diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..e6b956b448a3775fe5459fb263aad74a9da665c4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_backward_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _adaptive_avg_pool2d_backward_out(const at::Tensor & grad_output, const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor adaptive_avg_pool2d_backward_cpu(const at::Tensor & grad_output, const at::Tensor & self);
+TORCH_API at::Tensor adaptive_avg_pool2d_backward_cuda(const at::Tensor & grad_output, const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5db47506572e419face854e4ac2bc84a761ce602
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_native.h
@@ -0,0 +1,25 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _adaptive_avg_pool2d_out_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, at::Tensor & out);
+TORCH_API at::Tensor adaptive_avg_pool2d_cpu(const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API at::Tensor adaptive_avg_pool2d_cuda(const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API at::Tensor adaptive_avg_pool2d_quantized_cpu(const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API at::Tensor adaptive_avg_pool2d_quantized_cuda(const at::Tensor & self, at::IntArrayRef output_size);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..c4b1aa5c11d68bf01210b70ebd4871499d62496c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _adaptive_avg_pool2d {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_adaptive_avg_pool2d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_adaptive_avg_pool2d(Tensor self, SymInt[2] output_size) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef output_size);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef output_size);
+};
+
+struct TORCH_API _adaptive_avg_pool2d_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_adaptive_avg_pool2d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_adaptive_avg_pool2d.out(Tensor self, SymInt[2] output_size, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, c10::SymIntArrayRef output_size, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef output_size, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_add_batch_dim_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_add_batch_dim_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..93c39445fbfedc557d3f1d09bf801804a7184872
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_add_batch_dim_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor _add_batch_dim(const at::Tensor & self, int64_t batch_dim, int64_t level);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_addmm_activation_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_addmm_activation_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..556907110c403f7d72c45942ac6a4149da96f110
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_addmm_activation_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor _addmm_activation(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1, bool use_gelu=false);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Byte_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Byte_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8de0067cec8116c0f4e5b16ad93fd37c6cb46420
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Byte_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor _cast_Byte(const at::Tensor & self, bool non_blocking=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Int_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Int_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..51c1bb0907985a1c71c9812d89609d69a10c5bb4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Int_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor _cast_Int(const at::Tensor & self, bool non_blocking=false);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cholesky_solve_helper_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cholesky_solve_helper_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..89609b25562942767e5288ec3bd5d007ca05eebf
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cholesky_solve_helper_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _cholesky_solve_helper_out(const at::Tensor & self, const at::Tensor & A, bool upper, at::Tensor & out);
+TORCH_API at::Tensor _cholesky_solve_helper_cpu(const at::Tensor & self, const at::Tensor & A, bool upper);
+TORCH_API at::Tensor _cholesky_solve_helper_cuda(const at::Tensor & self, const at::Tensor & A, bool upper);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr.h
new file mode 100644
index 0000000000000000000000000000000000000000..f9c6030b39a5b395b996b434e6d47729b99b35a8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_convert_indices_from_coo_to_csr_ops.h>
+
+namespace at {
+
+
+// aten::_convert_indices_from_coo_to_csr(Tensor self, int size, *, bool out_int32=False) -> Tensor
+inline at::Tensor _convert_indices_from_coo_to_csr(const at::Tensor & self, int64_t size, bool out_int32=false) {
+    return at::_ops::_convert_indices_from_coo_to_csr::call(self, size, out_int32);
+}
+
+// aten::_convert_indices_from_coo_to_csr.out(Tensor self, int size, *, bool out_int32=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _convert_indices_from_coo_to_csr_out(at::Tensor & out, const at::Tensor & self, int64_t size, bool out_int32=false) {
+    return at::_ops::_convert_indices_from_coo_to_csr_out::call(self, size, out_int32, out);
+}
+// aten::_convert_indices_from_coo_to_csr.out(Tensor self, int size, *, bool out_int32=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _convert_indices_from_coo_to_csr_outf(const at::Tensor & self, int64_t size, bool out_int32, at::Tensor & out) {
+    return at::_ops::_convert_indices_from_coo_to_csr_out::call(self, size, out_int32, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_weight_to_int4pack_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_weight_to_int4pack_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..56f9d999abda53f327db2e2e55c5e8618c1e7036
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_weight_to_int4pack_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _convert_weight_to_int4pack {
+  using schema = at::Tensor (const at::Tensor &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_convert_weight_to_int4pack";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_convert_weight_to_int4pack(Tensor self, int innerKTiles) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t innerKTiles);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t innerKTiles);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cslt_compress.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cslt_compress.h
new file mode 100644
index 0000000000000000000000000000000000000000..9ded01bb61b1398be9444ea94a5c6226fcd7fc0a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cslt_compress.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_cslt_compress_ops.h>
+
+namespace at {
+
+
+// aten::_cslt_compress(Tensor input) -> Tensor
+inline at::Tensor _cslt_compress(const at::Tensor & input) {
+    return at::_ops::_cslt_compress::call(input);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_backward_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_backward_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7d9c1280be606d46f0dca7a109c1695decbf4963
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_backward_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API void _cudnn_rnn_backward_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::TensorList out3, const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, const at::Tensor & output, const ::std::optional<at::Tensor> & grad_output, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, int64_t mode, int64_t hidden_size, int64_t proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, at::IntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask);
+TORCH_API void _cudnn_rnn_backward_outf(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, const at::Tensor & output, const ::std::optional<at::Tensor> & grad_output, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, int64_t mode, int64_t hidden_size, int64_t proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, at::IntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::TensorList out3);
+TORCH_API void _cudnn_rnn_backward_symint_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::TensorList out3, const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, const at::Tensor & output, const ::std::optional<at::Tensor> & grad_output, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, int64_t mode, c10::SymInt hidden_size, c10::SymInt proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, c10::SymIntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask);
+TORCH_API void _cudnn_rnn_backward_symint_outf(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, const at::Tensor & output, const ::std::optional<at::Tensor> & grad_output, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, int64_t mode, c10::SymInt hidden_size, c10::SymInt proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, c10::SymIntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::TensorList out3);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..17c7cd83fdff006203931ff80766bacbad4a3737
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_backward_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _cudnn_rnn_backward {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor,::std::vector<at::Tensor>> (const at::Tensor &, at::TensorList, int64_t, const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, int64_t, c10::SymInt, c10::SymInt, int64_t, bool, double, bool, bool, c10::SymIntArrayRef, const ::std::optional<at::Tensor> &, const at::Tensor &, ::std::array<bool,4>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_cudnn_rnn_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_cudnn_rnn_backward(Tensor input, Tensor[] weight, int weight_stride0, Tensor weight_buf, Tensor hx, Tensor? cx, Tensor output, Tensor? grad_output, Tensor? grad_hy, Tensor? grad_cy, int mode, SymInt hidden_size, SymInt proj_size, int num_layers, bool batch_first, float dropout, bool train, bool bidirectional, SymInt[] batch_sizes, Tensor? dropout_state, Tensor reserve, bool[4] output_mask) -> (Tensor, Tensor, Tensor, Tensor[])";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,::std::vector<at::Tensor>> call(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, const at::Tensor & output, const ::std::optional<at::Tensor> & grad_output, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, int64_t mode, c10::SymInt hidden_size, c10::SymInt proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, c10::SymIntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,::std::vector<at::Tensor>> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, const at::Tensor & output, const ::std::optional<at::Tensor> & grad_output, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, int64_t mode, c10::SymInt hidden_size, c10::SymInt proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, c10::SymIntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask);
+};
+
+struct TORCH_API _cudnn_rnn_backward_out {
+  using schema = void (const at::Tensor &, at::TensorList, int64_t, const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, int64_t, c10::SymInt, c10::SymInt, int64_t, bool, double, bool, bool, c10::SymIntArrayRef, const ::std::optional<at::Tensor> &, const at::Tensor &, ::std::array<bool,4>, at::Tensor &, at::Tensor &, at::Tensor &, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_cudnn_rnn_backward";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_cudnn_rnn_backward.out(Tensor input, Tensor[] weight, int weight_stride0, Tensor weight_buf, Tensor hx, Tensor? cx, Tensor output, Tensor? grad_output, Tensor? grad_hy, Tensor? grad_cy, int mode, SymInt hidden_size, SymInt proj_size, int num_layers, bool batch_first, float dropout, bool train, bool bidirectional, SymInt[] batch_sizes, Tensor? dropout_state, Tensor reserve, bool[4] output_mask, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2, Tensor(d!)[] out3) -> ()";
+  static void call(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, const at::Tensor & output, const ::std::optional<at::Tensor> & grad_output, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, int64_t mode, c10::SymInt hidden_size, c10::SymInt proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, c10::SymIntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::TensorList out3);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, const at::Tensor & output, const ::std::optional<at::Tensor> & grad_output, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, int64_t mode, c10::SymInt hidden_size, c10::SymInt proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, c10::SymIntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::TensorList out3);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..03347a95c484d1a8b665444a2ce0c25810676472
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_cuda_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _cudnn_rnn(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const ::std::optional<at::Tensor> & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, int64_t mode, int64_t hidden_size, int64_t proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, at::IntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _cudnn_rnn_symint(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const ::std::optional<at::Tensor> & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, int64_t mode, c10::SymInt hidden_size, c10::SymInt proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, c10::SymIntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_efficientzerotensor_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_efficientzerotensor_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..3a5b2621bdffe1e0bcc38ab9b3da2f2e7dac111c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_efficientzerotensor_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _efficientzerotensor {
+  using schema = at::Tensor (c10::SymIntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_efficientzerotensor";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_efficientzerotensor(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API _efficientzerotensor_out {
+  using schema = at::Tensor & (c10::SymIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_efficientzerotensor";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_efficientzerotensor.out(SymInt[] size, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymIntArrayRef size, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9a1e5fccc1bdf89994d03ad7e7940bcc985cb156
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_backward_cuda_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor _embedding_bag_backward(const at::Tensor & grad, const at::Tensor & indices, const at::Tensor & offsets, const at::Tensor & offset2bag, const at::Tensor & bag_size, const at::Tensor & maximum_indices, int64_t num_weights, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, int64_t padding_idx=-1);
+TORCH_API at::Tensor _embedding_bag_backward_symint(const at::Tensor & grad, const at::Tensor & indices, const at::Tensor & offsets, const at::Tensor & offset2bag, const at::Tensor & bag_size, const at::Tensor & maximum_indices, c10::SymInt num_weights, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, int64_t padding_idx=-1);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_sparse_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_sparse_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..9dab94042c8e41cfa7e15bd9b71b15a1d1678028
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_sparse_backward.h
@@ -0,0 +1,47 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_embedding_bag_sparse_backward_ops.h>
+
+namespace at {
+
+
+// aten::_embedding_bag_sparse_backward(Tensor grad, Tensor indices, Tensor offsets, Tensor offset2bag, Tensor bag_size, SymInt num_weights, bool scale_grad_by_freq, int mode, Tensor? per_sample_weights, int padding_idx=-1) -> Tensor
+inline at::Tensor _embedding_bag_sparse_backward(const at::Tensor & grad, const at::Tensor & indices, const at::Tensor & offsets, const at::Tensor & offset2bag, const at::Tensor & bag_size, int64_t num_weights, bool scale_grad_by_freq, int64_t mode, const ::std::optional<at::Tensor> & per_sample_weights, int64_t padding_idx=-1) {
+    return at::_ops::_embedding_bag_sparse_backward::call(grad, indices, offsets, offset2bag, bag_size, num_weights, scale_grad_by_freq, mode, per_sample_weights, padding_idx);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor _embedding_bag_sparse_backward(const at::Tensor & grad, const at::Tensor & indices, const at::Tensor & offsets, const at::Tensor & offset2bag, const at::Tensor & bag_size, int64_t num_weights, bool scale_grad_by_freq, int64_t mode, const ::std::optional<at::Tensor> & per_sample_weights, int64_t padding_idx=-1) {
+    return at::_ops::_embedding_bag_sparse_backward::call(grad, indices, offsets, offset2bag, bag_size, num_weights, scale_grad_by_freq, mode, per_sample_weights, padding_idx);
+  }
+}
+
+// aten::_embedding_bag_sparse_backward(Tensor grad, Tensor indices, Tensor offsets, Tensor offset2bag, Tensor bag_size, SymInt num_weights, bool scale_grad_by_freq, int mode, Tensor? per_sample_weights, int padding_idx=-1) -> Tensor
+inline at::Tensor _embedding_bag_sparse_backward_symint(const at::Tensor & grad, const at::Tensor & indices, const at::Tensor & offsets, const at::Tensor & offset2bag, const at::Tensor & bag_size, c10::SymInt num_weights, bool scale_grad_by_freq, int64_t mode, const ::std::optional<at::Tensor> & per_sample_weights, int64_t padding_idx=-1) {
+    return at::_ops::_embedding_bag_sparse_backward::call(grad, indices, offsets, offset2bag, bag_size, num_weights, scale_grad_by_freq, mode, per_sample_weights, padding_idx);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor _embedding_bag_sparse_backward(const at::Tensor & grad, const at::Tensor & indices, const at::Tensor & offsets, const at::Tensor & offset2bag, const at::Tensor & bag_size, c10::SymInt num_weights, bool scale_grad_by_freq, int64_t mode, const ::std::optional<at::Tensor> & per_sample_weights, int64_t padding_idx=-1) {
+    return at::_ops::_embedding_bag_sparse_backward::call(grad, indices, offsets, offset2bag, bag_size, num_weights, scale_grad_by_freq, mode, per_sample_weights, padding_idx);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_euclidean_dist.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_euclidean_dist.h
new file mode 100644
index 0000000000000000000000000000000000000000..d24a5036cfaae5aaea87836fdbae535ec431117e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_euclidean_dist.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_euclidean_dist_ops.h>
+
+namespace at {
+
+
+// aten::_euclidean_dist(Tensor x1, Tensor x2) -> Tensor
+inline at::Tensor _euclidean_dist(const at::Tensor & x1, const at::Tensor & x2) {
+    return at::_ops::_euclidean_dist::call(x1, x2);
+}
+
+// aten::_euclidean_dist.out(Tensor x1, Tensor x2, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _euclidean_dist_out(at::Tensor & out, const at::Tensor & x1, const at::Tensor & x2) {
+    return at::_ops::_euclidean_dist_out::call(x1, x2, out);
+}
+// aten::_euclidean_dist.out(Tensor x1, Tensor x2, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _euclidean_dist_outf(const at::Tensor & x1, const at::Tensor & x2, at::Tensor & out) {
+    return at::_ops::_euclidean_dist_out::call(x1, x2, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_channel_affine_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_channel_affine_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d809f673fd81ea2d225cace213b2f2be596ccc9b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_channel_affine_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _fake_quantize_learnable_per_channel_affine_out(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max, double grad_factor, at::Tensor & out);
+TORCH_API at::Tensor _fake_quantize_learnable_per_channel_affine(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max, double grad_factor=1.0);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams.h
new file mode 100644
index 0000000000000000000000000000000000000000..edbf835a599b7acd57c0dd81f62538006d0c6020
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams_ops.h>
+
+namespace at {
+
+
+// aten::_fake_quantize_per_tensor_affine_cachemask_tensor_qparams(Tensor self, Tensor scale, Tensor zero_point, Tensor fake_quant_enabled, int quant_min, int quant_max) -> (Tensor output, Tensor mask)
+inline ::std::tuple<at::Tensor,at::Tensor> _fake_quantize_per_tensor_affine_cachemask_tensor_qparams(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, const at::Tensor & fake_quant_enabled, int64_t quant_min, int64_t quant_max) {
+    return at::_ops::_fake_quantize_per_tensor_affine_cachemask_tensor_qparams::call(self, scale, zero_point, fake_quant_enabled, quant_min, quant_max);
+}
+
+// aten::_fake_quantize_per_tensor_affine_cachemask_tensor_qparams.out(Tensor self, Tensor scale, Tensor zero_point, Tensor fake_quant_enabled, int quant_min, int quant_max, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _fake_quantize_per_tensor_affine_cachemask_tensor_qparams_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, const at::Tensor & fake_quant_enabled, int64_t quant_min, int64_t quant_max) {
+    return at::_ops::_fake_quantize_per_tensor_affine_cachemask_tensor_qparams_out::call(self, scale, zero_point, fake_quant_enabled, quant_min, quant_max, out0, out1);
+}
+// aten::_fake_quantize_per_tensor_affine_cachemask_tensor_qparams.out(Tensor self, Tensor scale, Tensor zero_point, Tensor fake_quant_enabled, int quant_min, int quant_max, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _fake_quantize_per_tensor_affine_cachemask_tensor_qparams_outf(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, const at::Tensor & fake_quant_enabled, int64_t quant_min, int64_t quant_max, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::_fake_quantize_per_tensor_affine_cachemask_tensor_qparams_out::call(self, scale, zero_point, fake_quant_enabled, quant_min, quant_max, out0, out1);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fill_mem_eff_dropout_mask_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fill_mem_eff_dropout_mask_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..b7dc4f0cd344b523ee8fecee742c4f61bd23f78e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fill_mem_eff_dropout_mask_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _fill_mem_eff_dropout_mask_ {
