Update app.py

app.py

@@ -74,7 +74,7 @@ def query_model(model_name: str, messages: List[Dict[str, str]]) -> str:
     except Exception as e:
         return f"{model_name} error: {str(e)}"
 
-def respond(message: str, history: List[List[str]], session_id: str) -> Generator[tuple[str, str], None, None]:
+def respond(message: str, history: List[List[str]], session_id: str) -> str:
     """Handle sequential model responses with context preservation"""
     # Load or initialize session
     session = session_manager.load_session(session_id)
@@ -97,8 +97,9 @@ def respond(message: str, history: List[List[str]], session_id: str) -> Generato
         "content": message
     })
     
+    responses = []
+    
     # First model response
-    yield message, "💭 *Qwen2.5-Coder-32B-Instruct is thinking...*"
     response1 = query_model("Qwen2.5-Coder-32B-Instruct", messages)
     session["history"].append({
         "timestamp": datetime.now().isoformat(),
@@ -107,10 +108,9 @@ def respond(message: str, history: List[List[str]], session_id: str) -> Generato
         "content": response1
     })
     messages.append({"role": "assistant", "content": f"Qwen2.5-Coder-32B-Instruct: {response1}"})
-    yield message, f"🔵 **Qwen2.5-Coder-32B-Instruct**\n{response1}"
+    responses.append(f"🔵 **Qwen2.5-Coder-32B-Instruct**\n{response1}")
     
     # Second model response
-    yield message, f"🔵 **Qwen2.5-Coder-32B-Instruct**\n{response1}\n\n💭 *Qwen2.5-72B-Instruct is thinking...*"
     response2 = query_model("Qwen2.5-72B-Instruct", messages)
     session["history"].append({
         "timestamp": datetime.now().isoformat(),
@@ -119,10 +119,9 @@ def respond(message: str, history: List[List[str]], session_id: str) -> Generato
         "content": response2
     })
     messages.append({"role": "assistant", "content": f"Qwen2.5-72B-Instruct: {response2}"})
-    yield message, f"🔵 **Qwen2.5-Coder-32B-Instruct**\n{response1}\n\n🟣 **Qwen2.5-72B-Instruct**\n{response2}"
+    responses.append(f"🟣 **Qwen2.5-72B-Instruct**\n{response2}")
     
     # Final model response
-    yield message, f"🔵 **Qwen2.5-Coder-32B-Instruct**\n{response1}\n\n🟣 **Qwen2.5-72B-Instruct**\n{response2}\n\n💭 *Llama3.3-70B-Instruct is thinking...*"
     response3 = query_model("Llama3.3-70B-Instruct", messages)
     session["history"].append({
         "timestamp": datetime.now().isoformat(),
@@ -131,12 +130,13 @@ def respond(message: str, history: List[List[str]], session_id: str) -> Generato
         "content": response3
     })
     messages.append({"role": "assistant", "content": f"Llama3.3-70B-Instruct: {response3}"})
+    responses.append(f"🟡 **Llama3.3-70B-Instruct**\n{response3}")
     
     # Save final session state
     session_manager.save_session(session_id, session)
     
-    # Return final combined response
-    yield message, f"🔵 **Qwen2.5-Coder-32B-Instruct**\n{response1}\n\n🟣 **Qwen2.5-72B-Instruct**\n{response2}\n\n🟡 **Llama3.3-70B-Instruct**\n{response3}"
+    # Return responses
+    return "\n\n".join(responses)
 
 # Custom CSS for styling
 css = """
@@ -160,11 +160,11 @@ demo = gr.ChatInterface(
     chatbot=gr.Chatbot(
         height=600,
         show_label=False,
-        avatar_images=("👤", "🤖"),
         bubble_full_width=False,
         show_copy_button=True,
         container=True,
         sanitize_html=False,
+        render_markdown=True
     ),
     theme=gr.themes.Soft(
         primary_hue="blue",