Enable Flashinfer TRTLLM-GEN-MoE FP8 blockwise kernel for Qwen3-Next on Blackwell (#12543)

python/sglang/srt/models/qwen2_moe.py

@@ -57,6 +57,7 @@ from sglang.srt.layers.moe import get_moe_a2a_backend
 from sglang.srt.layers.moe.ep_moe.layer import get_moe_impl_class
 from sglang.srt.layers.moe.fused_moe_triton import FusedMoE
 from sglang.srt.layers.moe.topk import TopK
+from sglang.srt.layers.moe.utils import RoutingMethodType
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
 from sglang.srt.layers.radix_attention import RadixAttention
 from sglang.srt.layers.rotary_embedding import get_rope
@@ -162,6 +163,7 @@ class Qwen2MoeSparseMoeBlock(nn.Module):
             intermediate_size=config.moe_intermediate_size,
             quant_config=quant_config,
             prefix=add_prefix("experts", prefix),
+            routing_method_type=RoutingMethodType.RenormalizeNaive,
         )
 
         self.gate = ReplicatedLinear(