Update app.py

app.py

@@ -76,30 +76,23 @@ class BasicAgent:
 
 
     def __call__(self, question: str) -> str:
-        
-      try:
+    try:
         print(f"\n➡️ Agent received question: {question[:80]}")
 
-        if getattr(self, "model_type", None) == "local":
-            # Use local model
-            messages = [{"role": "user", "content": question}]
-            response = self.pipe(messages, max_new_tokens=200)
-            answer = response[0]["generated_text"]
-        else:
-            # Use Hugging Face router (OpenAI-style)
-            response = self.client.chat.completions.create(
-                model=self.model_id,
-                messages=[
-                    {"role": "system", "content": "You are a helpful assistant."},
-                    {"role": "user", "content": question}
-                ],
-            )
-            answer = response.choices[0].message["content"].strip()
+        # Use local model
+        response = self.pipe(question, max_new_tokens=150)
+        raw_answer = response[0]["generated_text"]
+
+        # Clean output: remove question and artifacts
+        answer = raw_answer.replace(question, "").strip()
+
+        # Guarantee a valid string (the API rejects non-strings)
+        answer = str(answer)
 
         print(f"✅ Agent returning answer: {answer[:100]}...")
         return answer
 
-      except Exception as e:
+    except Exception as e:
         print(f"❌ Agent encountered an error: {e}")
         traceback.print_exc()
         return f"Error generating answer: {e}"
@@ -107,6 +100,7 @@ class BasicAgent:
 
 
 
+
 def run_and_submit_all( profile: gr.OAuthProfile | None):
     """
     Fetches all questions, runs the BasicAgent on them, submits all answers,