fix interface error

csrc/quantization/fp8/amd/quant_utils.cuh

@@ -160,15 +160,15 @@ template <>
 __inline__ __device__ Float4_
 scaled_vec_conversion<Float4_, uint32_t>(const uint32_t& a, const float scale, Fp8KVCacheDataType kv_type) {
   Float4_ res;
-  res.x = scaled_vec_conversion<float2, uint16_t>((uint16_t)a, scale);
-  res.y = scaled_vec_conversion<float2, uint16_t>((uint16_t)(a >> 16U), scale);
+  res.x = scaled_vec_conversion<float2, uint16_t>((uint16_t)a, scale, kv_type);
+  res.y = scaled_vec_conversion<float2, uint16_t>((uint16_t)(a >> 16U), scale, kv_type);
   return res;
 }
 
 // fp8x4 -> float4
 template <>
 __inline__ __device__ float4
-scaled_vec_conversion<float4, uint32_t>(const uint32_t& a, float scale, kv_type) {
+scaled_vec_conversion<float4, uint32_t>(const uint32_t& a, float scale, Fp8KVCacheDataType kv_type) {
   Float4_ res = scaled_vec_conversion<Float4_, uint32_t>(a, scale, kv_type);
   return {res.x.x, res.x.y, res.y.x, res.y.y};
 }