cleanup deps 3/n (#4541)

python/sglang/srt/layers/quantization/fp8.py

@@ -152,8 +152,6 @@ class Fp8Config(QuantizationConfig):
     def get_quant_method(
         self, layer: torch.nn.Module, prefix: str
     ) -> Optional["QuantizeMethodBase"]:
-        from vllm.attention.layer import Attention  # Avoid circular import
-
         from sglang.srt.layers.moe.fused_moe_triton import FusedMoE
 
         if isinstance(layer, LinearBase):
@@ -162,8 +160,6 @@ class Fp8Config(QuantizationConfig):
             return Fp8LinearMethod(self)
         elif isinstance(layer, FusedMoE):
             return Fp8MoEMethod(self)
-        elif isinstance(layer, Attention):
-            return Fp8KVCacheMethod(self)
         return None
 
     def get_scaled_act_names(self) -> List[str]: