Update app.py

app.py

@@ -72,45 +72,28 @@ def query_model(model_name: str, messages: List[Dict[str, str]]) -> str:
         return f"{model_name} error: {str(e)}"
 
 def respond(message: str, history: List[List[str]]) -> str:
-    """Handle sequential model responses with continuous context"""
+    """Handle continuous contextual conversations"""
+    # Build full message history from previous interactions
+    messages = []
+    for user_msg, bot_msg in history:
+        messages.append({"role": "user", "content": user_msg})
+        messages.append({"role": "assistant", "content": bot_msg})
     
-    # Build full conversation history from previous interactions
-    conversation = []
-    if history:
-        for user_msg, assistant_msg in history:
-            conversation.append({"role": "user", "content": user_msg})
-            if assistant_msg:
-                # Split assistant message into individual model responses
-                responses = assistant_msg.split("\n\n")
-                for resp in responses:
-                    if resp:
-                        conversation.append({"role": "assistant", "content": resp})
+    # Add new user message
+    messages.append({"role": "user", "content": message})
     
-    # Add current message
-    conversation.append({"role": "user", "content": message})
-    
-    # Get first model's response
-    response1 = query_model("Qwen2.5-Coder-32B-Instruct", conversation)
+    # First model sees current prompt + full history
+    response1 = query_model("Qwen2.5-Coder-32B-Instruct", messages)
     yield f"**Qwen2.5-Coder-32B-Instruct**:\n{response1}"
     
-    # Add first response to context
-    conversation.append({
-        "role": "assistant",
-        "content": f"**Qwen2.5-Coder-32B-Instruct**:\n{response1}"
-    })
-    
-    # Get second model's response
-    response2 = query_model("Qwen2.5-72B-Instruct", conversation)
+    # Second model sees current prompt + history + first response
+    messages.append({"role": "assistant", "content": response1})
+    response2 = query_model("Qwen2.5-72B-Instruct", messages)
     yield f"**Qwen2.5-72B-Instruct**:\n{response2}"
     
-    # Add second response to context
-    conversation.append({
-        "role": "assistant",
-        "content": f"**Qwen2.5-72B-Instruct**:\n{response2}"
-    })
-    
-    # Get final model's response
-    response3 = query_model("Llama3.3-70B-Instruct", conversation)
+    # Third model sees current prompt + history + both responses
+    messages.append({"role": "assistant", "content": response2})
+    response3 = query_model("Llama3.3-70B-Instruct", messages)
     yield f"**Llama3.3-70B-Instruct**:\n{response3}"
 
 # Create the Gradio interface