Update app.py

app.py

@@ -11,6 +11,7 @@ model_name = "bloom-560m"
 model = AutoModelForCausalLM.from_pretrained(f'jslin09/{model_name}-finetuned-fraud').to('cuda')
 tokenizer = BloomTokenizerFast.from_pretrained(f'bigscience/{model_name}', bos_token = '<s>', eos_token = '</s>', pad_token = '<pad>')
 
+@spaces.GPU
 def rnd_generate(prompt):
     rnd_seed = random.randint(10, 500)
     set_seed(rnd_seed)
@@ -24,6 +25,7 @@ def rnd_generate(prompt):
                        top_p=0.9)
     return tokenizer.decode(results[0])
 
+@spaces.GPU
 def generate(prompt):
     result_length = len(prompt) + 4
     inputs = tokenizer(prompt, return_tensors="pt") # 回傳的張量使用 Pytorch的格式。如果是 Tensorflow 格式的話，則指定為 "tf"。