diff --git a/fbgemm_gpu/codegen/training/backward/embedding_backward_split_host_template.cpp b/fbgemm_gpu/codegen/training/backward/embedding_backward_split_host_template.cpp
index 16a04c1d2..9628abb88 100644
--- a/fbgemm_gpu/codegen/training/backward/embedding_backward_split_host_template.cpp
+++ b/fbgemm_gpu/codegen/training/backward/embedding_backward_split_host_template.cpp
@@ -564,8 +564,8 @@ class {{ autograd_func }} :
     {%- if not nobag and dense and not vbe %}
     const Tensor& offsets,
     const int64_t pooling_mode,
-    const c10::optional<Tensor>& indice_weights,
-    const c10::optional<Tensor>& feature_requires_grad
+    const std::optional<Tensor>& indice_weights,
+    const std::optional<Tensor>& feature_requires_grad
     {%- elif not nobag %}
     const Tensor& offsets,
     const int64_t pooling_mode,
@@ -610,9 +610,9 @@ class {{ autograd_func }} :
     {{ args.split_function_args | join(", ") }}
     {%- else %}
     {%- if vbe %}
-    const c10::optional<Tensor>& B_offsets,
-    const c10::optional<Tensor>& vbe_output_offsets_feature_rank,
-    const c10::optional<Tensor>& vbe_B_offsets_rank_per_feature,
+    const std::optional<Tensor>& B_offsets,
+    const std::optional<Tensor>& vbe_output_offsets_feature_rank,
+    const std::optional<Tensor>& vbe_B_offsets_rank_per_feature,
     const c10::SymInt max_B,
     const c10::SymInt max_B_feature_rank,
     const c10::SymInt vbe_output_size
@@ -1025,7 +1025,7 @@ Tensor {{ bwd_mdesc }}_embedding_codegen_lookup_{{ optimizer }}_function(
     {%- endif %}
     const bool apply_global_weight_decay = false,
     {%- if ssd %}
-    const c10::optional<at::TensorList>& ssd_tensors = c10::nullopt,
+    const std::optional<at::TensorList>& ssd_tensors = c10::nullopt,
     {%- endif %}
     const double gwd_lower_bound = 0
     {%- else %}
diff --git a/fbgemm_gpu/codegen/training/pt2/embedding_split_host_pt2_autograd_template.cpp b/fbgemm_gpu/codegen/training/pt2/embedding_split_host_pt2_autograd_template.cpp
index 3666de5b9..9c0eb320b 100644
--- a/fbgemm_gpu/codegen/training/pt2/embedding_split_host_pt2_autograd_template.cpp
+++ b/fbgemm_gpu/codegen/training/pt2/embedding_split_host_pt2_autograd_template.cpp
@@ -1005,7 +1005,7 @@ Tensor {{ bwd_mdesc }}_embedding_codegen_lookup_{{ optimizer }}_function_pt2(
     {%- endif %}
     const bool apply_global_weight_decay = false,
     {%- if ssd %}
-    const c10::optional<at::TensorList>& ssd_tensors = c10::nullopt,
+    const std::optional<at::TensorList>& ssd_tensors = c10::nullopt,
     {%- endif %}
     const double gwd_lower_bound = 0
 ) {
diff --git a/fbgemm_gpu/docs/src/nitpick.ignore b/fbgemm_gpu/docs/src/nitpick.ignore
index 2bb503fc5..a8dcceda7 100644
--- a/fbgemm_gpu/docs/src/nitpick.ignore
+++ b/fbgemm_gpu/docs/src/nitpick.ignore
@@ -12,10 +12,10 @@ cpp:identifier at::Tensor
 cpp:identifier block_type_t
 cpp:identifier c10
 cpp:identifier c10::Error
-cpp:identifier c10::optional<at::SymInt>
-cpp:identifier c10::optional<at::Tensor>
-cpp:identifier c10::optional<int64_t>
-cpp:identifier c10::optional<Tensor>
+cpp:identifier std::optional<at::SymInt>
+cpp:identifier std::optional<at::Tensor>
+cpp:identifier std::optional<int64_t>
+cpp:identifier std::optional<Tensor>
 cpp:identifier c10::SymInt
 cpp:identifier c10::SymIntArrayRef
 cpp:identifier int32_t