Update app.py

app.py

@@ -325,7 +325,6 @@ Grounded in auditory neuroscience (Norman-Haignere 2025, Vollan 2025). Zero embe
                     label="",
                     height=480,
                     show_label=False,
-                    type="tuples",
                 )
 
                 retrieval_info = gr.Markdown("", elem_id="retrieval-info")
@@ -443,14 +442,16 @@ Commands: type `:coverage` to see sweep focus • `:summary` for index stats •
                 return
 
             if state.retriever is None:
-                chat_history = chat_history + [(msg, "⚠️ Please load a document first.")]
+                chat_history.append({"role": "user", "content": msg})
+                chat_history.append({"role": "assistant", "content": "⚠️ Please load a document first."})
                 yield chat_history, ""
                 return
 
             # Handle commands
             cmd_response, is_cmd = handle_command(msg)
             if is_cmd:
-                chat_history = chat_history + [(msg, cmd_response)]
+                chat_history.append({"role": "user", "content": msg})
+                chat_history.append({"role": "assistant", "content": cmd_response})
                 yield chat_history, ""
                 return
 
@@ -481,10 +482,14 @@ Commands: type `:coverage` to see sweep focus • `:summary` for index stats •
             # Stream response
             client = get_client(hf_token)
             partial = ""
-            chat_history = chat_history + [(msg, "")]
+            
+            # Gradio 6 messages format
+            chat_history.append({"role": "user", "content": msg})
+            chat_history.append({"role": "assistant", "content": ""})
+            
             for token in stream_response(client, model, messages):
                 partial += token
-                chat_history[-1] = (msg, partial)
+                chat_history[-1]["content"] = partial
                 yield chat_history, retrieval_display
 
             # Save to history