[chore][1/N] Avoid using default mutable parameters (#11478)

Signed-off-by: Kai-Hsun Chen <khchen@x.ai>

python/sglang/srt/layers/quantization/compressed_tensors/compressed_tensors.py

@@ -86,7 +86,7 @@ class CompressedTensorsConfig(QuantizationConfig):
         sparsity_ignore_list: List[str],
         kv_cache_scheme: Optional[Dict[str, Any]] = None,
         config: Optional[Dict[str, Any]] = None,
-        packed_modules_mapping: Dict[str, List[str]] = {},
+        packed_modules_mapping: Optional[Dict[str, List[str]]] = None,
     ):
         super().__init__()
         self.ignore = ignore
@@ -97,7 +97,7 @@ class CompressedTensorsConfig(QuantizationConfig):
         self.sparsity_scheme_map = sparsity_scheme_map
         self.sparsity_ignore_list = sparsity_ignore_list
         self.config = config
-        self.packed_modules_mapping = packed_modules_mapping
+        self.packed_modules_mapping = packed_modules_mapping or {}
 
     def get_linear_method(self) -> CompressedTensorsLinearMethod:
         return CompressedTensorsLinearMethod(self)

python/sglang/srt/utils/common.py

@@ -492,7 +492,7 @@ def make_layers(
     pp_size: Optional[int] = None,
     prefix: str = "",
     return_tuple: bool = False,
-    offloader_kwargs: Dict[str, Any] = {},
+    offloader_kwargs: Optional[Dict[str, Any]] = None,
 ) -> Tuple[torch.nn.Module, int, int]:
     """Make a list of layers with the given layer function"""
     # circula imports
@@ -517,7 +517,7 @@ def make_layers(
                 layer_fn(idx=idx, prefix=add_prefix(idx, prefix))
                 for idx in range(start_layer, end_layer)
             ),
-            **offloader_kwargs,
+            **(offloader_kwargs or {}),
         )
         + [
             PPMissingLayer(return_tuple=return_tuple)