+  using schema = at::Tensor & (at::Tensor &, double, int64_t, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fill_mem_eff_dropout_mask_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_fill_mem_eff_dropout_mask_(Tensor(a!) self, float dropout_p, int seed, int offset) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, double dropout_p, int64_t seed, int64_t offset);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, double dropout_p, int64_t seed, int64_t offset);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_flash_attention_forward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_flash_attention_forward.h
new file mode 100644
index 0000000000000000000000000000000000000000..1fff717f8b449efab5e11753b268a3215a0e43d4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_flash_attention_forward.h
@@ -0,0 +1,47 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_flash_attention_forward_ops.h>
+
+namespace at {
+
+
+// aten::_flash_attention_forward(Tensor query, Tensor key, Tensor value, Tensor? cum_seq_q, Tensor? cum_seq_k, SymInt max_q, SymInt max_k, float dropout_p, bool is_causal, bool return_debug_mask, *, float? scale=None, SymInt? window_size_left=None, SymInt? window_size_right=None, Tensor? seqused_k=None, Tensor? alibi_slopes=None) -> (Tensor output, Tensor softmax_logsumexp, Tensor philox_seed, Tensor philox_offset, Tensor debug_attn_mask)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _flash_attention_forward(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & cum_seq_q, const ::std::optional<at::Tensor> & cum_seq_k, int64_t max_q, int64_t max_k, double dropout_p, bool is_causal, bool return_debug_mask, ::std::optional<double> scale=::std::nullopt, ::std::optional<int64_t> window_size_left=::std::nullopt, ::std::optional<int64_t> window_size_right=::std::nullopt, const ::std::optional<at::Tensor> & seqused_k={}, const ::std::optional<at::Tensor> & alibi_slopes={}) {
+    return at::_ops::_flash_attention_forward::call(query, key, value, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, return_debug_mask, scale, window_size_left.has_value() ? ::std::make_optional(c10::SymInt(*window_size_left)) : ::std::nullopt, window_size_right.has_value() ? ::std::make_optional(c10::SymInt(*window_size_right)) : ::std::nullopt, seqused_k, alibi_slopes);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _flash_attention_forward(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & cum_seq_q, const ::std::optional<at::Tensor> & cum_seq_k, int64_t max_q, int64_t max_k, double dropout_p, bool is_causal, bool return_debug_mask, ::std::optional<double> scale=::std::nullopt, ::std::optional<int64_t> window_size_left=::std::nullopt, ::std::optional<int64_t> window_size_right=::std::nullopt, const ::std::optional<at::Tensor> & seqused_k={}, const ::std::optional<at::Tensor> & alibi_slopes={}) {
+    return at::_ops::_flash_attention_forward::call(query, key, value, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, return_debug_mask, scale, window_size_left.has_value() ? ::std::make_optional(c10::SymInt(*window_size_left)) : ::std::nullopt, window_size_right.has_value() ? ::std::make_optional(c10::SymInt(*window_size_right)) : ::std::nullopt, seqused_k, alibi_slopes);
+  }
+}
+
+// aten::_flash_attention_forward(Tensor query, Tensor key, Tensor value, Tensor? cum_seq_q, Tensor? cum_seq_k, SymInt max_q, SymInt max_k, float dropout_p, bool is_causal, bool return_debug_mask, *, float? scale=None, SymInt? window_size_left=None, SymInt? window_size_right=None, Tensor? seqused_k=None, Tensor? alibi_slopes=None) -> (Tensor output, Tensor softmax_logsumexp, Tensor philox_seed, Tensor philox_offset, Tensor debug_attn_mask)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _flash_attention_forward_symint(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & cum_seq_q, const ::std::optional<at::Tensor> & cum_seq_k, c10::SymInt max_q, c10::SymInt max_k, double dropout_p, bool is_causal, bool return_debug_mask, ::std::optional<double> scale=::std::nullopt, ::std::optional<c10::SymInt> window_size_left=::std::nullopt, ::std::optional<c10::SymInt> window_size_right=::std::nullopt, const ::std::optional<at::Tensor> & seqused_k={}, const ::std::optional<at::Tensor> & alibi_slopes={}) {
+    return at::_ops::_flash_attention_forward::call(query, key, value, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, return_debug_mask, scale, window_size_left, window_size_right, seqused_k, alibi_slopes);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _flash_attention_forward(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & cum_seq_q, const ::std::optional<at::Tensor> & cum_seq_k, c10::SymInt max_q, c10::SymInt max_k, double dropout_p, bool is_causal, bool return_debug_mask, ::std::optional<double> scale=::std::nullopt, ::std::optional<c10::SymInt> window_size_left=::std::nullopt, ::std::optional<c10::SymInt> window_size_right=::std::nullopt, const ::std::optional<at::Tensor> & seqused_k={}, const ::std::optional<at::Tensor> & alibi_slopes={}) {
+    return at::_ops::_flash_attention_forward::call(query, key, value, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, return_debug_mask, scale, window_size_left, window_size_right, seqused_k, alibi_slopes);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foobar_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foobar_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..e4e8550ce77e6bceb602c6fa088fd08cad767bfc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foobar_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _foobar_out(const at::Tensor & self, bool arg1, bool arg2, bool arg3, at::Tensor & out);
+TORCH_API at::Tensor foobar(const at::Tensor & self, bool arg1=true, bool arg2=true, bool arg3=true);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_abs_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_abs_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a1db2a2841e87d1580b5ddbca5e9f401e909b9a4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_abs_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::vector<at::Tensor> _foreach_abs(at::TensorList self);
+TORCH_API void _foreach_abs_out(at::TensorList out, at::TensorList self);
+TORCH_API void _foreach_abs_outf(at::TensorList self, at::TensorList out);
+TORCH_API void _foreach_abs_(at::TensorList self);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_exp_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_exp_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3336692ddb41f4dbb6532842233f09e890c9b86f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_exp_cuda_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::vector<at::Tensor> _foreach_exp(at::TensorList self);
+TORCH_API void _foreach_exp_(at::TensorList self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_exp_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_exp_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..ecb7204569b0c81dc62536c8267530f8f6b3e2b8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_exp_native.h
@@ -0,0 +1,25 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_exp_slow(at::TensorList self);
+TORCH_API void _foreach_exp_out(at::TensorList self, at::TensorList out);
+TORCH_API void foreach_tensor_exp_slow_(at::TensorList self);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_exp_cuda(at::TensorList self);
+TORCH_API void foreach_tensor_exp_cuda_(at::TensorList self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_mul_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_mul_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a217f275da078d3d9d944157dd37cdb6b63c82b0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_mul_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,38 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::vector<at::Tensor> _foreach_mul(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_mul_out(at::TensorList out, at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_mul_outf(at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+TORCH_API void _foreach_mul_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> _foreach_mul(at::TensorList self, at::TensorList other);
+TORCH_API void _foreach_mul_out(at::TensorList out, at::TensorList self, at::TensorList other);
+TORCH_API void _foreach_mul_outf(at::TensorList self, at::TensorList other, at::TensorList out);
+TORCH_API void _foreach_mul_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> _foreach_mul(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_mul_out(at::TensorList out, at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_mul_outf(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+TORCH_API void _foreach_mul_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API ::std::vector<at::Tensor> _foreach_mul(at::TensorList self, const at::Tensor & other);
+TORCH_API void _foreach_mul_out(at::TensorList out, at::TensorList self, const at::Tensor & other);
+TORCH_API void _foreach_mul_outf(at::TensorList self, const at::Tensor & other, at::TensorList out);
+TORCH_API void _foreach_mul_(at::TensorList self, const at::Tensor & other);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_mul_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_mul_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..06378aa37f2c17403516b29b252eb33ffbfc76be
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_mul_cuda_dispatch.h
@@ -0,0 +1,30 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::vector<at::Tensor> _foreach_mul(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_mul_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> _foreach_mul(at::TensorList self, at::TensorList other);
+TORCH_API void _foreach_mul_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> _foreach_mul(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_mul_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API ::std::vector<at::Tensor> _foreach_mul(at::TensorList self, const at::Tensor & other);
+TORCH_API void _foreach_mul_(at::TensorList self, const at::Tensor & other);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sigmoid_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sigmoid_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..f1a55a55d2e1e918c03d016f01e465369e376f96
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sigmoid_native.h
@@ -0,0 +1,25 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_sigmoid_slow(at::TensorList self);
+TORCH_API void _foreach_sigmoid_out(at::TensorList self, at::TensorList out);
+TORCH_API void foreach_tensor_sigmoid_slow_(at::TensorList self);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_sigmoid_cuda(at::TensorList self);
+TORCH_API void foreach_tensor_sigmoid_cuda_(at::TensorList self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_tan_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_tan_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f63082652e6a799953da96969263b338ac11adfd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_tan_ops.h
@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _foreach_tan {
+  using schema = ::std::vector<at::Tensor> (at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_tan";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_tan(Tensor[] self) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(at::TensorList self);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_tan_ {
+  using schema = void (at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_tan_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_tan_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_tan_out {
+  using schema = void (at::TensorList, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_tan";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_tan.out(Tensor[] self, *, Tensor(a!)[] out) -> ()";
+  static void call(at::TensorList self, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::TensorList out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adagrad_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adagrad_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..194cc10c5fb0634ffcf7352906eed2edd777f50f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adagrad_cpu_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API void _fused_adagrad_(at::TensorList self, at::TensorList grads, at::TensorList state_sums, at::TensorList state_steps, double lr, double lr_decay, double weight_decay, double eps, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adam_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adam_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e1e8b58d19cc0d28c2f7cfed1ec13d363eb20e4a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adam_cpu_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API void _fused_adam_(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, double lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+TORCH_API void _fused_adam_(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, const at::Tensor & lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_dropout_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_dropout_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..91d7808b6a1b19514a95d985b5e244a3088f3572
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_dropout_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _fused_dropout {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, double, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fused_dropout";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_fused_dropout(Tensor self, float p, Generator? generator=None) -> (Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self, double p, ::std::optional<at::Generator> generator);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, double p, ::std::optional<at::Generator> generator);
+};
+
+struct TORCH_API _fused_dropout_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const at::Tensor &, double, ::std::optional<at::Generator>, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fused_dropout";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_fused_dropout.out(Tensor self, float p, Generator? generator=None, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & self, double p, ::std::optional<at::Generator> generator, at::Tensor & out0, at::Tensor & out1);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, double p, ::std::optional<at::Generator> generator, at::Tensor & out0, at::Tensor & out1);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_bin_edges.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_bin_edges.h
new file mode 100644
index 0000000000000000000000000000000000000000..5e57bf4e44df14b22d49fb54d931a269d2a7d7d8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_bin_edges.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_histogramdd_bin_edges_ops.h>
+
+namespace at {
+
+
+// aten::_histogramdd_bin_edges(Tensor self, int[] bins, *, float[]? range=None, Tensor? weight=None, bool density=False) -> Tensor[]
+inline ::std::vector<at::Tensor> _histogramdd_bin_edges(const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range=::std::nullopt, const ::std::optional<at::Tensor> & weight={}, bool density=false) {
+    return at::_ops::_histogramdd_bin_edges::call(self, bins, range, weight, density);
+}
+
+// aten::_histogramdd_bin_edges.out(Tensor self, int[] bins, *, float[]? range=None, Tensor? weight=None, bool density=False, Tensor(a!)[] out) -> ()
+inline void _histogramdd_bin_edges_out(at::TensorList out, const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range=::std::nullopt, const ::std::optional<at::Tensor> & weight={}, bool density=false) {
+    return at::_ops::_histogramdd_bin_edges_out::call(self, bins, range, weight, density, out);
+}
+// aten::_histogramdd_bin_edges.out(Tensor self, int[] bins, *, float[]? range=None, Tensor? weight=None, bool density=False, Tensor(a!)[] out) -> ()
+inline void _histogramdd_bin_edges_outf(const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range, const ::std::optional<at::Tensor> & weight, bool density, at::TensorList out) {
+    return at::_ops::_histogramdd_bin_edges_out::call(self, bins, range, weight, density, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_cts_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_cts_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3fa5922b641e16962e38bdefbce4837e8087ae25
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_cts_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & _histogramdd_from_bin_cts_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range=::std::nullopt, const ::std::optional<at::Tensor> & weight={}, bool density=false);
+TORCH_API at::Tensor & _histogramdd_from_bin_cts_outf(const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range, const ::std::optional<at::Tensor> & weight, bool density, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_tensors.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_tensors.h
new file mode 100644
index 0000000000000000000000000000000000000000..29292a8a407b23a3c51d95039f2d0a3d2e039f6d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_tensors.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_histogramdd_from_bin_tensors_ops.h>
+
+namespace at {
+
+
+// aten::_histogramdd_from_bin_tensors(Tensor self, Tensor[] bins, *, Tensor? weight=None, bool density=False) -> Tensor
+inline at::Tensor _histogramdd_from_bin_tensors(const at::Tensor & self, at::TensorList bins, const ::std::optional<at::Tensor> & weight={}, bool density=false) {
+    return at::_ops::_histogramdd_from_bin_tensors::call(self, bins, weight, density);
+}
+
+// aten::_histogramdd_from_bin_tensors.out(Tensor self, Tensor[] bins, *, Tensor? weight=None, bool density=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _histogramdd_from_bin_tensors_out(at::Tensor & out, const at::Tensor & self, at::TensorList bins, const ::std::optional<at::Tensor> & weight={}, bool density=false) {
+    return at::_ops::_histogramdd_from_bin_tensors_out::call(self, bins, weight, density, out);
+}
+// aten::_histogramdd_from_bin_tensors.out(Tensor self, Tensor[] bins, *, Tensor? weight=None, bool density=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _histogramdd_from_bin_tensors_outf(const at::Tensor & self, at::TensorList bins, const ::std::optional<at::Tensor> & weight, bool density, at::Tensor & out) {
+    return at::_ops::_histogramdd_from_bin_tensors_out::call(self, bins, weight, density, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_check_errors.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_check_errors.h
new file mode 100644
index 0000000000000000000000000000000000000000..a5ebdb2bacfaed677afe006f36fafcaf32b57f81
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_check_errors.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_linalg_check_errors_ops.h>
+
+namespace at {
+
+
+// aten::_linalg_check_errors(Tensor info, str api_name, *, bool is_matrix) -> ()
+inline void _linalg_check_errors(const at::Tensor & info, c10::string_view api_name, bool is_matrix) {
+    return at::_ops::_linalg_check_errors::call(info, api_name, is_matrix);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_det_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_det_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..aca8896e7a3eca51a1dbc582a9f96fa5d16475be
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_det_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _linalg_det(const at::Tensor & A);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_solve_ex_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_solve_ex_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..34648efe13164c74c1cbef8ca0a38dec5ce4fc83
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_solve_ex_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> _linalg_solve_ex(const at::Tensor & A, const at::Tensor & B, bool left=true, bool check_errors=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _linalg_solve_ex_out(at::Tensor & result, at::Tensor & LU, at::Tensor & pivots, at::Tensor & info, const at::Tensor & A, const at::Tensor & B, bool left=true, bool check_errors=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _linalg_solve_ex_outf(const at::Tensor & A, const at::Tensor & B, bool left, bool check_errors, at::Tensor & result, at::Tensor & LU, at::Tensor & pivots, at::Tensor & info);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_backward_data_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_backward_data_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..28f511e6def083582952f466f308a53910f129e4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_backward_data_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _log_softmax_backward_data {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, int64_t, at::ScalarType);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_log_softmax_backward_data";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_log_softmax_backward_data(Tensor grad_output, Tensor output, int dim, ScalarType input_dtype) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, at::ScalarType input_dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, at::ScalarType input_dtype);
+};
+
+struct TORCH_API _log_softmax_backward_data_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, int64_t, at::ScalarType, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_log_softmax_backward_data";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_log_softmax_backward_data.out(Tensor grad_output, Tensor output, int dim, ScalarType input_dtype, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, at::ScalarType input_dtype, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, at::ScalarType input_dtype, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_lstm_mps_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_lstm_mps_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a59e4bdce00a283e01fad6d46f8ec3115cd93f1d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_lstm_mps_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _lstm_mps_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, at::Tensor & out4, at::Tensor & out5, const at::Tensor & input, at::TensorList hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional, bool batch_first);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _lstm_mps_outf(const at::Tensor & input, at::TensorList hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional, bool batch_first, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, at::Tensor & out4, at::Tensor & out5);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_make_per_tensor_quantized_tensor_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_make_per_tensor_quantized_tensor_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9011b93155e7c5e6ac2f3c8095e2580063a8b95b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_make_per_tensor_quantized_tensor_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _make_per_tensor_quantized_tensor {
+  using schema = at::Tensor (const at::Tensor &, double, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_make_per_tensor_quantized_tensor";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_make_per_tensor_quantized_tensor(Tensor self, float scale, int zero_point) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, double scale, int64_t zero_point);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, double scale, int64_t zero_point);
