label) for label in labels])

# search for most similar data
similarities = text_features @ image_features
best_match = labels[np.argmax(similarities)]

print(f"The image is most similar to: {best_match}")
# قطة جالسة
```


![image/png](https://cdn-uploads.huggingface.co/production/uploads/6527e89a8808d80ccff88b7a/d5i4ItET9AZN9xgv8ify5.png)