skip silu_and_mul_quant

csrc/ops.h

@@ -260,8 +260,8 @@ void rotary_embedding_tgi(
 
 void silu_and_mul(torch::Tensor& out, torch::Tensor& input);
 
-void silu_and_mul_quant(torch::Tensor& out, torch::Tensor& input,
-                        torch::Tensor& scale);
+// void silu_and_mul_quant(torch::Tensor& out, torch::Tensor& input,
+//                         torch::Tensor& scale);
 
 void mul_and_silu(torch::Tensor& out, torch::Tensor& input);
 

csrc/torch_bindings.cpp

@@ -259,9 +259,9 @@ TORCH_LIBRARY_EXPAND(TORCH_EXTENSION_NAME, ops) {
   ops.def("silu_and_mul(Tensor! result, Tensor input) -> ()");
   ops.impl("silu_and_mul", torch::kCUDA, &silu_and_mul);
 
-  ops.def(
-      "silu_and_mul_quant(Tensor! result, Tensor input, Tensor scale) -> ()");
-  ops.impl("silu_and_mul_quant", torch::kCUDA, &silu_and_mul_quant);
+//   ops.def(
+//       "silu_and_mul_quant(Tensor! result, Tensor input, Tensor scale) -> ()");
+//   ops.impl("silu_and_mul_quant", torch::kCUDA, &silu_and_mul_quant);
 
   ops.def("mul_and_silu(Tensor! out, Tensor input) -> ()");
   ops.impl("mul_and_silu", torch::kCUDA, &mul_and_silu);