nvm going back

app.py

@@ -31,27 +31,6 @@ def load_model(model_name):
         print(f"Error loading model {model_name}: {e}")
         return None, None
 
-# def get_sentiment_prediction(text, model, tokenizer):
-#     """Get sentiment prediction from model"""
-#     if model is None:
-#         # Fallback to dummy predictions for demo
-#         return {
-#             "label": "NM",
-#             "probabilities": {"Negative": 0.01, "Neutral": 0.01, "Positive": 0.01}
-#         }
-    
-#     try:
-#         # Build full prompt for analysis
-#         prefix = "Analyze the sentiment of this statement extracted from a financial news article. Provide your answer as either negative, positive, or neutral.. Text: "
-#         suffix = ".. Answer: "
-#         full_prompt = f"{prefix}{text}{suffix}"
-#         # Added a small comment here.
-#         result = model.generate(prompt=text)
-#         return result
-#     except Exception as e:
-#         print(f"Error in prediction: {e}")
-#         return {"label": "NA", "probabilities": {"Negative": 0.0, "Neutral": 0.0, "Positive": 0.0}}
-
 def get_sentiment_prediction(text, model, tokenizer):
     """Get sentiment prediction from model"""
     if model is None:
@@ -62,23 +41,12 @@ def get_sentiment_prediction(text, model, tokenizer):
         }
     
     try:
-        # Print model's label mapping for debugging
-        if hasattr(model, 'id2label'):
-            print(f"[DEBUG] Model id2label mapping: {model.id2label}")
-        
-        result = model.generate(prompt=text)
-        
-        # Print raw prediction
-        print(f"[DEBUG] Text (first 50 chars): {text[:50]}")
-        print(f"[DEBUG] Raw prediction: {result}")
-        
-        # Normalize label to title case to match your offline results
-        if 'label' in result:
-            original_label = result['label']
-            normalized_label = original_label.capitalize()
-            result['label'] = normalized_label
-            print(f"[DEBUG] Normalized label: {original_label} → {normalized_label}")
-        
+        # Build full prompt for analysis
+        prefix = "Analyze the sentiment of this statement extracted from a financial news article. Provide your answer as either negative, positive, or neutral.. Text: "
+        suffix = ".. Answer: "
+        full_prompt = f"{prefix}{text}{suffix}"
+        # Added a small comment here.
+        result = model.generate(prompt=full_prompt)
         return result
     except Exception as e:
         print(f"Error in prediction: {e}")