Update app.py

app.py

@@ -46,7 +46,7 @@ model_id = "CardinalOperations/ORLM-LLaMA-3-8B"
 # model.eval()
 
 subprocess.run(f'huggingface-cli download {model_id} --local-dir ./local_model',  shell=True)
-model = LLM(model='./local_model', tensor_parallel_size=torch.cuda.device_count())
+model = LLM(model='./local_model', tensor_parallel_size=1)
 print("init model done.")
 
 @spaces.GPU(duration=60)