Update README.md

README.md

@@ -1,4 +1,61 @@
 ---
-library_name: transformers
+language:
+- ar
+- mey
+tags:
+- punctuation-restoration
+- token-classification
+- marbert
+- mauritania
+- dialect
 license: apache-2.0
----
+datasets:
+- hassaniya-punctuation-dataset
+---
+
+# Hassaniya Punctuation Restoration Model (الحسانية)
+
+This is the first open-source model dedicated to restoring punctuation for the **Hassaniya Arabic dialect** (spoken in Mauritania, Southern Morocco, etc.). 
+
+It restores the following punctuation marks:
+- Period (.)
+- Comma (،)
+- Question Mark (؟)
+- Exclamation Mark (!)
+- Colon (:)
+
+## Model Details
+- **Base Model:** `UBC-NLP/MARBERTv2` (State-of-the-art for Arabic Dialects).
+- **Training Data:** ~10,000 sentences of Hassaniya dialogue and narrative text.
+- **Task:** Token Classification (The model tags each word with the punctuation that should follow it).
+
+## How to use
+
+You can use this model directly with the Hugging Face `pipeline`:
+
+```python
+from transformers import pipeline
+
+# Load model
+model_name = "Emin009/hassaniya-punctuation-restoration"
+punct_pipe = pipeline("token-classification", model=model_name, aggregation_strategy="simple")
+
+# Define punctuation map
+punct_map = {"PERIOD": ".", "COMMA": "،", "QUESTION": "؟", "EXCLAM": "!", "COLON": ":"}
+
+def restore_punct(text):
+    results = punct_pipe(text)
+    output_text = ""
+    last_idx = 0
+    for res in results:
+        output_text += text[last_idx:res['end']]
+        if res['entity_group'] in punct_map:
+            output_text += punct_map[res['entity_group']]
+        last_idx = res['end']
+    output_text += text[last_idx:]
+    return output_text
+
+# Test
+text = "منهو هذا الراجل جا منين"
+print(restore_punct(text))
+# Output: منهو هذا الراجل؟ جا منين؟