Update prototype_source/max_autotune_on_CPU_tutorial.rst

Co-authored-by: Svetlana Karslioglu <[email protected]>

Author: chunyuan-w

prototype_source/max_autotune_on_CPU_tutorial.rst

@@ -98,7 +98,7 @@ In this example, C++ template outperforms ATen kernel so that it will be selecte
 We could check the generated output code by setting ``export TORCH_LOGS="+output_code"``.
 When C++ template is selected, we won't have ``torch.ops.mkldnn._linear_pointwise.default`` (for bfloat16) or ``torch.ops.mkl._mkl_linear.default`` (for float32)
 in the generated code anymore, instead, we'll find kernel based on CPP GEMM template ``cpp_fused__to_copy_relu_1``
-(only part of the code is demonstrated below for simplicity) with the bias and relu epilogues fused inside the CPP GEMM template kernel.
+(only part of the code is demonstrated below for simplicity) with the bias and relu epilogues fused inside the C++ GEMM template kernel.
 
 The generated code differs by CPU architecture and is implementation-specific, which is subject to change.