Update app.py

app.py

@@ -1,6 +1,7 @@
+import spaces
 import os
 import torch
-from transformers import AutoTokenizer, AutoModelForCausalLM
+from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
 import gradio as gr
 from threading import Thread
 
@@ -32,6 +33,7 @@ tokenizer = AutoTokenizer.from_pretrained(MODEL, trust_remote_code=True)
 model = AutoModelForCausalLM.from_pretrained(MODEL, torch_dtype=torch.bfloat16, trust_remote_code=True, device_map="auto")
 model = model.eval()
 
+@spaces.GPU
 def stream_chat(
     message: str, 
     history: list,
@@ -44,12 +46,10 @@ def stream_chat(
     print(f'message: {message}')
     print(f'history: {history}')
 
-    # Prepare the conversation history
     chat_history = []
     for prompt, answer in history:
         chat_history.append((prompt, answer))
 
-    # Generate the response
     for response, _ in model.stream_chat(
         tokenizer,
         message,
@@ -70,11 +70,12 @@ with gr.Blocks(css=CSS, theme="soft") as demo:
         fn=stream_chat,
         chatbot=chatbot,
         fill_height=True,
-        additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False),
+        additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False, render=False),
         additional_inputs=[
             gr.Textbox(
                 value="You are a helpful assistant capable of generating long-form content.",
                 label="System Prompt",
+                render=False,
             ),
             gr.Slider(
                 minimum=0,
@@ -82,6 +83,7 @@ with gr.Blocks(css=CSS, theme="soft") as demo:
                 step=0.1,
                 value=0.5,
                 label="Temperature",
+                render=False,
             ),
             gr.Slider(
                 minimum=1024,
@@ -89,6 +91,7 @@ with gr.Blocks(css=CSS, theme="soft") as demo:
                 step=1024,
                 value=32768,
                 label="Max new tokens",
+                render=False,
             ),
             gr.Slider(
                 minimum=0.0,
@@ -96,6 +99,7 @@ with gr.Blocks(css=CSS, theme="soft") as demo:
                 step=0.1,
                 value=1.0,
                 label="Top p",
+                render=False,
             ),
             gr.Slider(
                 minimum=1,
@@ -103,6 +107,7 @@ with gr.Blocks(css=CSS, theme="soft") as demo:
                 step=1,
                 value=50,
                 label="Top k",
+                render=False,
             ),
         ],
         examples=[