Update app.py

app.py

@@ -30,6 +30,7 @@ def find_attached_file(filename, attached_files):
 async def echo(message, history, state, persona, use_generated_summaries):
     attached_file = None
     system_instruction = Template(prompt_tmpl['summarization']['system_prompt']).safe_substitute(persona=persona)
+    system_instruction_cutoff = Template(prompt_tmpl['summarization']['system_prompt_cutoff'])
     use_generated_summaries = True if use_generated_summaries == "Yes" else False
 
     if message['files']:
@@ -51,28 +52,52 @@ async def echo(message, history, state, persona, use_generated_summaries):
                 "expiration_time": path_gcp.expiration_time,
             })
             attached_file = path_wrap
-
-    user_message_parts = [types.Part.from_text(text=message['text'])]
-    if attached_file: user_message_parts.append(attached_file)
-    user_message = [types.Content(role='user', parts=user_message_parts)]
-    state['messages'] = state['messages'] + user_message
-
+            
     response_chunks = ""
     model_contents = ""
     if use_generated_summaries:
         if "summary_history" in state and len(state["summary_history"]):
-            model_contents += state["summary_history"][-1]
+            user_message_parts = [
+                types.Part.from_text(text=f"""Summary\n:{state["summary_history"][-1]}\n-------"""),
+                types.Part.from_text(text=message['text'])
+            ]
+            if attached_file: user_message_parts.append(attached_file)
+            model_contents = [types.Content(role='user', parts=user_message_parts)]
+
+            model_content_stream = await client.models.generate_content_stream(
+                model=args.model, 
+                contents=model_contents, 
+                config=types.GenerateContentConfig(
+                    system_instruction=system_instruction_cutoff, seed=args.seed
+                ),
+            )                
         else:
+            user_message_parts = [types.Part.from_text(text=message['text'])]
+            if attached_file: user_message_parts.append(attached_file)
+            user_message = [types.Content(role='user', parts=user_message_parts)]
+            
             model_contents = state['messages']
+            state['messages'] = model_contents + user_message
+
+            model_content_stream = await client.models.generate_content_stream(
+                model=args.model, 
+                contents=model_contents, 
+                config=types.GenerateContentConfig(seed=args.seed),
+            )    
     else:
+        user_message_parts = [types.Part.from_text(text=message['text'])]
+        if attached_file: user_message_parts.append(attached_file)
+        user_message = [types.Content(role='user', parts=user_message_parts)]
+        
         model_contents = state['messages']
-    model_content_stream = await client.models.generate_content_stream(
-        model=args.model, 
-        contents=model_contents, 
-        config=types.GenerateContentConfig(
-            system_instruction=system_instruction, seed=args.seed
-        ),
-    )
+        state['messages'] = model_contents + user_message
+
+        model_content_stream = await client.models.generate_content_stream(
+            model=args.model, 
+            contents=model_contents, 
+            config=types.GenerateContentConfig(seed=args.seed),
+        )    
+
     async for chunk in model_content_stream:
         response_chunks += chunk.text
         # when model generates too fast, Gradio does not respond that in real-time.