Update app.py

app.py

@@ -9,6 +9,7 @@ Original file is located at
 
 import torch
 import gradio as gr
+import spaces
 from threading import Thread
 from unsloth import FastLanguageModel
 from transformers import TextStreamer
@@ -46,7 +47,7 @@ FastLanguageModel.for_inference(model)
 text_streamer = TextStreamer(tokenizer, skip_prompt=False,skip_special_tokens=True)
 
 # Get the device based on GPU availability
-device = 'cuda' if torch.cuda.is_available() else 'cpu'
+device = 'cuda'
 
 # Move model into device
 model = model.to(device)
@@ -61,6 +62,7 @@ class StopOnTokens(StoppingCriteria):
 
 # Current implementation does not support conversation based on history.
 # Highly recommend to experiment on various hyper parameters to compare qualities.
+@spaces.GPU(duration=gpu_timeout)
 def predict(message, history):
     stop = StopOnTokens()
     messages = alpaca_prompt.format(