fix lint

Signed-off-by: realliujiaxu <[email protected]>

Author: realliujiaxu

vllm_ascend/ops/linear.py

@@ -188,18 +188,19 @@ def apply(
         return output, output_bias
 
 
-def get_custom_tp_group_column(disable_tp, prefix, layer):
+def get_custom_tp_group_column(
+        disable_tp, prefix,
+        layer) -> tuple[Optional[CustomTensorParallelBase], int, int]:
     if disable_tp:
         return None, 0, 1
 
+    custom_tp_group = None
     if "gate_up_proj" in prefix and mlp_tp_enable():
         custom_tp_group = MLPCustomColumnParallel(layer)
     elif "gate_up_proj" in prefix and dense_optim_enable():
         custom_tp_group = DenseOptimMergedColumnParallel(layer)
     elif dense_optim_enable():
         custom_tp_group = DenseOptimQKVParallelLinear(layer, prefix)
-    else:
-        custom_tp_group = None
 
     if custom_tp_group is not None:
         return custom_tp_group, custom_tp_group.tp_rank, custom_tp_group.tp_size
@@ -329,6 +330,7 @@ def apply(
                                                       self.hcomm_info,
                                                       bias=bias_)
         else:
+            assert self.quant_method is not None
             output = self.quant_method.apply(self.layer,
                                              input_parallel,
                                              bias=bias_)