Update handler.py

handler.py

@@ -43,31 +43,20 @@ class EndpointHandler:
             return self.tokenizer.decode(tokens)
         return text
     
+    
     def __call__(self, data: Dict[str, Any]) -> Dict[str, Any]:
         user_prompt = data.pop("inputs", data)
         
-        # Add the user's message to the conversation history
-        self.conversation_history += f"<user>: {user_prompt}\n"
-        
-        # Ensure the conversation history is within token limit
-        self.conversation_history = self._ensure_token_limit(self.conversation_history)
-
-        # Add the permanent context, user's prompt, and conversation history
-        permanent_context = "<context>: You are a life coaching bot with the goal of providing guidance, improving understanding, reducing suffering and improving life. Gain as much understanding of the user before providing guidance."
-        structured_prompt = f"{permanent_context}\n{self.conversation_history}<bot> response:"
+        # Permanent context
+        permanent_context = "<context>: You are a life coaching bot..."
+        structured_prompt = f"{permanent_context}\<bot> response:"
         
         result = self.pipeline(structured_prompt, generation_config=self.generate_config)
         
-        # Extract only the bot's response without the structuring text
+        # Ensure _extract_response is defined and works as intended
         response_text = self._extract_response(result[0]['generated_text'])
         
-        # Remove the last "<bot>" from the response_text
+        # Trimming response
         response_text = response_text.rsplit("[END", 1)[0].strip()
-        
-        # Add the bot's response to the conversation history
-        self.conversation_history += f"<bot>: {response_text}\n"
-        self.conversation_history = self._ensure_token_limit(self.conversation_history)
     
-        return [{"generated_text": response_text}]
-        
         return {"response": response_text}