move import of AsecndQuantConfig to pre_register

Signed-off-by: angazenn <zengyanjia@huawei.com>

vllm_ascend/__init__.py

@@ -18,8 +18,4 @@
 
 def register():
     """Register the NPU platform."""
-    # TODO: https://github.com/vllm-project/vllm/pull/12432 Once this pr is merged, 
-    # the following module can be imported using pre_register_and_update function.
-    from vllm_ascend.quantization.quant_config import AscendQuantConfig  # noqa: F401
-
     return "vllm_ascend.platform.NPUPlatform"

vllm_ascend/platform.py

@@ -94,6 +94,21 @@ def synchronize(cls):
     def mem_get_info(cls) -> Tuple[int, int]:
         return torch.npu.mem_get_info()
 
+    @classmethod
+    def pre_register_and_update(cls,
+                                parser: Optional[FlexibleArgumentParser] = None
+                                ) -> None:
+        """
+        Do some pre-registeration or update action for the current platform.
+        This function is called before global VllmConfig is initialized or cli
+        arguments are parsed. It's used for out-of-tree platforms to register or
+        update the configuration.
+        For example, the out-of-tree quantization config can be imported and
+        registered here dynamically.
+        """
+
+        from vllm_ascend.quantization.quant_config import AscendQuantConfig  # noqa: F401
+
     @classmethod
     def check_and_update_config(cls, vllm_config: VllmConfig) -> None:
         # Register ops when setup.