+};
+
+struct TORCH_API _make_per_tensor_quantized_tensor_out {
+  using schema = at::Tensor & (const at::Tensor &, double, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_make_per_tensor_quantized_tensor";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_make_per_tensor_quantized_tensor.out(Tensor self, float scale, int zero_point, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, double scale, int64_t zero_point, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, double scale, int64_t zero_point, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_masked_scale_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_masked_scale_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..374ea2de7422c5ccc263b3d46648176a4d637f74
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_masked_scale_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _masked_scale_out(const at::Tensor & self, const at::Tensor & mask, double scale, at::Tensor & out);
+TORCH_API at::Tensor masked_scale_cuda(const at::Tensor & self, const at::Tensor & mask, double scale);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_masked_softmax_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_masked_softmax_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..989f874bc7473ddd1e45b18a0828a502304e4079
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_masked_softmax_backward.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_masked_softmax_backward_ops.h>
+
+namespace at {
+
+
+// aten::_masked_softmax_backward(Tensor grad_output, Tensor output, Tensor mask, int? dim=None) -> Tensor
+inline at::Tensor _masked_softmax_backward(const at::Tensor & grad_output, const at::Tensor & output, const at::Tensor & mask, ::std::optional<int64_t> dim=::std::nullopt) {
+    return at::_ops::_masked_softmax_backward::call(grad_output, output, mask, dim);
+}
+
+// aten::_masked_softmax_backward.out(Tensor grad_output, Tensor output, Tensor mask, int? dim=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _masked_softmax_backward_out(at::Tensor & out, const at::Tensor & grad_output, const at::Tensor & output, const at::Tensor & mask, ::std::optional<int64_t> dim=::std::nullopt) {
+    return at::_ops::_masked_softmax_backward_out::call(grad_output, output, mask, dim, out);
+}
+// aten::_masked_softmax_backward.out(Tensor grad_output, Tensor output, Tensor mask, int? dim=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _masked_softmax_backward_outf(const at::Tensor & grad_output, const at::Tensor & output, const at::Tensor & mask, ::std::optional<int64_t> dim, at::Tensor & out) {
+    return at::_ops::_masked_softmax_backward_out::call(grad_output, output, mask, dim, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_mkldnn_transpose_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_mkldnn_transpose_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9469088ab234157b1dc198c9e19363895a124312
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_mkldnn_transpose_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & _mkldnn_transpose_out(at::Tensor & out, const at::Tensor & self, int64_t dim0, int64_t dim1);
+TORCH_API at::Tensor & _mkldnn_transpose_outf(const at::Tensor & self, int64_t dim0, int64_t dim1, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_native_multi_head_attention_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_multi_head_attention_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7995360ae249d850d4c5ec25b39b6f6edbff61b0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_multi_head_attention_cpu_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> _native_multi_head_attention(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, int64_t embed_dim, int64_t num_head, const at::Tensor & qkv_weight, const at::Tensor & qkv_bias, const at::Tensor & proj_weight, const at::Tensor & proj_bias, const ::std::optional<at::Tensor> & mask={}, bool need_weights=true, bool average_attn_weights=true, ::std::optional<int64_t> mask_type=::std::nullopt);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_neg_view_copy_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_neg_view_copy_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a9c0d7b840e402186378c89844a5871fe695d28d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_neg_view_copy_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & _neg_view_copy_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & _neg_view_copy_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_from_padded_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_from_padded_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7b58fb64779165a7eba1d06fe0c710d4bbb28aa1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_from_padded_cpu_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor _nested_from_padded(const at::Tensor & padded, const at::Tensor & cpu_nested_shape_example, bool fuse_transform_0213=false);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_lengths_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_lengths_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..ad7409936d1258c375acebaabc672c542b28c370
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_lengths_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _nested_get_lengths {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_nested_get_lengths";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_nested_get_lengths(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_max_seqlen_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_max_seqlen_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..df8394bae10e1da4e571154352ef3d822891ee04
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_max_seqlen_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _nested_get_max_seqlen {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_nested_get_max_seqlen";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_nested_get_max_seqlen(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_sum_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_sum_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..93f4738c5e461644a996d3e9bf190cbf99b3ce38
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_sum_backward.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_nested_sum_backward_ops.h>
+
+namespace at {
+
+
+// aten::_nested_sum_backward(Tensor grad, Tensor self, int[1]? dim, bool keepdim=False) -> Tensor
+inline at::Tensor _nested_sum_backward(const at::Tensor & grad, const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim=false) {
+    return at::_ops::_nested_sum_backward::call(grad, self, dim, keepdim);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_size_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_size_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f2a02ec4c81260ef7afa307797315c883d6ea9f8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_size_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & _nested_tensor_size_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & _nested_tensor_size_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_jagged_copy_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_jagged_copy_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1906c77671de42d3cde35a824d0681a5c809b972
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_jagged_copy_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _nested_view_from_jagged_copy {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, int64_t, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_nested_view_from_jagged_copy";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_nested_view_from_jagged_copy(Tensor self, Tensor offsets, Tensor dummy, Tensor? lengths=None, int ragged_idx=1, Tensor? min_seqlen=None, Tensor? max_seqlen=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & offsets, const at::Tensor & dummy, const ::std::optional<at::Tensor> & lengths, int64_t ragged_idx, const ::std::optional<at::Tensor> & min_seqlen, const ::std::optional<at::Tensor> & max_seqlen);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & offsets, const at::Tensor & dummy, const ::std::optional<at::Tensor> & lengths, int64_t ragged_idx, const ::std::optional<at::Tensor> & min_seqlen, const ::std::optional<at::Tensor> & max_seqlen);
+};
+
+struct TORCH_API _nested_view_from_jagged_copy_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, int64_t, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_nested_view_from_jagged_copy";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_nested_view_from_jagged_copy.out(Tensor self, Tensor offsets, Tensor dummy, Tensor? lengths=None, int ragged_idx=1, Tensor? min_seqlen=None, Tensor? max_seqlen=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & offsets, const at::Tensor & dummy, const ::std::optional<at::Tensor> & lengths, int64_t ragged_idx, const ::std::optional<at::Tensor> & min_seqlen, const ::std::optional<at::Tensor> & max_seqlen, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & offsets, const at::Tensor & dummy, const ::std::optional<at::Tensor> & lengths, int64_t ragged_idx, const ::std::optional<at::Tensor> & min_seqlen, const ::std::optional<at::Tensor> & max_seqlen, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_new_zeros_with_same_feature_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_new_zeros_with_same_feature_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..0f7e0c73649e9b9701f99f4e85e698dcfa282ae4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_new_zeros_with_same_feature_meta.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_new_zeros_with_same_feature_meta_ops.h>
+
+namespace at {
+
+
+// aten::_new_zeros_with_same_feature_meta(Tensor self, Tensor other, *, int self_num_batch_dims=0) -> Tensor
+inline at::Tensor _new_zeros_with_same_feature_meta(const at::Tensor & self, const at::Tensor & other, int64_t self_num_batch_dims=0) {
+    return at::_ops::_new_zeros_with_same_feature_meta::call(self, other, self_num_batch_dims);
+}
+
+// aten::_new_zeros_with_same_feature_meta.out(Tensor self, Tensor other, *, int self_num_batch_dims=0, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _new_zeros_with_same_feature_meta_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other, int64_t self_num_batch_dims=0) {
+    return at::_ops::_new_zeros_with_same_feature_meta_out::call(self, other, self_num_batch_dims, out);
+}
+// aten::_new_zeros_with_same_feature_meta.out(Tensor self, Tensor other, *, int self_num_batch_dims=0, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _new_zeros_with_same_feature_meta_outf(const at::Tensor & self, const at::Tensor & other, int64_t self_num_batch_dims, at::Tensor & out) {
+    return at::_ops::_new_zeros_with_same_feature_meta_out::call(self, other, self_num_batch_dims, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_new_zeros_with_same_feature_meta_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_new_zeros_with_same_feature_meta_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c3cec2c29433521430615a473a2aec1b47007e4f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_new_zeros_with_same_feature_meta_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor _new_zeros_with_same_feature_meta(const at::Tensor & self, const at::Tensor & other, int64_t self_num_batch_dims=0);
+TORCH_API at::Tensor & _new_zeros_with_same_feature_meta_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other, int64_t self_num_batch_dims=0);
+TORCH_API at::Tensor & _new_zeros_with_same_feature_meta_outf(const at::Tensor & self, const at::Tensor & other, int64_t self_num_batch_dims, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_pad_packed_sequence_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_pad_packed_sequence_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..f46eab8c8f89d3199f1deb7386b00b43968612ec
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_pad_packed_sequence_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> _pad_packed_sequence(const at::Tensor & data, const at::Tensor & batch_sizes, bool batch_first, const at::Scalar & padding_value, int64_t total_length);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_rowwise_prune_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_rowwise_prune_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..be3daf5ff8cdb9bb96a6753d24e0c688f7b54e2d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_rowwise_prune_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> _rowwise_prune(const at::Tensor & weight, const at::Tensor & mask, at::ScalarType compressed_indices_dtype);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_saturate_weight_to_fp16_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_saturate_weight_to_fp16_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..84083760141ce21aed73f8794464be6d9f785f96
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_saturate_weight_to_fp16_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor _saturate_weight_to_fp16(const at::Tensor & weight);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_flash_attention_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_flash_attention_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..3e53fb4a1b3ec90c9f5fe4fa59795a81aaa06ff8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_flash_attention_backward.h
@@ -0,0 +1,47 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_scaled_dot_product_flash_attention_backward_ops.h>
+
+namespace at {
+
+
+// aten::_scaled_dot_product_flash_attention_backward(Tensor grad_out, Tensor query, Tensor key, Tensor value, Tensor out, Tensor logsumexp, Tensor cum_seq_q, Tensor cum_seq_k, SymInt max_q, SymInt max_k, float dropout_p, bool is_causal, Tensor philox_seed, Tensor philox_offset, *, float? scale=None) -> (Tensor grad_query, Tensor grad_key, Tensor grad_value)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _scaled_dot_product_flash_attention_backward(const at::Tensor & grad_out, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const at::Tensor & out, const at::Tensor & logsumexp, const at::Tensor & cum_seq_q, const at::Tensor & cum_seq_k, int64_t max_q, int64_t max_k, double dropout_p, bool is_causal, const at::Tensor & philox_seed, const at::Tensor & philox_offset, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_flash_attention_backward::call(grad_out, query, key, value, out, logsumexp, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, philox_seed, philox_offset, scale);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _scaled_dot_product_flash_attention_backward(const at::Tensor & grad_out, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const at::Tensor & out, const at::Tensor & logsumexp, const at::Tensor & cum_seq_q, const at::Tensor & cum_seq_k, int64_t max_q, int64_t max_k, double dropout_p, bool is_causal, const at::Tensor & philox_seed, const at::Tensor & philox_offset, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_flash_attention_backward::call(grad_out, query, key, value, out, logsumexp, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, philox_seed, philox_offset, scale);
+  }
+}
+
+// aten::_scaled_dot_product_flash_attention_backward(Tensor grad_out, Tensor query, Tensor key, Tensor value, Tensor out, Tensor logsumexp, Tensor cum_seq_q, Tensor cum_seq_k, SymInt max_q, SymInt max_k, float dropout_p, bool is_causal, Tensor philox_seed, Tensor philox_offset, *, float? scale=None) -> (Tensor grad_query, Tensor grad_key, Tensor grad_value)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _scaled_dot_product_flash_attention_backward_symint(const at::Tensor & grad_out, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const at::Tensor & out, const at::Tensor & logsumexp, const at::Tensor & cum_seq_q, const at::Tensor & cum_seq_k, c10::SymInt max_q, c10::SymInt max_k, double dropout_p, bool is_causal, const at::Tensor & philox_seed, const at::Tensor & philox_offset, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_flash_attention_backward::call(grad_out, query, key, value, out, logsumexp, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, philox_seed, philox_offset, scale);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _scaled_dot_product_flash_attention_backward(const at::Tensor & grad_out, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const at::Tensor & out, const at::Tensor & logsumexp, const at::Tensor & cum_seq_q, const at::Tensor & cum_seq_k, c10::SymInt max_q, c10::SymInt max_k, double dropout_p, bool is_causal, const at::Tensor & philox_seed, const at::Tensor & philox_offset, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_flash_attention_backward::call(grad_out, query, key, value, out, logsumexp, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, philox_seed, philox_offset, scale);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_fused_attention_overrideable_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_fused_attention_overrideable_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..542fdea1d61503c7f3674ec66071374ddd9fb7e1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_fused_attention_overrideable_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _scaled_dot_product_fused_attention_overrideable {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt,at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, double, bool, bool, ::std::optional<double>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_scaled_dot_product_fused_attention_overrideable";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_scaled_dot_product_fused_attention_overrideable(Tensor query, Tensor key, Tensor value, Tensor? attn_bias=None, float dropout_p=0.0, bool is_causal=False, bool return_debug_mask=False, *, float? scale=None) -> (Tensor output, Tensor logsumexp, Tensor cum_seq_q, Tensor cum_seq_k, SymInt max_q, SymInt max_k, Tensor philox_seed, Tensor philox_offset, Tensor debug_attn_mask)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt,at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_bias, double dropout_p, bool is_causal, bool return_debug_mask, ::std::optional<double> scale);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt,at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_bias, double dropout_p, bool is_causal, bool return_debug_mask, ::std::optional<double> scale);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_shape_as_tensor.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_shape_as_tensor.h
new file mode 100644
index 0000000000000000000000000000000000000000..f929928a47096606b5d0f9f78d0fe91520372508
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_shape_as_tensor.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_shape_as_tensor_ops.h>
+
+namespace at {
+
+
+// aten::_shape_as_tensor(Tensor self) -> Tensor
+inline at::Tensor _shape_as_tensor(const at::Tensor & self) {
+    return at::_ops::_shape_as_tensor::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_shape_as_tensor_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_shape_as_tensor_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..e76e888f37427233062f7d6eb7c150b03eb4b3eb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_shape_as_tensor_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _shape_as_tensor {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_shape_as_tensor";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_shape_as_tensor(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_slow_conv2d_forward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_slow_conv2d_forward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c800e80f9f285f917babacbc1932fa999131179e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_slow_conv2d_forward_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor slow_conv2d_forward_cpu(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding);
+TORCH_API at::Tensor & slow_conv2d_forward_out_cpu(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::Tensor & output);
+TORCH_API at::Tensor slow_conv2d_forward_cuda(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding);
+TORCH_API at::Tensor & slow_conv2d_forward_out_cuda(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::Tensor & output);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_initialize_state_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_initialize_state_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..02920ab7bf0662ffddd7cdb1403ab187d255a319
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_initialize_state_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _sobol_engine_initialize_state_(at::Tensor & self, int64_t dimension);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_compressed_tensor_with_dims.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_compressed_tensor_with_dims.h
new file mode 100644
index 0000000000000000000000000000000000000000..795bd1faae2dcad68ab0237901c6b77a880be4fc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_compressed_tensor_with_dims.h
@@ -0,0 +1,34 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_sparse_compressed_tensor_with_dims_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_compressed_tensor_with_dims(int nnz, int dense_dim, int[] size, int[] blocksize, ScalarType index_dtype, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor _sparse_compressed_tensor_with_dims(int64_t nnz, int64_t dense_dim, at::IntArrayRef size, at::IntArrayRef blocksize, at::ScalarType index_dtype, at::TensorOptions options) {
+    return at::_ops::_sparse_compressed_tensor_with_dims::call(nnz, dense_dim, size, blocksize, index_dtype, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::_sparse_compressed_tensor_with_dims(int nnz, int dense_dim, int[] size, int[] blocksize, ScalarType index_dtype, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor _sparse_compressed_tensor_with_dims(int64_t nnz, int64_t dense_dim, at::IntArrayRef size, at::IntArrayRef blocksize, at::ScalarType index_dtype, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::_sparse_compressed_tensor_with_dims::call(nnz, dense_dim, size, blocksize, index_dtype, dtype, layout, device, pin_memory);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_and_tensors_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_and_tensors_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..d902380446d6b508bc56cfba878ff65aa1dd8108
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_and_tensors_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _sparse_coo_tensor_with_dims_and_tensors {
+  using schema = at::Tensor (int64_t, int64_t, c10::SymIntArrayRef, const at::Tensor &, const at::Tensor &, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sparse_coo_tensor_with_dims_and_tensors";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_sparse_coo_tensor_with_dims_and_tensors(int sparse_dim, int dense_dim, SymInt[] size, Tensor indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False, bool? is_coalesced=None) -> Tensor";
