Update app.py

app.py

@@ -180,73 +180,64 @@ def test():
 def call_test():
     return {"message": test()}
 
-with gr.Blocks(css=CSS, theme="soft", fill_height=True) as demo:
-    gr.HTML(TITLE)
-    gr.HTML(DESCRIPTION)
-    gr.ChatInterface(
-        fn=stream_chat,
-        multimodal=True,
-        textbox=chat_input,
-        chatbot=chatbot,
-        fill_height=True,
-        additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False, render=False),
-        additional_inputs=[
-            gr.Slider(
-                minimum=0,
-                maximum=1,
-                step=0.1,
-                value=0.8,
-                label="Temperature",
-                render=False,
-            ),
-            gr.Slider(
-                minimum=1024,
-                maximum=8192,
-                step=1,
-                value=4096,
-                label="Max Length",
-                render=False,
-            ),
-            gr.Slider(
-                minimum=0.0,
-                maximum=1.0,
-                step=0.1,
-                value=1.0,
-                label="top_p",
-                render=False,
-            ),
-            gr.Slider(
-                minimum=1,
-                maximum=20,
-                step=1,
-                value=10,
-                label="top_k",
-                render=False,
-            ),
-            gr.Slider(
-                minimum=0.0,
-                maximum=2.0,
-                step=0.1,
-                value=1.0,
-                label="Repetition penalty",
-                render=False,
-            ),
-        ],
-    ),
-    gr.Examples(EXAMPLES, [chat_input])
+def run_gradio():
+    with gr.Blocks(css=CSS, theme="soft", fill_height=True) as demo:
+        gr.HTML(TITLE)
+        gr.HTML(DESCRIPTION)
+        gr.ChatInterface(
+            fn=stream_chat,
+            multimodal=True,
+            textbox=chat_input,
+            chatbot=chatbot,
+            fill_height=True,
+            additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False, render=False),
+            additional_inputs=[
+                gr.Slider(
+                    minimum=0,
+                    maximum=1,
+                    step=0.1,
+                    value=0.8,
+                    label="Temperature",
+                    render=False,
+                ),
+                gr.Slider(
+                    minimum=1024,
+                    maximum=8192,
+                    step=1,
+                    value=4096,
+                    label="Max Length",
+                    render=False,
+                ),
+                gr.Slider(
+                    minimum=0.0,
+                    maximum=1.0,
+                    step=0.1,
+                    value=1.0,
+                    label="top_p",
+                    render=False,
+                ),
+                gr.Slider(
+                    minimum=1,
+                    maximum=20,
+                    step=1,
+                    value=10,
+                    label="top_k",
+                    render=False,
+                ),
+                gr.Slider(
+                    minimum=0.0,
+                    maximum=2.0,
+                    step=0.1,
+                    value=1.0,
+                    label="Repetition penalty",
+                    render=False,
+                ),
+            ],
+        ),
+        gr.Examples(EXAMPLES, [chat_input])
+    demo.queue(api_open=False).launch(show_api=False, share=False)
 
 if __name__ == "__main__":
-    def run_fastapi():
-        uvicorn.run(app, host="0.0.0.0", port=8000)
-
-    def run_gradio():
-        demo.queue(api_open=False).launch(show_api=False, share=False)
-
-    fastapi_thread = Thread(target=run_fastapi)
-    fastapi_thread.start()
-
     gradio_thread = Thread(target=run_gradio)
     gradio_thread.start()
-
-    fastapi_thread.join()
-    gradio_thread.join()
+    uvicorn.run(app, host="0.0.0.0", port=8000)