Update app.py

app.py

@@ -411,7 +411,10 @@ def clean_response(response_text):
     cleaned_response = re.sub(r'page_content=".*?"\),?', '', cleaned_response, flags=re.DOTALL)
     cleaned_response = re.sub(r'\[.*?\]', '', cleaned_response, flags=re.DOTALL)  # Remove content in brackets
     cleaned_response = re.sub(r'\s+', ' ', cleaned_response).strip()
+    #Remove any unwanted follow-up questions or unnecessary text
+    cleaned_response = re.sub(r'Question:.*\nAnswer:', '', response_text, flags=re.DOTALL).strip()
     return cleaned_response
+    
 
 def generate_answer(message, choice, retrieval_mode, selected_model):
     logging.debug(f"generate_answer called with choice: {choice} and retrieval_mode: {retrieval_mode}")
@@ -453,7 +456,7 @@ def generate_answer(message, choice, retrieval_mode, selected_model):
                 logging.debug(f"Phi-3.5 Prompt: {prompt}")
 
                 response = selected_model(prompt, **{
-                    "max_new_tokens": 512,  # Increased to handle longer responses
+                    "max_new_tokens": 128,  # Increased to handle longer responses
                     "return_full_text": False,
                     "temperature": 0.7,  # Adjusted to avoid cutting off
                     "do_sample": True,   # Allow sampling to increase response diversity