+  static at::Tensor call(int64_t sparse_dim, int64_t dense_dim, c10::SymIntArrayRef size, const at::Tensor & indices, const at::Tensor & values, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<bool> is_coalesced);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t sparse_dim, int64_t dense_dim, c10::SymIntArrayRef size, const at::Tensor & indices, const at::Tensor & values, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<bool> is_coalesced);
+};
+
+struct TORCH_API _sparse_coo_tensor_with_dims_and_tensors_out {
+  using schema = at::Tensor & (int64_t, int64_t, c10::SymIntArrayRef, const at::Tensor &, const at::Tensor &, ::std::optional<bool>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sparse_coo_tensor_with_dims_and_tensors";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_sparse_coo_tensor_with_dims_and_tensors.out(int sparse_dim, int dense_dim, SymInt[] size, Tensor indices, Tensor values, *, bool? is_coalesced=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t sparse_dim, int64_t dense_dim, c10::SymIntArrayRef size, const at::Tensor & indices, const at::Tensor & values, ::std::optional<bool> is_coalesced, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t sparse_dim, int64_t dense_dim, c10::SymIntArrayRef size, const at::Tensor & indices, const at::Tensor & values, ::std::optional<bool> is_coalesced, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..02c9debb68bc2683248e5739326880e294ef3d63
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _sparse_coo_tensor_with_dims {
+  using schema = at::Tensor (int64_t, int64_t, at::IntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sparse_coo_tensor_with_dims";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_sparse_coo_tensor_with_dims(int sparse_dim, int dense_dim, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor";
+  static at::Tensor call(int64_t sparse_dim, int64_t dense_dim, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t sparse_dim, int64_t dense_dim, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API _sparse_coo_tensor_with_dims_out {
+  using schema = at::Tensor & (int64_t, int64_t, at::IntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sparse_coo_tensor_with_dims";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_sparse_coo_tensor_with_dims.out(int sparse_dim, int dense_dim, int[] size, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t sparse_dim, int64_t dense_dim, at::IntArrayRef size, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t sparse_dim, int64_t dense_dim, at::IntArrayRef size, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_log_softmax_backward_data.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_log_softmax_backward_data.h
new file mode 100644
index 0000000000000000000000000000000000000000..41f3eed237864b269383685b38c933e5035c87fd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_log_softmax_backward_data.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_sparse_log_softmax_backward_data_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_log_softmax_backward_data(Tensor grad_output, Tensor output, int dim, Tensor self) -> Tensor
+inline at::Tensor _sparse_log_softmax_backward_data(const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, const at::Tensor & self) {
+    return at::_ops::_sparse_log_softmax_backward_data::call(grad_output, output, dim, self);
+}
+
+// aten::_sparse_log_softmax_backward_data.out(Tensor grad_output, Tensor output, int dim, Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_log_softmax_backward_data_out(at::Tensor & out, const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, const at::Tensor & self) {
+    return at::_ops::_sparse_log_softmax_backward_data_out::call(grad_output, output, dim, self, out);
+}
+// aten::_sparse_log_softmax_backward_data.out(Tensor grad_output, Tensor output, int dim, Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_log_softmax_backward_data_outf(const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::_sparse_log_softmax_backward_data_out::call(grad_output, output, dim, self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_apply_dense_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_apply_dense_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4712db3954bbe6f710a8fe1b7eae71b2a0a4ffa4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_apply_dense_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor _sparse_semi_structured_apply_dense(const at::Tensor & input, const at::Tensor & thread_masks);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_mm_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_mm_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..3554b0b65730b289c075e3c5e0b802198793df5f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_mm_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _sparse_semi_structured_mm {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, ::std::optional<at::ScalarType>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sparse_semi_structured_mm";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_sparse_semi_structured_mm(Tensor mat1, Tensor mat1_meta, Tensor mat2, *, ScalarType? out_dtype=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & mat1, const at::Tensor & mat1_meta, const at::Tensor & mat2, ::std::optional<at::ScalarType> out_dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & mat1, const at::Tensor & mat1_meta, const at::Tensor & mat2, ::std::optional<at::ScalarType> out_dtype);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_sparse_matmul_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_sparse_matmul_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ada52eef057dfebe7bf8e6c4491d8f1fad7236cb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_sparse_matmul_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & _sparse_sparse_matmul_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & _sparse_sparse_matmul_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_spsolve.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_spsolve.h
new file mode 100644
index 0000000000000000000000000000000000000000..d440ac0b68ff7442f28140c6d486c95658db93d3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_spsolve.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_spsolve_ops.h>
+
+namespace at {
+
+
+// aten::_spsolve(Tensor A, Tensor B, *, bool left=True) -> Tensor
+inline at::Tensor _spsolve(const at::Tensor & A, const at::Tensor & B, bool left=true) {
+    return at::_ops::_spsolve::call(A, B, left);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_spsolve_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_spsolve_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..25ea35916be1a93b3e6dfd307e9a328c2a2785dc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_spsolve_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _spsolve {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_spsolve";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_spsolve(Tensor A, Tensor B, *, bool left=True) -> Tensor";
+  static at::Tensor call(const at::Tensor & A, const at::Tensor & B, bool left);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, const at::Tensor & B, bool left);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_filled_intlist.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_filled_intlist.h
new file mode 100644
index 0000000000000000000000000000000000000000..37c77a156fc1452350749a4dd917d9a426035f9d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_filled_intlist.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_test_optional_filled_intlist_ops.h>
+
+namespace at {
+
+
+// aten::_test_optional_filled_intlist(Tensor values, int[2]? addends) -> Tensor
+inline at::Tensor _test_optional_filled_intlist(const at::Tensor & values, at::OptionalIntArrayRef addends) {
+    return at::_ops::_test_optional_filled_intlist::call(values, addends);
+}
+
+// aten::_test_optional_filled_intlist.out(Tensor values, int[2]? addends, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _test_optional_filled_intlist_out(at::Tensor & out, const at::Tensor & values, at::OptionalIntArrayRef addends) {
+    return at::_ops::_test_optional_filled_intlist_out::call(values, addends, out);
+}
+// aten::_test_optional_filled_intlist.out(Tensor values, int[2]? addends, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _test_optional_filled_intlist_outf(const at::Tensor & values, at::OptionalIntArrayRef addends, at::Tensor & out) {
+    return at::_ops::_test_optional_filled_intlist_out::call(values, addends, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_serialization_subcmul_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_serialization_subcmul_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..0579a1fdbf5260c85a74eb7adf70fbd17b3304f4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_serialization_subcmul_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor _test_serialization_subcmul(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_string_default_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_string_default_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..c45cb175b91fe525c43711e33391485b9ff08bdb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_string_default_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _test_string_default {
+  using schema = at::Tensor (const at::Tensor &, c10::string_view, c10::string_view);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_test_string_default";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_test_string_default(Tensor dummy, str a=\"\\\"'\\\\\", str b='\"\\'\\\\') -> Tensor";
+  static at::Tensor call(const at::Tensor & dummy, c10::string_view a, c10::string_view b);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & dummy, c10::string_view a, c10::string_view b);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_warn_in_autograd_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_warn_in_autograd_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e37de508b4b2edabfd32b89effe0b260e7c05858
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_warn_in_autograd_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor _test_warn_in_autograd(const at::Tensor & self);
+TORCH_API at::Tensor & _test_warn_in_autograd_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & _test_warn_in_autograd_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_differentiable_gru_cell_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_differentiable_gru_cell_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d11630c4c2e1248d4f758a66ceb287c0958ab9ae
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_differentiable_gru_cell_backward_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _thnn_differentiable_gru_cell_backward(const at::Tensor & grad_hy, const at::Tensor & input_gates, const at::Tensor & hidden_gates, const at::Tensor & hx, const ::std::optional<at::Tensor> & input_bias, const ::std::optional<at::Tensor> & hidden_bias);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_to_cpu_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_cpu_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9a439e44a0c2cee9bcab243f2ab7deddf6cc4c60
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_cpu_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _to_cpu {
+  using schema = ::std::vector<at::Tensor> (at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_to_cpu";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_to_cpu(Tensor[] tensors) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(at::TensorList tensors);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList tensors);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_semi_structured_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_semi_structured_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..b15f272fd2e401a22bee9dc5fe35e586684c68ba
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_semi_structured_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _to_sparse_semi_structured {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_to_sparse_semi_structured";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_to_sparse_semi_structured(Tensor dense) -> (Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & dense);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & dense);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bilinear2d_aa_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bilinear2d_aa_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..4395c5e701cc48df82011203e45d388d930a45ee
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bilinear2d_aa_ops.h
@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _upsample_bilinear2d_aa_vec {
+  using schema = at::Tensor (const at::Tensor &, at::OptionalSymIntArrayRef, bool, ::std::optional<at::ArrayRef<double>>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_upsample_bilinear2d_aa";
+  static constexpr const char* overload_name = "vec";
+  static constexpr const char* schema_str = "_upsample_bilinear2d_aa.vec(Tensor input, SymInt[]? output_size, bool align_corners, float[]? scale_factors) -> Tensor";
+  static at::Tensor call(const at::Tensor & input, at::OptionalSymIntArrayRef output_size, bool align_corners, ::std::optional<at::ArrayRef<double>> scale_factors);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, at::OptionalSymIntArrayRef output_size, bool align_corners, ::std::optional<at::ArrayRef<double>> scale_factors);
+};
+
+struct TORCH_API _upsample_bilinear2d_aa_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, bool, ::std::optional<double>, ::std::optional<double>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_upsample_bilinear2d_aa";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_upsample_bilinear2d_aa.out(Tensor self, SymInt[2] output_size, bool align_corners, float? scales_h=None, float? scales_w=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+};
+
+struct TORCH_API _upsample_bilinear2d_aa {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, bool, ::std::optional<double>, ::std::optional<double>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_upsample_bilinear2d_aa";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_upsample_bilinear2d_aa(Tensor self, SymInt[2] output_size, bool align_corners, float? scales_h=None, float? scales_w=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..a2525692d7a49f7d877fbefc81d74cda23951323
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_backward_native.h
@@ -0,0 +1,26 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/_upsample_nearest_exact1d_backward_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured__upsample_nearest_exact1d_backward_out_cpu : public at::meta::structured__upsample_nearest_exact1d_backward {
+void impl(const at::Tensor & grad_output, at::ArrayRef<int64_t> output_size, at::ArrayRef<int64_t> input_size, ::std::optional<double> scales, const at::Tensor & grad_input);
+};
+struct TORCH_API structured__upsample_nearest_exact1d_backward_out_cuda : public at::meta::structured__upsample_nearest_exact1d_backward {
+void impl(const at::Tensor & grad_output, at::ArrayRef<int64_t> output_size, at::ArrayRef<int64_t> input_size, ::std::optional<double> scales, const at::Tensor & grad_input);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d3f82b0ded6cff40a9f01a84d39f2a9c8b7ba9db
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_meta_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor _upsample_nearest_exact1d(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor _upsample_nearest_exact1d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor & _upsample_nearest_exact1d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor & _upsample_nearest_exact1d_outf(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales, at::Tensor & out);
+TORCH_API at::Tensor & _upsample_nearest_exact1d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor & _upsample_nearest_exact1d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact2d_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact2d_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..461a64944296dd35e500e979ba04cce9c8cb3fd3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact2d_backward_native.h
@@ -0,0 +1,26 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/_upsample_nearest_exact2d_backward_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured__upsample_nearest_exact2d_backward_out_cpu : public at::meta::structured__upsample_nearest_exact2d_backward {
+void impl(const at::Tensor & grad_output, at::ArrayRef<int64_t> output_size, at::ArrayRef<int64_t> input_size, ::std::optional<double> scales_h, ::std::optional<double> scales_w, const at::Tensor & grad_input);
+};
+struct TORCH_API structured__upsample_nearest_exact2d_backward_out_cuda : public at::meta::structured__upsample_nearest_exact2d_backward {
+void impl(const at::Tensor & grad_output, at::ArrayRef<int64_t> output_size, at::ArrayRef<int64_t> input_size, ::std::optional<double> scales_h, ::std::optional<double> scales_w, const at::Tensor & grad_input);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_compressed_sparse_indices_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_compressed_sparse_indices_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..97a07a59888e7fe4d2e02c72980a92249d7b6b03
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_compressed_sparse_indices_cpu_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API void _validate_compressed_sparse_indices(bool is_crow, const at::Tensor & compressed_idx, const at::Tensor & plain_idx, int64_t cdim, int64_t dim, int64_t nnz);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_compressed_tensor_args_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_compressed_tensor_args_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..75c35a9462fcba4d794410aeaa33419b5d3b344b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_compressed_tensor_args_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API void _validate_sparse_compressed_tensor_args(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, at::IntArrayRef size, at::Layout layout);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_coo_tensor_args_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_coo_tensor_args_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fba4818cfca3f686bdd60260e2ec75b7a3ba7f28
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_coo_tensor_args_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API void _validate_sparse_coo_tensor_args(const at::Tensor & indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<bool> is_coalesced=::std::nullopt);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_csr_tensor_args_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_csr_tensor_args_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ead2abe9411a1170c83a6f8d5b646e8aebfbc757
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_csr_tensor_args_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API void _validate_sparse_csr_tensor_args(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, at::IntArrayRef size);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_int8pack_mm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_int8pack_mm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..bdd37fa2c2a89ca9b9c3ecba89144e3e8557fa80
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_int8pack_mm_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor _weight_int8pack_mm_cpu(const at::Tensor & self, const at::Tensor & mat2, const at::Tensor & scales);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4d0f12a3d040af148407c30630901ca4f5742ad4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_backward_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> _weight_norm_interface_backward(const at::Tensor & grad_w, const at::Tensor & saved_v, const at::Tensor & saved_g, const at::Tensor & saved_norms, int64_t dim);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/absolute_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/absolute_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..ae419587fa6f6cddb953346e2b2bf33b7a891357
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/absolute_ops.h
@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API absolute {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::absolute";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "absolute(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+struct TORCH_API absolute_ {
+  using schema = at::Tensor & (at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::absolute_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "absolute_(Tensor(a!) self) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self);
+};
+
+struct TORCH_API absolute_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::absolute";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "absolute.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool2d_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool2d_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..93b04891e650fa468be661ac36820ce3caa9f141
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool2d_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor adaptive_avg_pool2d(const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API at::Tensor adaptive_avg_pool2d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool3d_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool3d_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..834df66dab5af9edc168aac4e6315f1f75ef0336
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool3d_cpu_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor & adaptive_avg_pool3d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API at::Tensor & adaptive_avg_pool3d_outf(const at::Tensor & self, at::IntArrayRef output_size, at::Tensor & out);
+TORCH_API at::Tensor & adaptive_avg_pool3d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size);
+TORCH_API at::Tensor & adaptive_avg_pool3d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fb791ade54edefa89b33384498c10e2f819f80d5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> adaptive_max_pool3d(const at::Tensor & self, at::IntArrayRef output_size);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b44a051c075db291286ad9e291f67ad7d092fc84
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> adaptive_max_pool3d(const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> adaptive_max_pool3d_out(at::Tensor & out, at::Tensor & indices, const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> adaptive_max_pool3d_outf(const at::Tensor & self, at::IntArrayRef output_size, at::Tensor & out, at::Tensor & indices);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/add_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/add_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a8bd8b68cdc44cfe9e07653eb56cbba81c3f0021
