[6/N] MoE Refactor: Cleanup MoE-related configs (#8849)

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

@@ -33,7 +33,8 @@ from sglang.srt.layers.quantization.unquant import UnquantizedLinearMethod
 from sglang.srt.layers.quantization.utils import get_scalar_types, replace_parameter
 
 if TYPE_CHECKING:
-    from sglang.srt.layers.moe.topk import TopKOutput
+    from sglang.srt.layers.moe.moe_runner import MoeRunnerConfig
+    from sglang.srt.layers.moe.topk import StandardTopKOutput
 
 from sglang.srt.utils import is_cuda, is_hip
 
@@ -739,13 +740,12 @@ class AWQMoEMethod(FusedMoEMethodBase):
         self,
         layer: torch.nn.Module,
         x: torch.Tensor,
-        topk_output: TopKOutput,
-        *,
-        activation: str = "silu",
-        **kwargs,
+        topk_output: StandardTopKOutput,
+        moe_runner_config: MoeRunnerConfig,
     ) -> torch.Tensor:
-
-        assert activation == "silu", "Only SiLU activation is supported."
+        assert (
+            moe_runner_config.activation == "silu"
+        ), "Only SiLU activation is supported."
 
         # The input must currently be float16
         orig_dtype = x.dtype