Update app.py

app.py

@@ -35,8 +35,7 @@ def generate(prompt, history, system_prompt, temperature=0.9, max_new_tokens=904
     output = ""
     for response in stream:
         output += response.token.text
-    yield output
-    return output
+    return [(prompt, output)]
 
 additional_inputs = [
     gr.Textbox(label="System Prompt", max_lines=1, interactive=True),
@@ -53,7 +52,8 @@ with app:
     submit_btn = gr.Button("Submit")
 
     def process_message(message, history):
-        history.append((message, generate(message, history, additional_inputs[0].value)))
+        response = generate(message, history, additional_inputs[0].value)
+        history.extend(response)
         return history, ""
 
     submit_btn.click(process_message, inputs=[text_input, chatbot], outputs=[chatbot, text_input])