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/add_cpu_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor add(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & add_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & add_outf(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha, at::Tensor & out);
+TORCH_API at::Tensor & add_(at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul.h
new file mode 100644
index 0000000000000000000000000000000000000000..da7ea2feb4a351433132ff8d876453b9cbde7761
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/addcmul_ops.h>
+
+namespace at {
+
+
+// aten::addcmul.out(Tensor self, Tensor tensor1, Tensor tensor2, *, Scalar value=1, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & addcmul_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value=1) {
+    return at::_ops::addcmul_out::call(self, tensor1, tensor2, value, out);
+}
+// aten::addcmul.out(Tensor self, Tensor tensor1, Tensor tensor2, *, Scalar value=1, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & addcmul_outf(const at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value, at::Tensor & out) {
+    return at::_ops::addcmul_out::call(self, tensor1, tensor2, value, out);
+}
+
+// aten::addcmul(Tensor self, Tensor tensor1, Tensor tensor2, *, Scalar value=1) -> Tensor
+inline at::Tensor addcmul(const at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value=1) {
+    return at::_ops::addcmul::call(self, tensor1, tensor2, value);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addmv_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addmv_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c3e680d43e86bebf389220b8ce84559c9fd7e904
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addmv_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor addmv(const at::Tensor & self, const at::Tensor & mat, const at::Tensor & vec, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & addmv_(at::Tensor & self, const at::Tensor & mat, const at::Tensor & vec, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/align_tensors_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/align_tensors_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ffce8f8fa2733cd380611b0fe3d5333d61e90f0e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/align_tensors_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API ::std::vector<at::Tensor> align_tensors(at::TensorList tensors);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/all_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/all_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5062377415a33ba2d9a2bccbf10f104662415123
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/all_meta_dispatch.h
@@ -0,0 +1,31 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor all(const at::Tensor & self, int64_t dim, bool keepdim=false);
+TORCH_API at::Tensor & all_out(at::Tensor & out, const at::Tensor & self, int64_t dim, bool keepdim=false);
+TORCH_API at::Tensor & all_outf(const at::Tensor & self, int64_t dim, bool keepdim, at::Tensor & out);
+TORCH_API at::Tensor all(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim=false);
+TORCH_API at::Tensor & all_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim=false);
+TORCH_API at::Tensor & all_outf(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim, at::Tensor & out);
+TORCH_API at::Tensor all(const at::Tensor & self);
+TORCH_API at::Tensor & all_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & all_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/allclose_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/allclose_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..3b302d9213645149b15743f635107364d485e976
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/allclose_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API bool allclose(const at::Tensor & self, const at::Tensor & other, double rtol=1e-05, double atol=1e-08, bool equal_nan=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arccosh_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arccosh_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1827d44cd2e3c3997b01e82f2e50321bc88a6955
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arccosh_ops.h
@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API arccosh {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::arccosh";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "arccosh(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+struct TORCH_API arccosh_ {
+  using schema = at::Tensor & (at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::arccosh_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "arccosh_(Tensor(a!) self) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self);
+};
+
+struct TORCH_API arccosh_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::arccosh";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "arccosh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arctan2.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arctan2.h
new file mode 100644
index 0000000000000000000000000000000000000000..714907d4f909d2f83f708b39acedb09399638805
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arctan2.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/arctan2_ops.h>
+
+namespace at {
+
+
+// aten::arctan2(Tensor self, Tensor other) -> Tensor
+inline at::Tensor arctan2(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::arctan2::call(self, other);
+}
+
+// aten::arctan2.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arctan2_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::arctan2_out::call(self, other, out);
+}
+// aten::arctan2.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arctan2_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::arctan2_out::call(self, other, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arctan_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arctan_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..80b5fb5777ca3aa6d00707769729e5158b4d27de
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arctan_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor arctan(const at::Tensor & self);
+TORCH_API at::Tensor & arctan_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & arctan_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arctanh.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arctanh.h
new file mode 100644
index 0000000000000000000000000000000000000000..6e0e59442c2a874fd0fbbe6da7b48125c418ed7b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arctanh.h
@@ -0,0 +1,44 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/arctanh_ops.h>
+
+namespace at {
+
+
+// aten::arctanh(Tensor self) -> Tensor
+inline at::Tensor arctanh(const at::Tensor & self) {
+    return at::_ops::arctanh::call(self);
+}
+
+// aten::arctanh_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & arctanh_(at::Tensor & self) {
+    return at::_ops::arctanh_::call(self);
+}
+
+// aten::arctanh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arctanh_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::arctanh_out::call(self, out);
+}
+// aten::arctanh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arctanh_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::arctanh_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/argmin.h b/lib/python3.12/site-packages/torch/include/ATen/ops/argmin.h
new file mode 100644
index 0000000000000000000000000000000000000000..589f902c59bcf82bcbe12eeb9f4a52c3c39356ab
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/argmin.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/argmin_ops.h>
+
+namespace at {
+
+
+// aten::argmin(Tensor self, int? dim=None, bool keepdim=False) -> Tensor
+inline at::Tensor argmin(const at::Tensor & self, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false) {
+    return at::_ops::argmin::call(self, dim, keepdim);
+}
+
+// aten::argmin.out(Tensor self, int? dim=None, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & argmin_out(at::Tensor & out, const at::Tensor & self, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false) {
+    return at::_ops::argmin_out::call(self, dim, keepdim, out);
+}
+// aten::argmin.out(Tensor self, int? dim=None, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & argmin_outf(const at::Tensor & self, ::std::optional<int64_t> dim, bool keepdim, at::Tensor & out) {
+    return at::_ops::argmin_out::call(self, dim, keepdim, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/asin_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/asin_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..70012b262494c2d6edf2f941592a90a5aa572105
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/asin_cpu_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor asin(const at::Tensor & self);
+TORCH_API at::Tensor & asin_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & asin_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & asin_(at::Tensor & self);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/atanh_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/atanh_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6fe4d216cdb3e2014fb950d3197472900ce3fa3a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/atanh_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor atanh(const at::Tensor & self);
+TORCH_API at::Tensor & atanh_(at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool3d_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool3d_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f6b85ae97730dd8bd7d4ca32abd59f49cec82c67
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool3d_backward_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor avg_pool3d_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override);
+TORCH_API at::Tensor & avg_pool3d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override);
+TORCH_API at::Tensor & avg_pool3d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override, at::Tensor & grad_input);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bartlett_window_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bartlett_window_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..fece24d6fc1ea1477e4656fdca252909dbfc2b96
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bartlett_window_ops.h
@@ -0,0 +1,61 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API bartlett_window {
+  using schema = at::Tensor (int64_t, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::bartlett_window";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "bartlett_window(int window_length, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(int64_t window_length, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API bartlett_window_periodic {
+  using schema = at::Tensor (int64_t, bool, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::bartlett_window";
+  static constexpr const char* overload_name = "periodic";
+  static constexpr const char* schema_str = "bartlett_window.periodic(int window_length, bool periodic, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(int64_t window_length, bool periodic, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, bool periodic, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API bartlett_window_out {
+  using schema = at::Tensor & (int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::bartlett_window";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "bartlett_window.out(int window_length, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t window_length, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, at::Tensor & out);
+};
+
+struct TORCH_API bartlett_window_periodic_out {
+  using schema = at::Tensor & (int64_t, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::bartlett_window";
+  static constexpr const char* overload_name = "periodic_out";
+  static constexpr const char* schema_str = "bartlett_window.periodic_out(int window_length, bool periodic, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t window_length, bool periodic, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, bool periodic, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_gather_stats.h b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_gather_stats.h
new file mode 100644
index 0000000000000000000000000000000000000000..3d9f9ef5759d1658dc509da7f7dd8387a2dffc7f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_gather_stats.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/batch_norm_gather_stats_ops.h>
+
+namespace at {
+
+
+// aten::batch_norm_gather_stats(Tensor input, Tensor mean, Tensor invstd, Tensor? running_mean, Tensor? running_var, float momentum, float eps, int count) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> batch_norm_gather_stats(const at::Tensor & input, const at::Tensor & mean, const at::Tensor & invstd, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, double momentum, double eps, int64_t count) {
+    return at::_ops::batch_norm_gather_stats::call(input, mean, invstd, running_mean, running_var, momentum, eps, count);
+}
+
+// aten::batch_norm_gather_stats.out(Tensor input, Tensor mean, Tensor invstd, Tensor? running_mean, Tensor? running_var, float momentum, float eps, int count, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> batch_norm_gather_stats_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & input, const at::Tensor & mean, const at::Tensor & invstd, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, double momentum, double eps, int64_t count) {
+    return at::_ops::batch_norm_gather_stats_out::call(input, mean, invstd, running_mean, running_var, momentum, eps, count, out0, out1);
+}
+// aten::batch_norm_gather_stats.out(Tensor input, Tensor mean, Tensor invstd, Tensor? running_mean, Tensor? running_var, float momentum, float eps, int count, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> batch_norm_gather_stats_outf(const at::Tensor & input, const at::Tensor & mean, const at::Tensor & invstd, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, double momentum, double eps, int64_t count, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::batch_norm_gather_stats_out::call(input, mean, invstd, running_mean, running_var, momentum, eps, count, out0, out1);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_gather_stats_with_counts.h b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_gather_stats_with_counts.h
new file mode 100644
index 0000000000000000000000000000000000000000..c0fa67ec24175e12a48e5aa5de750c47c413bedd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_gather_stats_with_counts.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/batch_norm_gather_stats_with_counts_ops.h>
+
+namespace at {
+
+
+// aten::batch_norm_gather_stats_with_counts(Tensor input, Tensor mean, Tensor invstd, Tensor? running_mean, Tensor? running_var, float momentum, float eps, Tensor counts) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> batch_norm_gather_stats_with_counts(const at::Tensor & input, const at::Tensor & mean, const at::Tensor & invstd, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, double momentum, double eps, const at::Tensor & counts) {
+    return at::_ops::batch_norm_gather_stats_with_counts::call(input, mean, invstd, running_mean, running_var, momentum, eps, counts);
+}
+
+// aten::batch_norm_gather_stats_with_counts.out(Tensor input, Tensor mean, Tensor invstd, Tensor? running_mean, Tensor? running_var, float momentum, float eps, Tensor counts, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> batch_norm_gather_stats_with_counts_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & input, const at::Tensor & mean, const at::Tensor & invstd, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, double momentum, double eps, const at::Tensor & counts) {
+    return at::_ops::batch_norm_gather_stats_with_counts_out::call(input, mean, invstd, running_mean, running_var, momentum, eps, counts, out0, out1);
+}
+// aten::batch_norm_gather_stats_with_counts.out(Tensor input, Tensor mean, Tensor invstd, Tensor? running_mean, Tensor? running_var, float momentum, float eps, Tensor counts, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> batch_norm_gather_stats_with_counts_outf(const at::Tensor & input, const at::Tensor & mean, const at::Tensor & invstd, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, double momentum, double eps, const at::Tensor & counts, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::batch_norm_gather_stats_with_counts_out::call(input, mean, invstd, running_mean, running_var, momentum, eps, counts, out0, out1);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_stats_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_stats_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a39dafbb2673a5be368f3ec61f3ea2b41ffa74dc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_stats_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API batch_norm_stats {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, double);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::batch_norm_stats";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "batch_norm_stats(Tensor input, float eps) -> (Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & input, double eps);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, double eps);
+};
+
+struct TORCH_API batch_norm_stats_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const at::Tensor &, double, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::batch_norm_stats";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "batch_norm_stats.out(Tensor input, float eps, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & input, double eps, at::Tensor & out0, at::Tensor & out1);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, double eps, at::Tensor & out0, at::Tensor & out1);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_update_stats_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_update_stats_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fc796bcf9105192c602ca3b73bcfece0bdc668aa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_update_stats_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> batch_norm_update_stats(const at::Tensor & input, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, double momentum);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/binary_cross_entropy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/binary_cross_entropy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..541c3aefcaa4b33617e419e5475358885411993b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/binary_cross_entropy_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor binary_cross_entropy_cpu(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean);
+TORCH_API at::Tensor & binary_cross_entropy_out_cpu(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, at::Tensor & out);
+TORCH_API at::Tensor binary_cross_entropy_cuda(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean);
+TORCH_API at::Tensor & binary_cross_entropy_out_cuda(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bincount_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bincount_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f1bfbbd45a40b96d6c5db4b83cc9acd67927f30c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bincount_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API bincount {
+  using schema = at::Tensor (const at::Tensor &, const ::std::optional<at::Tensor> &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::bincount";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "bincount(Tensor self, Tensor? weights=None, int minlength=0) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const ::std::optional<at::Tensor> & weights, int64_t minlength);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const ::std::optional<at::Tensor> & weights, int64_t minlength);
+};
+
+struct TORCH_API bincount_out {
+  using schema = at::Tensor & (const at::Tensor &, const ::std::optional<at::Tensor> &, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::bincount";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "bincount.out(Tensor self, Tensor? weights=None, int minlength=0, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const ::std::optional<at::Tensor> & weights, int64_t minlength, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const ::std::optional<at::Tensor> & weights, int64_t minlength, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_and_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_and_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..2cdb117a6dfa8ef515d349fd61dca62f0bd9b831
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_and_native.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/bitwise_and_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_bitwise_and_out : public at::meta::structured_bitwise_and_Tensor {
+void impl(const at::Tensor & self, const at::Tensor & other, const at::Tensor & out);
+};
+TORCH_API at::Tensor bitwise_and(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & bitwise_and_out(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & bitwise_and_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor bitwise_and(const at::Scalar & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_and_Scalar_Tensor_out(const at::Scalar & self, const at::Tensor & other, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_left_shift.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_left_shift.h
new file mode 100644
index 0000000000000000000000000000000000000000..ca17861e3ca247f0436f1819e464407d8f7d9770
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_left_shift.h
@@ -0,0 +1,67 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/bitwise_left_shift_ops.h>
+
+namespace at {
+
+
+// aten::bitwise_left_shift.Tensor(Tensor self, Tensor other) -> Tensor
+inline at::Tensor bitwise_left_shift(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::bitwise_left_shift_Tensor::call(self, other);
+}
+
+// aten::bitwise_left_shift.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & bitwise_left_shift_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::bitwise_left_shift_Tensor_out::call(self, other, out);
+}
+// aten::bitwise_left_shift.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & bitwise_left_shift_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::bitwise_left_shift_Tensor_out::call(self, other, out);
+}
+
+// aten::bitwise_left_shift.Tensor_Scalar(Tensor self, Scalar other) -> Tensor
+inline at::Tensor bitwise_left_shift(const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::bitwise_left_shift_Tensor_Scalar::call(self, other);
+}
+
+// aten::bitwise_left_shift.Tensor_Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & bitwise_left_shift_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::bitwise_left_shift_Tensor_Scalar_out::call(self, other, out);
+}
+// aten::bitwise_left_shift.Tensor_Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & bitwise_left_shift_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out) {
