Update app.py

app.py

@@ -1,53 +1,109 @@
 import gradio as gr
-from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
+import joblib
 import os
+import numpy as np
+
 
 
 def load_model():
-    """Load model from root directory"""
-    model_path = "."
-    
+    """Load joblib model and components"""
     try:
-        print(f"Loading model from root directory...")
-        tokenizer = AutoTokenizer.from_pretrained(model_path)
-        model = AutoModelForSequenceClassification.from_pretrained(model_path)
-        classifier = pipeline(
-            "text-classification",
-            model=model,
-            tokenizer=tokenizer,
-            return_all_scores=True
-        )
-        print(f"✅ Successfully loaded model")
-        return classifier
+        print("Loading joblib model...")
+        
+        # Load model (try different possible names)
+        model = None
+        model_file = None
+        
+        if os.path.exists("model.safetensors"):
+            print("Found model.safetensors")
+            model = joblib.load("model.safetensors")
+            model_file = "model.safetensors"
+        elif os.path.exists("model.jobilib"):
+            print("Found model.jobilib")
+            model = joblib.load("model.jobilib")
+            model_file = "model.jobilib"
+        elif os.path.exists("tfidf_logreg_best.jobilib"):
+            print("Found tfidf_logreg_best.jobilib")
+            model = joblib.load("tfidf_logreg_best.jobilib")
+            model_file = "tfidf_logreg_best.jobilib"
+        else:
+            # List available files
+            files = os.listdir(".")
+            print(f"Available files: {files}")
+            raise FileNotFoundError("No model file found")
+        
+        # Load vectorizer/tokenizer
+        vectorizer = None
+        if os.path.exists("vocab"):
+            print("Found vocab file")
+            vectorizer = joblib.load("vocab")
+        elif os.path.exists("tokenizer"):
+            print("Found tokenizer file")
+            vectorizer = joblib.load("tokenizer")
+        
+        # Load label encoder
+        label_encoder = None
+        if os.path.exists("label_encoder.jobilib"):
+            print("Found label_encoder.jobilib")
+            label_encoder = joblib.load("label_encoder.jobilib")
+        
+        print(f"✅ Model loaded successfully from {model_file}")
+        return {
+            "model": model,
+            "vectorizer": vectorizer,
+            "label_encoder": label_encoder
+        }
+        
     except Exception as e:
-        print(f"❌ Error loading model: {str(e)}")
-        classifier = pipeline(
-            "text-classification",
-            model="bert-base-uncased",
-            return_all_scores=True
-        )
-        return classifier
+        print(f"❌ Error loading joblib model: {str(e)}")
+        return None
+
+# Load model
+model_components = load_model()
 
-classifier = load_model()
 
-# ============================================================
-# PREDICTION LOGIC WITH CORRECT LABELS
-# ============================================================
 
 def predict(text):
-    """Predict cyberbullying category"""
+    """Predict cyberbullying category using joblib model"""
     if not text.strip():
         return "<div class='warn'>⚠️ Please enter some text.</div>"
     
     try:
-        results = classifier(text)[0]
-        best = max(results, key=lambda x: x["score"])
-        label = best["label"]
-        score = best["score"]
+        if model_components is None:
+            return "<div class='warn'>❌ Model not loaded properly</div>"
+        
+        model = model_components["model"]
+        vectorizer = model_components["vectorizer"]
+        label_encoder = model_components["label_encoder"]
+        
+        # Vectorize the text
+        if vectorizer is not None:
+            text_vector = vectorizer.transform([text])
+        else:
+            return "<div class='warn'>❌ Vectorizer not found</div>"
+        
+        # Get prediction
+        prediction = model.predict(text_vector)[0]
         
-        print(f"Label: {label}, Score: {score}")
+        # Get probability if available
+        try:
+            probabilities = model.predict_proba(text_vector)[0]
+            score = max(probabilities)
+        except:
+            score = 0.8  # Default score
         
-        # Your model labels
+        # Decode label if encoder exists
+        if label_encoder is not None:
+            try:
+                label = label_encoder.inverse_transform([prediction])[0]
+            except:
+                label = str(prediction)
+        else:
+            label = str(prediction)
+        
+        print(f"Prediction: {label}, Score: {score}")
+        
+        # Category definitions
         cyberbullying_types = {
             "age": {"emoji": "👶", "color": "#ff6b6b", "text": "Age-Based Cyberbullying"},
             "gender": {"emoji": "⚥️", "color": "#ff8c42", "text": "Gender-Based Cyberbullying"},
@@ -57,11 +113,12 @@ def predict(text):
             "not_cyberbullying": {"emoji": "✅", "color": "#00ff64", "text": "Safe Message"}
         }
         
-        # Get the category info
-        category = cyberbullying_types.get(label.lower(), cyberbullying_types["not_cyberbullying"])
+        # Get category (handle case variations)
+        label_lower = str(label).lower().strip()
+        category = cyberbullying_types.get(label_lower, cyberbullying_types.get(label, cyberbullying_types["not_cyberbullying"]))
         
         # Safe message
-        if label.lower() == "not_cyberbullying":
+        if label_lower == "not_cyberbullying":
             return f"""
             <div class='safe'>
                 <div class='checkmark'>{category['emoji']}</div>
@@ -88,13 +145,12 @@ def predict(text):
             
     except Exception as e:
         import traceback
+        error_msg = traceback.format_exc()
         print(f"ERROR: {str(e)}")
-        print(traceback.format_exc())
+        print(error_msg)
         return f"<div class='warn'>❌ Error: {str(e)}</div>"
 
-# ============================================================
-# GRADIO INTERFACE WITH PURPLE-TO-BLUE GRADIENT
-# ============================================================
+
 
 with gr.Blocks(theme=gr.themes.Soft(), css="""
 <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/animate.css/4.1.1/animate.min.css"/>
@@ -456,6 +512,6 @@ textarea:focus {
         outputs=output
     )
 
-# Launch the app
+
 if __name__ == "__main__":
     demo.launch()