Update app.py

app.py

@@ -35,7 +35,7 @@ def generate(prompt, history, system_prompt, temperature=0.9, max_new_tokens=904
     output = ""
     for response in stream:
         output += response.token.text
-    return [(prompt, output)]
+        yield (prompt, output)
 
 additional_inputs = [
     gr.Textbox(label="System Prompt", max_lines=1, interactive=True),
@@ -49,13 +49,10 @@ app = gr.Blocks(theme=gr.themes.Soft())
 with app:
     chatbot = gr.Chatbot()
     text_input = gr.Textbox(label="Your message")
-    submit_btn = gr.Button("Submit")
 
     def process_message(message, history):
-        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])
+        for response in generate(message, history, additional_inputs[0].value):
+            yield response
 
+    text_input.submit(process_message, inputs=[text_input, chatbot], outputs=[chatbot, text_input])
     app.launch(show_api=False)