Update README.md
Browse files
README.md
CHANGED
|
@@ -1,27 +1,27 @@
|
|
| 1 |
-
|
| 2 |
-
---
|
| 3 |
-
|
| 4 |
-
## 📝 `README.md` pour le Processor
|
| 5 |
-
|
| 6 |
-
```markdown
|
| 7 |
-
# Wolof GIT Processor (Image + Text)
|
| 8 |
|
| 9 |
Ce processor combine :
|
| 10 |
|
| 11 |
-
- Le tokenizer
|
| 12 |
-
- Le processor image du modèle `microsoft/git-base`
|
| 13 |
|
| 14 |
-
Il
|
| 15 |
|
| 16 |
-
|
|
|
|
|
|
|
| 17 |
|
| 18 |
```python
|
| 19 |
from transformers import AutoProcessor
|
|
|
|
| 20 |
|
| 21 |
processor = AutoProcessor.from_pretrained("ZigZeug/wolof-git-processor")
|
| 22 |
|
|
|
|
|
|
|
|
|
|
| 23 |
inputs = processor(
|
| 24 |
-
images=
|
| 25 |
-
text=
|
| 26 |
return_tensors="pt"
|
| 27 |
)
|
|
|
|
| 1 |
+
# 🤖 Wolof GIT Processor (Image + Text)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
|
| 3 |
Ce processor combine :
|
| 4 |
|
| 5 |
+
- Le tokenizer spécialisé en wolof [`ZigZeug/wolof-tokenizer-v1`](https://huggingface.co/ZigZeug/wolof-tokenizer-v1)
|
| 6 |
+
- Le processor image du modèle [`microsoft/git-base`](https://huggingface.co/microsoft/git-base)
|
| 7 |
|
| 8 |
+
Il est conçu pour être utilisé dans des tâches de **vision-to-text**, comme le **captioning d'image en wolof**, le **fine-tuning multimodal**, ou des applications éducatives et culturelles dans les langues africaines.
|
| 9 |
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
## 💡 Exemple d'utilisation
|
| 13 |
|
| 14 |
```python
|
| 15 |
from transformers import AutoProcessor
|
| 16 |
+
from PIL import Image
|
| 17 |
|
| 18 |
processor = AutoProcessor.from_pretrained("ZigZeug/wolof-git-processor")
|
| 19 |
|
| 20 |
+
image = Image.open("chemin/vers/image.jpg")
|
| 21 |
+
text = "Jàngalekat bi dafay def ay jagleel ci klas bi."
|
| 22 |
+
|
| 23 |
inputs = processor(
|
| 24 |
+
images=image,
|
| 25 |
+
text=text,
|
| 26 |
return_tensors="pt"
|
| 27 |
)
|