Update README.md

README.md

@@ -2,34 +2,66 @@
 language: ar
 base_model: faisalq/SaudiBERT
 tags:
-- eou
-- turn-taking
 - arabic
 - saudi
+- eou
+- turn-taking
+- conversational-ai
+license: mit
 ---
 
 # Saudi Arabic End-of-Utterance (EOU) Model
 
-This is a fine-tuned **SaudiBERT** model for **End-of-Utterance (EOU) detection** in Saudi Arabic conversational text.
+This model detects **End-of-Utterance (EOU)** events in **Saudi Arabic conversational text**.  
+It outputs the probability that a speaker has **finished their turn**, enabling natural turn-taking in real-time voice agents (e.g., LiveKit).
+
+---
 
 ## Task
-Binary classification:
-- 0 → Incomplete utterance
-- 1 → End of utterance
+Binary classification (probability output):
+
+- **0** → Incomplete utterance (speaker likely to continue)
+- **1** → Complete utterance (end of turn)
+
+---
+
+## Model Details
+- **Base model:** `faisalq/SaudiBERT`
+- **Architecture:** BERT Sequence Classification
+- **Output:** Single probability (sigmoid)
+- **Dialect focus:** Saudi Arabic (ar-SA)
+
+---
 
 ## Training
-- Base model: faisalq/SaudiBERT
-- Data: Saudi Arabic conversational dataset
-- Loss: Focal Loss
-- Metric: F1-score
+- **Dataset:** Saudi Arabic conversational EOU dataset  
+  https://huggingface.co/datasets/HussainKAUST/saudi-eou-dataset
+- **Data source:** Synthetic Saudi dialogue with natural pauses and incomplete turns
+- **Loss:** Focal Loss (class imbalance handling)
+- **Epochs:** 6
+
+---
+
+## Evaluation Results
+- **Validation F1:** ~0.83  
+- **Test F1:** ~0.75  
+- **Test Accuracy:** ~0.81
+
+---
+
+## Usage Example
 
-## Usage
 ```python
 from transformers import AutoTokenizer, AutoModelForSequenceClassification
 import torch
 
-tok = AutoTokenizer.from_pretrained("HussainKAUST/saudi-eou-model")
-mdl = AutoModelForSequenceClassification.from_pretrained("HussainKAUST/saudi-eou-model")
+tokenizer = AutoTokenizer.from_pretrained("HussainKAUST/saudi-eou-model")
+model = AutoModelForSequenceClassification.from_pretrained("HussainKAUST/saudi-eou-model")
+
+text = "ابي احجز موعد بس ..."
+inputs = tokenizer(text, return_tensors="pt")
+
+with torch.no_grad():
+    prob = torch.sigmoid(model(**inputs).logits).item()
 
-x = tok("ابي احجز موعد بس ...", return_tensors="pt")
-p = torch.sigmoid(mdl(**x).logits).item()
+print("EOU probability:", prob)