+    return at::_ops::bitwise_left_shift_Tensor_Scalar_out::call(self, other, out);
+}
+
+// aten::bitwise_left_shift.Scalar_Tensor(Scalar self, Tensor other) -> Tensor
+inline at::Tensor bitwise_left_shift(const at::Scalar & self, const at::Tensor & other) {
+    return at::_ops::bitwise_left_shift_Scalar_Tensor::call(self, other);
+}
+
+// aten::bitwise_left_shift.Scalar_Tensor_out(Scalar self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & bitwise_left_shift_out(at::Tensor & out, const at::Scalar & self, const at::Tensor & other) {
+    return at::_ops::bitwise_left_shift_Scalar_Tensor_out::call(self, other, out);
+}
+// aten::bitwise_left_shift.Scalar_Tensor_out(Scalar self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & bitwise_left_shift_outf(const at::Scalar & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::bitwise_left_shift_Scalar_Tensor_out::call(self, other, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_left_shift_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_left_shift_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..4609f5a6f6f36aa89233ff65d8e5fdfc381e5488
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_left_shift_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_bitwise_left_shift_Tensor : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Tensor & other);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d580df33865f7a34ac9a1048466bd04b19fd0083
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_native.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/bitwise_right_shift_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_bitwise_right_shift_out : public at::meta::structured_bitwise_right_shift_Tensor {
+void impl(const at::Tensor & self, const at::Tensor & other, const at::Tensor & out);
+};
+TORCH_API at::Tensor bitwise_right_shift(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & bitwise_right_shift_out(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & bitwise_right_shift_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor bitwise_right_shift(const at::Scalar & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_right_shift_Scalar_Tensor_out(const at::Scalar & self, const at::Tensor & other, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_xor_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_xor_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..dc2b899f572ad76950d09fe9b8990e14f903c34e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_xor_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor bitwise_xor(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_xor_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/broadcast_to.h b/lib/python3.12/site-packages/torch/include/ATen/ops/broadcast_to.h
new file mode 100644
index 0000000000000000000000000000000000000000..594be8fdfafc288564eb4e121312eb8dbf9d9952
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/broadcast_to.h
@@ -0,0 +1,47 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/broadcast_to_ops.h>
+
+namespace at {
+
+
+// aten::broadcast_to(Tensor(a) self, SymInt[] size) -> Tensor(a)
+inline at::Tensor broadcast_to(const at::Tensor & self, at::IntArrayRef size) {
+    return at::_ops::broadcast_to::call(self, c10::fromIntArrayRefSlow(size));
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor broadcast_to(const at::Tensor & self, at::IntArrayRef size) {
+    return at::_ops::broadcast_to::call(self, c10::fromIntArrayRefSlow(size));
+  }
+}
+
+// aten::broadcast_to(Tensor(a) self, SymInt[] size) -> Tensor(a)
+inline at::Tensor broadcast_to_symint(const at::Tensor & self, c10::SymIntArrayRef size) {
+    return at::_ops::broadcast_to::call(self, size);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor broadcast_to(const at::Tensor & self, c10::SymIntArrayRef size) {
+    return at::_ops::broadcast_to::call(self, size);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cat.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cat.h
new file mode 100644
index 0000000000000000000000000000000000000000..9c350d18271fea8be910ac47296037de9eb6ff64
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cat.h
@@ -0,0 +1,53 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/cat_ops.h>
+
+namespace at {
+
+
+// aten::cat(Tensor[] tensors, int dim=0) -> Tensor
+inline at::Tensor cat(const at::ITensorListRef & tensors, int64_t dim=0) {
+    return at::_ops::cat::call(tensors, dim);
+}
+
+// aten::cat.out(Tensor[] tensors, int dim=0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cat_out(at::Tensor & out, const at::ITensorListRef & tensors, int64_t dim=0) {
+    return at::_ops::cat_out::call(tensors, dim, out);
+}
+// aten::cat.out(Tensor[] tensors, int dim=0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cat_outf(const at::ITensorListRef & tensors, int64_t dim, at::Tensor & out) {
+    return at::_ops::cat_out::call(tensors, dim, out);
+}
+
+// aten::cat.names(Tensor[] tensors, Dimname dim) -> Tensor
+inline at::Tensor cat(at::TensorList tensors, at::Dimname dim) {
+    return at::_ops::cat_names::call(tensors, dim);
+}
+
+// aten::cat.names_out(Tensor[] tensors, Dimname dim, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cat_out(at::Tensor & out, at::TensorList tensors, at::Dimname dim) {
+    return at::_ops::cat_names_out::call(tensors, dim, out);
+}
+// aten::cat.names_out(Tensor[] tensors, Dimname dim, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cat_outf(at::TensorList tensors, at::Dimname dim, at::Tensor & out) {
+    return at::_ops::cat_names_out::call(tensors, dim, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cat_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cat_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..63b67a9c371ec090619e4ab8970d924a784b140e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cat_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor cat(const at::ITensorListRef & tensors, int64_t dim=0);
+TORCH_API at::Tensor & cat_out(at::Tensor & out, const at::ITensorListRef & tensors, int64_t dim=0);
+TORCH_API at::Tensor & cat_outf(const at::ITensorListRef & tensors, int64_t dim, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/ceil_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/ceil_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0f4613a6859dc0e3752f3cbece5267b16bbee823
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/ceil_cuda_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor ceil(const at::Tensor & self);
+TORCH_API at::Tensor & ceil_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & ceil_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & ceil_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cholesky_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cholesky_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9fc4959fa5491561f1955ceed7471dd36ccef9ac
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cholesky_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor cholesky(const at::Tensor & self, bool upper=false);
+TORCH_API at::Tensor & cholesky_out(at::Tensor & out, const at::Tensor & self, bool upper=false);
+TORCH_API at::Tensor & cholesky_outf(const at::Tensor & self, bool upper, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/chunk_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/chunk_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ee112f074ba1b4a8cf1c095ee47d59636d35facd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/chunk_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API ::std::vector<at::Tensor> chunk(const at::Tensor & self, int64_t chunks, int64_t dim=0);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/clip_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/clip_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1f079b33a98dd7afe1957f345550ff6b220d3e3b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/clip_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,30 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor clip(const at::Tensor & self, const ::std::optional<at::Scalar> & min, const ::std::optional<at::Scalar> & max=::std::nullopt);
+TORCH_API at::Tensor & clip_out(at::Tensor & out, const at::Tensor & self, const ::std::optional<at::Scalar> & min, const ::std::optional<at::Scalar> & max=::std::nullopt);
+TORCH_API at::Tensor & clip_outf(const at::Tensor & self, const ::std::optional<at::Scalar> & min, const ::std::optional<at::Scalar> & max, at::Tensor & out);
+TORCH_API at::Tensor & clip_(at::Tensor & self, const ::std::optional<at::Scalar> & min, const ::std::optional<at::Scalar> & max=::std::nullopt);
+TORCH_API at::Tensor clip(const at::Tensor & self, const ::std::optional<at::Tensor> & min={}, const ::std::optional<at::Tensor> & max={});
+TORCH_API at::Tensor & clip_out(at::Tensor & out, const at::Tensor & self, const ::std::optional<at::Tensor> & min={}, const ::std::optional<at::Tensor> & max={});
+TORCH_API at::Tensor & clip_outf(const at::Tensor & self, const ::std::optional<at::Tensor> & min, const ::std::optional<at::Tensor> & max, at::Tensor & out);
+TORCH_API at::Tensor & clip_(at::Tensor & self, const ::std::optional<at::Tensor> & min={}, const ::std::optional<at::Tensor> & max={});
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..167d26c6e6977fa00f391fde0392a06e5a8e2d41
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices_copy.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/col_indices_copy_ops.h>
+
+namespace at {
+
+
+// aten::col_indices_copy(Tensor self) -> Tensor
+inline at::Tensor col_indices_copy(const at::Tensor & self) {
+    return at::_ops::col_indices_copy::call(self);
+}
+
+// aten::col_indices_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & col_indices_copy_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::col_indices_copy_out::call(self, out);
+}
+// aten::col_indices_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & col_indices_copy_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::col_indices_copy_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices_copy_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices_copy_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..46984f03b8fbf037661b4ce983e0f052a35ce357
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices_copy_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API col_indices_copy {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::col_indices_copy";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "col_indices_copy(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+struct TORCH_API col_indices_copy_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::col_indices_copy";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "col_indices_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/conj_physical_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/conj_physical_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..617bc805032a1dc9e83f784aaf0cc7117b967bd7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/conj_physical_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor conj_physical(const at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/conv3d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/conv3d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..53a7e21142c83275eab25e2d4e9b851bef17bf71
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/conv3d_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor conv3d_symint(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef padding=c10::SymInt(0), c10::SymIntArrayRef dilation=c10::SymInt(1), c10::SymInt groups=1);
+TORCH_API at::Tensor conv3d_padding_symint(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::string_view padding="valid", c10::SymIntArrayRef dilation=c10::SymInt(1), c10::SymInt groups=1);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_backward_overrideable_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_backward_overrideable_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..12dce814d45553f1f1c2b34c8489f5a28ec9268f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_backward_overrideable_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> convolution_backward_overrideable(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & weight, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> convolution_backward_overrideable_symint(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & weight, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> convolution_backward_overrideable_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & weight, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> convolution_backward_overrideable_outf(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & weight, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> convolution_backward_overrideable_symint_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & weight, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> convolution_backward_overrideable_symint_outf(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & weight, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_overrideable_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_overrideable_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4756bdf974e8113dd7f68633f755c9c4fe525401
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_overrideable_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor convolution_overrideable(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups);
+TORCH_API at::Tensor convolution_overrideable_symint(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups);
+TORCH_API at::Tensor & convolution_overrideable_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups);
+TORCH_API at::Tensor & convolution_overrideable_outf(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups, at::Tensor & out);
+TORCH_API at::Tensor & convolution_overrideable_symint_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups);
+TORCH_API at::Tensor & convolution_overrideable_symint_outf(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/copy_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/copy_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6d7715af79952163406a5cd244df28fe0a883de6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/copy_meta_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor copy(const at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/corrcoef.h b/lib/python3.12/site-packages/torch/include/ATen/ops/corrcoef.h
new file mode 100644
index 0000000000000000000000000000000000000000..c6fa6fa7a7369890ba2204f6d112a02c708dd809
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/corrcoef.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/corrcoef_ops.h>
+
+namespace at {
+
+
+// aten::corrcoef(Tensor self) -> Tensor
+inline at::Tensor corrcoef(const at::Tensor & self) {
+    return at::_ops::corrcoef::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cosine_embedding_loss.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cosine_embedding_loss.h
new file mode 100644
index 0000000000000000000000000000000000000000..b9aa65c5259f11f7327cf64652626d35116a4a82
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cosine_embedding_loss.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/cosine_embedding_loss_ops.h>
+
+namespace at {
+
+
+// aten::cosine_embedding_loss(Tensor input1, Tensor input2, Tensor target, float margin=0.0, int reduction=Mean) -> Tensor
+inline at::Tensor cosine_embedding_loss(const at::Tensor & input1, const at::Tensor & input2, const at::Tensor & target, double margin=0.0, int64_t reduction=at::Reduction::Mean) {
+    return at::_ops::cosine_embedding_loss::call(input1, input2, target, margin, reduction);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cosine_embedding_loss_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cosine_embedding_loss_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..4720c35729d91f0f4f3d91c4f68bffda1320b247
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cosine_embedding_loss_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API cosine_embedding_loss {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, double, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::cosine_embedding_loss";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "cosine_embedding_loss(Tensor input1, Tensor input2, Tensor target, float margin=0.0, int reduction=Mean) -> Tensor";
+  static at::Tensor call(const at::Tensor & input1, const at::Tensor & input2, const at::Tensor & target, double margin, int64_t reduction);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input1, const at::Tensor & input2, const at::Tensor & target, double margin, int64_t reduction);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/crow_indices_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/crow_indices_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..65bdd8d370c934d7cda9abb72a357185c4c40c5d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/crow_indices_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor crow_indices_default(const at::Tensor & self);
+TORCH_API at::Tensor crow_indices_sparse_csr(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/crow_indices_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/crow_indices_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..46627bd15b39df8e0f0f1c45390c6f684fddfb60
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/crow_indices_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API crow_indices {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::crow_indices";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "crow_indices(Tensor(a) self) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_affine_grid_generator_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_affine_grid_generator_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3b868da7cf29b29f069787b2228af0550b8129a0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_affine_grid_generator_backward_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor cudnn_affine_grid_generator_backward(const at::Tensor & grad, int64_t N, int64_t C, int64_t H, int64_t W);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_batch_norm_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_batch_norm_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e78c94362ff6c927157f37d3f0dcf945e3d9c273
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_batch_norm_backward_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> cudnn_batch_norm_backward(const at::Tensor & input, const at::Tensor & grad_output, const at::Tensor & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_var, double epsilon, const at::Tensor & reserveSpace);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution.h
new file mode 100644
index 0000000000000000000000000000000000000000..8005d13e54159d21b20febf4bb1914865ee6dacc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/cudnn_convolution_ops.h>
+
+namespace at {
+
+
+// aten::cudnn_convolution(Tensor self, Tensor weight, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, bool allow_tf32) -> Tensor
+inline at::Tensor cudnn_convolution(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, bool allow_tf32) {
+    return at::_ops::cudnn_convolution::call(self, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic, allow_tf32);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor cudnn_convolution(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, bool allow_tf32) {
+    return at::_ops::cudnn_convolution::call(self, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic, allow_tf32);
+  }
+}
+
+// aten::cudnn_convolution(Tensor self, Tensor weight, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, bool allow_tf32) -> Tensor
+inline at::Tensor cudnn_convolution_symint(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32) {
+    return at::_ops::cudnn_convolution::call(self, weight, padding, stride, dilation, groups, benchmark, deterministic, allow_tf32);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor cudnn_convolution(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32) {
+    return at::_ops::cudnn_convolution::call(self, weight, padding, stride, dilation, groups, benchmark, deterministic, allow_tf32);
+  }
+}
+
+// aten::cudnn_convolution.out(Tensor self, Tensor weight, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, bool allow_tf32, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cudnn_convolution_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, bool allow_tf32) {
+    return at::_ops::cudnn_convolution_out::call(self, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic, allow_tf32, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & cudnn_convolution_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, bool allow_tf32) {
+    return at::_ops::cudnn_convolution_out::call(self, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic, allow_tf32, out);
+  }
+}
+
+// aten::cudnn_convolution.out(Tensor self, Tensor weight, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, bool allow_tf32, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cudnn_convolution_outf(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, bool allow_tf32, at::Tensor & out) {
+    return at::_ops::cudnn_convolution_out::call(self, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic, allow_tf32, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & cudnn_convolution_outf(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, bool allow_tf32, at::Tensor & out) {
+    return at::_ops::cudnn_convolution_out::call(self, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic, allow_tf32, out);
+  }
+}
+
+// aten::cudnn_convolution.out(Tensor self, Tensor weight, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, bool allow_tf32, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cudnn_convolution_symint_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32) {
+    return at::_ops::cudnn_convolution_out::call(self, weight, padding, stride, dilation, groups, benchmark, deterministic, allow_tf32, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & cudnn_convolution_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32) {
+    return at::_ops::cudnn_convolution_out::call(self, weight, padding, stride, dilation, groups, benchmark, deterministic, allow_tf32, out);
+  }
+}
+
+// aten::cudnn_convolution.out(Tensor self, Tensor weight, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, bool allow_tf32, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cudnn_convolution_symint_outf(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32, at::Tensor & out) {
+    return at::_ops::cudnn_convolution_out::call(self, weight, padding, stride, dilation, groups, benchmark, deterministic, allow_tf32, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & cudnn_convolution_outf(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32, at::Tensor & out) {
