LLM: fix pretraining_tp argument issue. (#9281)

python/llm/src/bigdl/llm/transformers/model.py

@@ -93,6 +93,9 @@ class _BaseAutoModelClass:
             kwargs["torch_dtype"] = kwargs.get("torch_dtype", 'auto')
             # Avoid tensor parallel F.Linear Operations
             if "pretraining_tp" in config_dict:
+                if "config" in kwargs:
+                    setattr(kwargs["config"], "pretraining_tp", 1)
+                else:
                     kwargs["pretraining_tp"] = 1
             q_k = load_in_low_bit if load_in_low_bit else "sym_int4"
             model = cls.load_convert(q_k, optimize_model, *args, **kwargs)