remove redundant routed_scaling_factor

vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe.py

@@ -838,7 +838,6 @@ class CompressedTensorsW8A8Fp8MoEMethod(CompressedTensorsMoEMethod):
         activation: str = "silu",
         enable_eplb: bool = False,
         use_nn_moe: Optional[bool] = False,
-        routed_scaling_factor: Optional[float] = None,
         use_fused_gate: Optional[bool] = False,
         expert_load_view: Optional[torch.Tensor] = None,
         logical_to_physical_map: Optional[torch.Tensor] = None,
@@ -1112,7 +1111,6 @@ class CompressedTensorsW8A8Int8MoEMethod(CompressedTensorsMoEMethod):
         activation: str = "silu",
         enable_eplb: bool = False,
         use_nn_moe: Optional[bool] = False,
-        routed_scaling_factor: Optional[float] = None,
         use_fused_gate: Optional[bool] = False,
         expert_load_view: Optional[torch.Tensor] = None,
         logical_to_physical_map: Optional[torch.Tensor] = None,

vllm/model_executor/layers/quantization/moe_wna16.py

@@ -348,7 +348,6 @@ class MoeWNA16Method(FusedMoEMethodBase):
         logical_to_physical_map: Optional[torch.Tensor] = None,
         logical_replica_count: Optional[torch.Tensor] = None,
         use_nn_moe: Optional[bool] = False,
-        routed_scaling_factor: Optional[float] = None,
         use_fused_gate: Optional[bool] = False,
     ) -> torch.Tensor:
         assert self.fused_experts is None
@@ -370,7 +369,6 @@ class MoeWNA16Method(FusedMoEMethodBase):
             routed_scaling_factor=routed_scaling_factor,
             e_score_correction_bias=e_score_correction_bias,
             indices_type=self.topk_indices_dtype,
-            routed_scaling_factor=routed_scaling_factor,
             use_fused_gate=use_fused_gate)
 
         weight_bits = self.quant_config.weight_bits

vllm/model_executor/models/deepseek_v2.py

@@ -170,8 +170,7 @@ class DeepseekV2MoE(nn.Module):
             routed_scaling_factor=self.routed_scaling_factor,
             e_score_correction_bias=self.gate.e_score_correction_bias,
             enable_eplb=self.enable_eplb,
-            num_redundant_experts=self.n_redundant_experts,
-            routed_scaling_factor=self.routed_scaling_factor)
+            num_redundant_experts=self.n_redundant_experts)
 
         if config.n_shared_experts is not None:
             intermediate_size = (config.moe_intermediate_size *

vllm/model_executor/models/deepseek_v3.py

@@ -104,7 +104,6 @@ class DeepseekV3MoE(nn.Module):
         self.tp_size = get_tensor_model_parallel_world_size()
         self.routed_scaling_factor = config.routed_scaling_factor
         self.n_shared_experts = config.n_shared_experts
-        self.routed_scaling_factor = config.routed_scaling_factor
         if self.tp_size > config.n_routed_experts:
             raise ValueError(
                 f"Tensor parallel size {self.tp_size} is greater than "