+    return at::_ops::cudnn_convolution_out::call(self, weight, padding, stride, dilation, groups, benchmark, deterministic, allow_tf32, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_relu.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_relu.h
new file mode 100644
index 0000000000000000000000000000000000000000..353ac734de6c7df83763d8cbd40d0fc3a5db88cc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_relu.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/cudnn_convolution_relu_ops.h>
+
+namespace at {
+
+
+// aten::cudnn_convolution_relu(Tensor self, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, SymInt groups) -> Tensor
+inline at::Tensor cudnn_convolution_relu(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, int64_t groups) {
+    return at::_ops::cudnn_convolution_relu::call(self, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), groups);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor cudnn_convolution_relu(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, int64_t groups) {
+    return at::_ops::cudnn_convolution_relu::call(self, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), groups);
+  }
+}
+
+// aten::cudnn_convolution_relu(Tensor self, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, SymInt groups) -> Tensor
+inline at::Tensor cudnn_convolution_relu_symint(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, c10::SymInt groups) {
+    return at::_ops::cudnn_convolution_relu::call(self, weight, bias, stride, padding, dilation, groups);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor cudnn_convolution_relu(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, c10::SymInt groups) {
+    return at::_ops::cudnn_convolution_relu::call(self, weight, bias, stride, padding, dilation, groups);
+  }
+}
+
+// aten::cudnn_convolution_relu.out(Tensor self, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, SymInt groups, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cudnn_convolution_relu_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, int64_t groups) {
+    return at::_ops::cudnn_convolution_relu_out::call(self, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), groups, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & cudnn_convolution_relu_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, int64_t groups) {
+    return at::_ops::cudnn_convolution_relu_out::call(self, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), groups, out);
+  }
+}
+
+// aten::cudnn_convolution_relu.out(Tensor self, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, SymInt groups, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cudnn_convolution_relu_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, int64_t groups, at::Tensor & out) {
+    return at::_ops::cudnn_convolution_relu_out::call(self, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), groups, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & cudnn_convolution_relu_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, int64_t groups, at::Tensor & out) {
+    return at::_ops::cudnn_convolution_relu_out::call(self, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), groups, out);
+  }
+}
+
+// aten::cudnn_convolution_relu.out(Tensor self, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, SymInt groups, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cudnn_convolution_relu_symint_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, c10::SymInt groups) {
+    return at::_ops::cudnn_convolution_relu_out::call(self, weight, bias, stride, padding, dilation, groups, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & cudnn_convolution_relu_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, c10::SymInt groups) {
+    return at::_ops::cudnn_convolution_relu_out::call(self, weight, bias, stride, padding, dilation, groups, out);
+  }
+}
+
+// aten::cudnn_convolution_relu.out(Tensor self, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, SymInt groups, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cudnn_convolution_relu_symint_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, c10::SymInt groups, at::Tensor & out) {
+    return at::_ops::cudnn_convolution_relu_out::call(self, weight, bias, stride, padding, dilation, groups, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & cudnn_convolution_relu_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, c10::SymInt groups, at::Tensor & out) {
+    return at::_ops::cudnn_convolution_relu_out::call(self, weight, bias, stride, padding, dilation, groups, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cummax_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cummax_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..754657f417aff727686eecddb02ec261d9b78bec
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cummax_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> cummax(const at::Tensor & self, int64_t dim);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> cummax_out(const at::Tensor & self, int64_t dim, at::Tensor & values, at::Tensor & indices);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> cummax(const at::Tensor & self, at::Dimname dim);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> cummax_out(const at::Tensor & self, at::Dimname dim, at::Tensor & values, at::Tensor & indices);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cummin.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cummin.h
new file mode 100644
index 0000000000000000000000000000000000000000..50ec4bf8bdfff26771d843423955590f324055f3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cummin.h
@@ -0,0 +1,53 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/cummin_ops.h>
+
+namespace at {
+
+
+// aten::cummin(Tensor self, int dim) -> (Tensor values, Tensor indices)
+inline ::std::tuple<at::Tensor,at::Tensor> cummin(const at::Tensor & self, int64_t dim) {
+    return at::_ops::cummin::call(self, dim);
+}
+
+// aten::cummin.out(Tensor self, int dim, *, Tensor(a!) values, Tensor(b!) indices) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::std::tuple<at::Tensor &,at::Tensor &> cummin_out(at::Tensor & values, at::Tensor & indices, const at::Tensor & self, int64_t dim) {
+    return at::_ops::cummin_out::call(self, dim, values, indices);
+}
+// aten::cummin.out(Tensor self, int dim, *, Tensor(a!) values, Tensor(b!) indices) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::std::tuple<at::Tensor &,at::Tensor &> cummin_outf(const at::Tensor & self, int64_t dim, at::Tensor & values, at::Tensor & indices) {
+    return at::_ops::cummin_out::call(self, dim, values, indices);
+}
+
+// aten::cummin.dimname(Tensor self, Dimname dim) -> (Tensor values, Tensor indices)
+inline ::std::tuple<at::Tensor,at::Tensor> cummin(const at::Tensor & self, at::Dimname dim) {
+    return at::_ops::cummin_dimname::call(self, dim);
+}
+
+// aten::cummin.dimname_out(Tensor self, Dimname dim, *, Tensor(a!) values, Tensor(b!) indices) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::std::tuple<at::Tensor &,at::Tensor &> cummin_out(at::Tensor & values, at::Tensor & indices, const at::Tensor & self, at::Dimname dim) {
+    return at::_ops::cummin_dimname_out::call(self, dim, values, indices);
+}
+// aten::cummin.dimname_out(Tensor self, Dimname dim, *, Tensor(a!) values, Tensor(b!) indices) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::std::tuple<at::Tensor &,at::Tensor &> cummin_outf(const at::Tensor & self, at::Dimname dim, at::Tensor & values, at::Tensor & indices) {
+    return at::_ops::cummin_dimname_out::call(self, dim, values, indices);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..38802f73a1f000de2ba8856097094f1ea9f2d783
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_cumsum : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/dequantize_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/dequantize_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..25d1ecbc5ef2d2ef01f1ee4bb92081e6ad5532b1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/dequantize_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor dequantize(const at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/detach.h b/lib/python3.12/site-packages/torch/include/ATen/ops/detach.h
new file mode 100644
index 0000000000000000000000000000000000000000..202ae57a9f58d119b79460fc673bbf3779e2c235
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/detach.h
@@ -0,0 +1,35 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/detach_ops.h>
+
+namespace at {
+
+
+// aten::detach(Tensor(a) self) -> Tensor(a)
+inline at::Tensor detach(const at::Tensor & self) {
+    return at::_ops::detach::call(self);
+}
+
+// aten::detach_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & detach_(at::Tensor & self) {
+    return at::_ops::detach_::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/detach_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/detach_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..06db87aa42c9ca4b6e9ee1fc82d69d5ff392f378
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/detach_copy.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/detach_copy_ops.h>
+
+namespace at {
+
+
+// aten::detach_copy(Tensor self) -> Tensor
+inline at::Tensor detach_copy(const at::Tensor & self) {
+    return at::_ops::detach_copy::call(self);
+}
+
+// aten::detach_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & detach_copy_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::detach_copy_out::call(self, out);
+}
+// aten::detach_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & detach_copy_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::detach_copy_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/detach_copy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/detach_copy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..53153012488ed8e4e1a6868e3e0bc1cfb3d0ca93
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/detach_copy_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor detach_copy(const at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/detach_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/detach_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c98cbdb93f5cf94749a0c4b8b2504a8701a4d22f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/detach_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor detach(const at::Tensor & self);
+TORCH_API at::Tensor & detach_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/diagflat_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/diagflat_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..48122bfd35d12077958aa4400982c313d3349a06
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/diagflat_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API diagflat {
+  using schema = at::Tensor (const at::Tensor &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::diagflat";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "diagflat(Tensor self, int offset=0) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t offset);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t offset);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/diagonal_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/diagonal_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..b9960fe05e89350afb966b6771b2908c73cbef66
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/diagonal_copy.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/diagonal_copy_ops.h>
+
+namespace at {
+
+
+// aten::diagonal_copy(Tensor self, int offset=0, int dim1=0, int dim2=1) -> Tensor
+inline at::Tensor diagonal_copy(const at::Tensor & self, int64_t offset=0, int64_t dim1=0, int64_t dim2=1) {
+    return at::_ops::diagonal_copy::call(self, offset, dim1, dim2);
+}
+
+// aten::diagonal_copy.out(Tensor self, int offset=0, int dim1=0, int dim2=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & diagonal_copy_out(at::Tensor & out, const at::Tensor & self, int64_t offset=0, int64_t dim1=0, int64_t dim2=1) {
+    return at::_ops::diagonal_copy_out::call(self, offset, dim1, dim2, out);
+}
+// aten::diagonal_copy.out(Tensor self, int offset=0, int dim1=0, int dim2=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & diagonal_copy_outf(const at::Tensor & self, int64_t offset, int64_t dim1, int64_t dim2, at::Tensor & out) {
+    return at::_ops::diagonal_copy_out::call(self, offset, dim1, dim2, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/divide_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/divide_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..90a00b93f7a130bb9bd9bdcf9270f18235fcc6e8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/divide_native.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor divide(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & divide_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & divide_(at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor divide(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & divide_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor divide(const at::Tensor & self, const at::Tensor & other, ::std::optional<c10::string_view> rounding_mode);
+TORCH_API at::Tensor & divide_out(const at::Tensor & self, const at::Tensor & other, ::std::optional<c10::string_view> rounding_mode, at::Tensor & out);
+TORCH_API at::Tensor & divide_(at::Tensor & self, const at::Tensor & other, ::std::optional<c10::string_view> rounding_mode);
+TORCH_API at::Tensor divide(const at::Tensor & self, const at::Scalar & other, ::std::optional<c10::string_view> rounding_mode);
+TORCH_API at::Tensor & divide_(at::Tensor & self, const at::Scalar & other, ::std::optional<c10::string_view> rounding_mode);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/divide_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/divide_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..56f1e54ee0b75f005b662d944dbb3ea6e5e5a14e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/divide_ops.h
@@ -0,0 +1,127 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API divide_Tensor {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::divide";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "divide.Tensor(Tensor self, Tensor other) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & other);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other);
+};
+
+struct TORCH_API divide__Tensor {
+  using schema = at::Tensor & (at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::divide_";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "divide_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & other);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & other);
+};
+
+struct TORCH_API divide_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::divide";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "divide.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+};
+
+struct TORCH_API divide_Scalar {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::divide";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "divide.Scalar(Tensor self, Scalar other) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Scalar & other);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & other);
+};
+
+struct TORCH_API divide__Scalar {
+  using schema = at::Tensor & (at::Tensor &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::divide_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "divide_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Scalar & other);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Scalar & other);
+};
+
+struct TORCH_API divide_Tensor_mode {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, ::std::optional<c10::string_view>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::divide";
+  static constexpr const char* overload_name = "Tensor_mode";
+  static constexpr const char* schema_str = "divide.Tensor_mode(Tensor self, Tensor other, *, str? rounding_mode) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & other, ::std::optional<c10::string_view> rounding_mode);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, ::std::optional<c10::string_view> rounding_mode);
+};
+
+struct TORCH_API divide__Tensor_mode {
+  using schema = at::Tensor & (at::Tensor &, const at::Tensor &, ::std::optional<c10::string_view>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::divide_";
+  static constexpr const char* overload_name = "Tensor_mode";
+  static constexpr const char* schema_str = "divide_.Tensor_mode(Tensor(a!) self, Tensor other, *, str? rounding_mode) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & other, ::std::optional<c10::string_view> rounding_mode);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & other, ::std::optional<c10::string_view> rounding_mode);
+};
+
+struct TORCH_API divide_out_mode {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, ::std::optional<c10::string_view>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::divide";
+  static constexpr const char* overload_name = "out_mode";
+  static constexpr const char* schema_str = "divide.out_mode(Tensor self, Tensor other, *, str? rounding_mode, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & other, ::std::optional<c10::string_view> rounding_mode, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, ::std::optional<c10::string_view> rounding_mode, at::Tensor & out);
+};
+
+struct TORCH_API divide_Scalar_mode {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &, ::std::optional<c10::string_view>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::divide";
+  static constexpr const char* overload_name = "Scalar_mode";
+  static constexpr const char* schema_str = "divide.Scalar_mode(Tensor self, Scalar other, *, str? rounding_mode) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Scalar & other, ::std::optional<c10::string_view> rounding_mode);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & other, ::std::optional<c10::string_view> rounding_mode);
+};
+
+struct TORCH_API divide__Scalar_mode {
+  using schema = at::Tensor & (at::Tensor &, const at::Scalar &, ::std::optional<c10::string_view>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::divide_";
+  static constexpr const char* overload_name = "Scalar_mode";
+  static constexpr const char* schema_str = "divide_.Scalar_mode(Tensor(a!) self, Scalar other, *, str? rounding_mode) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Scalar & other, ::std::optional<c10::string_view> rounding_mode);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Scalar & other, ::std::optional<c10::string_view> rounding_mode);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/elu_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/elu_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..1b203e9f8301a85ad072e8c4709e598010436cf9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/elu_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_elu : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Scalar & alpha, const at::Scalar & scale, const at::Scalar & input_scale);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/empty.h b/lib/python3.12/site-packages/torch/include/ATen/ops/empty.h
new file mode 100644
index 0000000000000000000000000000000000000000..8d7b6a93d477cf490c64dda9d43b6ff69c5ff222
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/empty.h
@@ -0,0 +1,131 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/empty_ops.h>
+
+namespace at {
+
+
+// aten::empty.names(int[] size, *, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor
+inline at::Tensor empty(at::IntArrayRef size, ::std::optional<at::DimnameList> names, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::empty_names::call(size, names, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt(), c10::impl::check_tensor_options_and_extract_memory_format(options, memory_format));
+}
+// aten::empty.names(int[] size, *, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor
+inline at::Tensor empty(at::IntArrayRef size, ::std::optional<at::DimnameList> names, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format) {
+    return at::_ops::empty_names::call(size, names, dtype, layout, device, pin_memory, memory_format);
+}
+
+// aten::empty.memory_format(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor
+inline at::Tensor empty(at::IntArrayRef size, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::empty_memory_format::call(c10::fromIntArrayRefSlow(size), c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt(), c10::impl::check_tensor_options_and_extract_memory_format(options, memory_format));
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor empty(at::IntArrayRef size, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::empty_memory_format::call(c10::fromIntArrayRefSlow(size), c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt(), c10::impl::check_tensor_options_and_extract_memory_format(options, memory_format));
+  }
+}
+
+// aten::empty.memory_format(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor
+inline at::Tensor empty(at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format) {
+    return at::_ops::empty_memory_format::call(c10::fromIntArrayRefSlow(size), dtype, layout, device, pin_memory, memory_format);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor empty(at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format) {
+    return at::_ops::empty_memory_format::call(c10::fromIntArrayRefSlow(size), dtype, layout, device, pin_memory, memory_format);
+  }
+}
+
+// aten::empty.memory_format(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor
+inline at::Tensor empty_symint(c10::SymIntArrayRef size, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::empty_memory_format::call(size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt(), c10::impl::check_tensor_options_and_extract_memory_format(options, memory_format));
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor empty(c10::SymIntArrayRef size, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::empty_memory_format::call(size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt(), c10::impl::check_tensor_options_and_extract_memory_format(options, memory_format));
+  }
+}
+
+// aten::empty.memory_format(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor
+inline at::Tensor empty_symint(c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format) {
+    return at::_ops::empty_memory_format::call(size, dtype, layout, device, pin_memory, memory_format);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor empty(c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format) {
+    return at::_ops::empty_memory_format::call(size, dtype, layout, device, pin_memory, memory_format);
+  }
+}
+
+// aten::empty.out(SymInt[] size, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & empty_out(at::Tensor & out, at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::empty_out::call(c10::fromIntArrayRefSlow(size), memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & empty_out(at::Tensor & out, at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::empty_out::call(c10::fromIntArrayRefSlow(size), memory_format, out);
+  }
+}
+
