remove quantization due to lack of cuda

model.py

@@ -26,7 +26,7 @@ def get_local_model(model_name_or_path:str)->pipeline:
     model = AutoModelForCausalLM.from_pretrained( 
         model_name_or_path,
         torch_dtype=torch.bfloat16,
-        load_in_4bit = True,
+        # load_in_4bit = True,
         token = hf_token
     )
     pipe = pipeline(