Update app.py

app.py

@@ -22,7 +22,7 @@ def init_args():
     args.batch_size = 1
     args.seq_length = 1024
     args.world_size = 1
-    args.use_int8 = True
+    args.use_int8 = False
     args.top_p = 0
     args.repetition_penalty_range = 1024
     args.repetition_penalty_slope = 0
@@ -38,7 +38,7 @@ 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)
     model.eval()