Update app.py

app.py

@@ -38,8 +38,9 @@ def init_model():
     global lm_generation
     torch.set_default_tensor_type(torch.HalfTensor)
     model = LLaMa(args)
-    # torch.set_default_tensor_type(torch.FloatTensor)
+    torch.set_default_tensor_type(torch.FloatTensor)
     model = load_model(model, args.load_model_path)
+    print('done load model.')
     model.eval()
 
     device = torch.device("cuda" if torch.cuda.is_available() else "cpu")