+// aten::empty.out(SymInt[] size, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & empty_outf(at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out) {
+    return at::_ops::empty_out::call(c10::fromIntArrayRefSlow(size), memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & empty_outf(at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out) {
+    return at::_ops::empty_out::call(c10::fromIntArrayRefSlow(size), memory_format, out);
+  }
+}
+
+// aten::empty.out(SymInt[] size, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & empty_symint_out(at::Tensor & out, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::empty_out::call(size, memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & empty_out(at::Tensor & out, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::empty_out::call(size, memory_format, out);
+  }
+}
+
+// aten::empty.out(SymInt[] size, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & empty_symint_outf(c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out) {
+    return at::_ops::empty_out::call(size, memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & empty_outf(c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out) {
+    return at::_ops::empty_out::call(size, memory_format, out);
+  }
+}
+
+// aten::empty.names_out(int[] size, *, Dimname[]? names, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & empty_out(at::Tensor & out, at::IntArrayRef size, ::std::optional<at::DimnameList> names, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::empty_names_out::call(size, names, memory_format, out);
+}
+// aten::empty.names_out(int[] size, *, Dimname[]? names, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & empty_outf(at::IntArrayRef size, ::std::optional<at::DimnameList> names, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out) {
+    return at::_ops::empty_names_out::call(size, names, memory_format, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/empty_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..31a4ad07dc5f2f73319c2aba1f24fc399098b97d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_cpu_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor empty(at::IntArrayRef size, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor empty(at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+TORCH_API at::Tensor empty_symint(c10::SymIntArrayRef size, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor empty_symint(c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/empty_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..4f6f37bc8b8049a074cb1c31a49205b4f8f6460a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_ops.h
@@ -0,0 +1,61 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API empty_names {
+  using schema = at::Tensor (at::IntArrayRef, ::std::optional<at::DimnameList>, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>, ::std::optional<at::MemoryFormat>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::empty";
+  static constexpr const char* overload_name = "names";
+  static constexpr const char* schema_str = "empty.names(int[] size, *, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor";
+  static at::Tensor call(at::IntArrayRef size, ::std::optional<at::DimnameList> names, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, at::IntArrayRef size, ::std::optional<at::DimnameList> names, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+};
+
+struct TORCH_API empty_memory_format {
+  using schema = at::Tensor (c10::SymIntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>, ::std::optional<at::MemoryFormat>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::empty";
+  static constexpr const char* overload_name = "memory_format";
+  static constexpr const char* schema_str = "empty.memory_format(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor";
+  static at::Tensor call(c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+};
+
+struct TORCH_API empty_out {
+  using schema = at::Tensor & (c10::SymIntArrayRef, ::std::optional<at::MemoryFormat>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::empty";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "empty.out(SymInt[] size, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+};
+
+struct TORCH_API empty_names_out {
+  using schema = at::Tensor & (at::IntArrayRef, ::std::optional<at::DimnameList>, ::std::optional<at::MemoryFormat>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::empty";
+  static constexpr const char* overload_name = "names_out";
+  static constexpr const char* schema_str = "empty.names_out(int[] size, *, Dimname[]? names, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(at::IntArrayRef size, ::std::optional<at::DimnameList> names, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::IntArrayRef size, ::std::optional<at::DimnameList> names, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/empty_strided_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_strided_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f149a94225dbce06d077b8ec4c68fc0fdfbe83ca
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_strided_meta_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor empty_strided(at::IntArrayRef size, at::IntArrayRef stride, at::TensorOptions options={});
+TORCH_API at::Tensor empty_strided(at::IntArrayRef size, at::IntArrayRef stride, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor empty_strided_symint(c10::SymIntArrayRef size, c10::SymIntArrayRef stride, at::TensorOptions options={});
+TORCH_API at::Tensor empty_strided_symint(c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/eq.h b/lib/python3.12/site-packages/torch/include/ATen/ops/eq.h
new file mode 100644
index 0000000000000000000000000000000000000000..633b869d0efda0480aefbeb7fc7b2e0a790e429a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/eq.h
@@ -0,0 +1,53 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/eq_ops.h>
+
+namespace at {
+
+
+// aten::eq.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & eq_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::eq_Scalar_out::call(self, other, out);
+}
+// aten::eq.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & eq_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out) {
+    return at::_ops::eq_Scalar_out::call(self, other, out);
+}
+
+// aten::eq.Scalar(Tensor self, Scalar other) -> Tensor
+inline at::Tensor eq(const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::eq_Scalar::call(self, other);
+}
+
+// aten::eq.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & eq_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::eq_Tensor_out::call(self, other, out);
+}
+// aten::eq.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & eq_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::eq_Tensor_out::call(self, other, out);
+}
+
+// aten::eq.Tensor(Tensor self, Tensor other) -> Tensor
+inline at::Tensor eq(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::eq_Tensor::call(self, other);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/eq_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/eq_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..01a49d7cb06e7c785c1b89b171e0ef3e5dfe83d8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/eq_meta.h
@@ -0,0 +1,32 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_eq_Scalar : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Scalar & other);
+};
+struct TORCH_API structured_eq_Tensor : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Tensor & other);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/equal_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/equal_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..35e7f850c65f6888c3cb88b4f04345db6845f8ee
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/equal_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API bool equal(const at::Tensor & self, const at::Tensor & other);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/equal_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/equal_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..f9d857926b3dae8ce8021d8f7332b3db30a9bbb3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/equal_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API bool cpu_equal(const at::Tensor & self, const at::Tensor & other);
+TORCH_API bool cuda_equal(const at::Tensor & self, const at::Tensor & other);
+TORCH_API bool equal_quantized_cpu(const at::Tensor & self, const at::Tensor & other);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/expand_as_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/expand_as_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..da4dda519eb9601faf86fd6117b46f126cd8c1f3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/expand_as_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API expand_as {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::expand_as";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "expand_as(Tensor(a) self, Tensor other) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & other);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/expm1_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/expm1_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8976a19ebc551f29df02a57471452849be779f17
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/expm1_cuda_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor expm1(const at::Tensor & self);
+TORCH_API at::Tensor & expm1_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & expm1_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & expm1_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/expm1_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/expm1_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9fdc88c0e0946b3874619eb238d25543b3d09f01
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/expm1_native.h
@@ -0,0 +1,29 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/expm1_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_expm1_out : public at::meta::structured_expm1 {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+TORCH_API at::Tensor expm1_sparse(const at::Tensor & self);
+TORCH_API at::Tensor & expm1_sparse_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & expm1_sparse_(at::Tensor & self);
+TORCH_API at::Tensor expm1_sparse_csr(const at::Tensor & self);
+TORCH_API at::Tensor & expm1_sparse_csr_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & expm1_sparse_csr_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/eye_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/eye_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7ff56d09b88d47bbb04866c0bb736ff1a7deed90
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/eye_cuda_dispatch.h
@@ -0,0 +1,30 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor & eye_out(at::Tensor & out, int64_t n);
+TORCH_API at::Tensor & eye_outf(int64_t n, at::Tensor & out);
+TORCH_API at::Tensor & eye_symint_out(at::Tensor & out, c10::SymInt n);
+TORCH_API at::Tensor & eye_symint_outf(c10::SymInt n, at::Tensor & out);
+TORCH_API at::Tensor & eye_out(at::Tensor & out, int64_t n, int64_t m);
+TORCH_API at::Tensor & eye_outf(int64_t n, int64_t m, at::Tensor & out);
+TORCH_API at::Tensor & eye_symint_out(at::Tensor & out, c10::SymInt n, c10::SymInt m);
+TORCH_API at::Tensor & eye_symint_outf(c10::SymInt n, c10::SymInt m, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/eye_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/eye_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..67741dfdc6fd1ba5586860c441a6fef11f22ab0e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/eye_native.h
@@ -0,0 +1,26 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor eye(int64_t n, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={});
+TORCH_API at::Tensor & eye_out_cpu(int64_t n, at::Tensor & out);
+TORCH_API at::Tensor & eye_out_cuda(int64_t n, at::Tensor & out);
+TORCH_API at::Tensor eye(int64_t n, int64_t m, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={});
+TORCH_API at::Tensor & eye_out_cpu(int64_t n, int64_t m, at::Tensor & out);
+TORCH_API at::Tensor & eye_out_cuda(int64_t n, int64_t m, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_tensor_affine_cachemask_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_tensor_affine_cachemask_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9659018e0b5af756a3a2596c41a5c8d9a780e11f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_tensor_affine_cachemask_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API fake_quantize_per_tensor_affine_cachemask {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, double, int64_t, int64_t, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fake_quantize_per_tensor_affine_cachemask";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "fake_quantize_per_tensor_affine_cachemask(Tensor self, float scale, int zero_point, int quant_min, int quant_max) -> (Tensor output, Tensor mask)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self, double scale, int64_t zero_point, int64_t quant_min, int64_t quant_max);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, double scale, int64_t zero_point, int64_t quant_min, int64_t quant_max);
+};
+
+struct TORCH_API fake_quantize_per_tensor_affine_cachemask_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const at::Tensor &, double, int64_t, int64_t, int64_t, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fake_quantize_per_tensor_affine_cachemask";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "fake_quantize_per_tensor_affine_cachemask.out(Tensor self, float scale, int zero_point, int quant_min, int quant_max, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & self, double scale, int64_t zero_point, int64_t quant_min, int64_t quant_max, at::Tensor & out0, at::Tensor & out1);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, double scale, int64_t zero_point, int64_t quant_min, int64_t quant_max, at::Tensor & out0, at::Tensor & out1);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_int8_weight_fp32_activation_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_int8_weight_fp32_activation_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..3e1c701a7df01eefc422975a9912ade387afdd27
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_int8_weight_fp32_activation_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API fbgemm_linear_int8_weight_fp32_activation {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Scalar &, const at::Scalar &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fbgemm_linear_int8_weight_fp32_activation";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "fbgemm_linear_int8_weight_fp32_activation(Tensor input, Tensor weight, Tensor packed, Tensor col_offsets, Scalar weight_scale, Scalar weight_zero_point, Tensor bias) -> Tensor";
+  static at::Tensor call(const at::Tensor & input, const at::Tensor & weight, const at::Tensor & packed, const at::Tensor & col_offsets, const at::Scalar & weight_scale, const at::Scalar & weight_zero_point, const at::Tensor & bias);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const at::Tensor & weight, const at::Tensor & packed, const at::Tensor & col_offsets, const at::Scalar & weight_scale, const at::Scalar & weight_zero_point, const at::Tensor & bias);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fft.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fft.h
new file mode 100644
index 0000000000000000000000000000000000000000..0bc872e2d2a4a2ebfe13bf438a5573bb05e27e1e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fft.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/fft_fft_ops.h>
+
+namespace at {
+
+
+// aten::fft_fft(Tensor self, SymInt? n=None, int dim=-1, str? norm=None) -> Tensor
+inline at::Tensor fft_fft(const at::Tensor & self, ::std::optional<int64_t> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fft::call(self, n.has_value() ? ::std::make_optional(c10::SymInt(*n)) : ::std::nullopt, dim, norm);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor fft_fft(const at::Tensor & self, ::std::optional<int64_t> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fft::call(self, n.has_value() ? ::std::make_optional(c10::SymInt(*n)) : ::std::nullopt, dim, norm);
+  }
+}
+
+// aten::fft_fft(Tensor self, SymInt? n=None, int dim=-1, str? norm=None) -> Tensor
+inline at::Tensor fft_fft_symint(const at::Tensor & self, ::std::optional<c10::SymInt> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fft::call(self, n, dim, norm);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor fft_fft(const at::Tensor & self, ::std::optional<c10::SymInt> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fft::call(self, n, dim, norm);
+  }
+}
+
+// aten::fft_fft.out(Tensor self, SymInt? n=None, int dim=-1, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_fft_out(at::Tensor & out, const at::Tensor & self, ::std::optional<int64_t> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fft_out::call(self, n.has_value() ? ::std::make_optional(c10::SymInt(*n)) : ::std::nullopt, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & fft_fft_out(at::Tensor & out, const at::Tensor & self, ::std::optional<int64_t> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fft_out::call(self, n.has_value() ? ::std::make_optional(c10::SymInt(*n)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_fft.out(Tensor self, SymInt? n=None, int dim=-1, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_fft_outf(const at::Tensor & self, ::std::optional<int64_t> n, int64_t dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_fft_out::call(self, n.has_value() ? ::std::make_optional(c10::SymInt(*n)) : ::std::nullopt, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & fft_fft_outf(const at::Tensor & self, ::std::optional<int64_t> n, int64_t dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_fft_out::call(self, n.has_value() ? ::std::make_optional(c10::SymInt(*n)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_fft.out(Tensor self, SymInt? n=None, int dim=-1, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_fft_symint_out(at::Tensor & out, const at::Tensor & self, ::std::optional<c10::SymInt> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fft_out::call(self, n, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & fft_fft_out(at::Tensor & out, const at::Tensor & self, ::std::optional<c10::SymInt> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fft_out::call(self, n, dim, norm, out);
+  }
+}
+
+// aten::fft_fft.out(Tensor self, SymInt? n=None, int dim=-1, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_fft_symint_outf(const at::Tensor & self, ::std::optional<c10::SymInt> n, int64_t dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_fft_out::call(self, n, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & fft_fft_outf(const at::Tensor & self, ::std::optional<c10::SymInt> n, int64_t dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_fft_out::call(self, n, dim, norm, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fftfreq_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fftfreq_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..19f3e6da9efcae6b16a07bf5e529c9a91230c301
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fftfreq_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API fft_fftfreq {
+  using schema = at::Tensor (int64_t, double, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fft_fftfreq";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "fft_fftfreq(int n, float d=1.0, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(int64_t n, double d, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t n, double d, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API fft_fftfreq_out {
+  using schema = at::Tensor & (int64_t, double, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fft_fftfreq";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "fft_fftfreq.out(int n, float d=1.0, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t n, double d, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t n, double d, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fftshift_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fftshift_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..df3f14aa3a9a5cf916d9c9cf359753b9e4201740
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fftshift_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API fft_fftshift {
+  using schema = at::Tensor (const at::Tensor &, at::OptionalIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fft_fftshift";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "fft_fftshift(Tensor self, int[1]? dim=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::OptionalIntArrayRef dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::OptionalIntArrayRef dim);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft2_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft2_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..274834f781293500deaf3cac9955d55f227abbfa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft2_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor fft_ifft2(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor fft_ifft2_symint(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_ifft2_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_ifft2_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+TORCH_API at::Tensor & fft_ifft2_symint_out(at::Tensor & out, const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_ifft2_symint_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..22d2cbdf57064774b4cc4e0fb3d76149fa7ef7ed
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor fft_ifft(const at::Tensor & self, ::std::optional<int64_t> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor fft_ifft_symint(const at::Tensor & self, ::std::optional<c10::SymInt> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_ifft_out(at::Tensor & out, const at::Tensor & self, ::std::optional<int64_t> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_ifft_outf(const at::Tensor & self, ::std::optional<int64_t> n, int64_t dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+TORCH_API at::Tensor & fft_ifft_symint_out(at::Tensor & out, const at::Tensor & self, ::std::optional<c10::SymInt> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_ifft_symint_outf(const at::Tensor & self, ::std::optional<c10::SymInt> n, int64_t dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..a603e3a7edcdd86873dbf59fd7c8f8d70f1ed22d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor fft_ifft_symint(const at::Tensor & self, ::std::optional<c10::SymInt> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_ifft_symint_out(const at::Tensor & self, ::std::optional<c10::SymInt> n, int64_t dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftn.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftn.h
new file mode 100644
index 0000000000000000000000000000000000000000..891c382301f717e6bf554f1727256cc1730f5626
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftn.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/fft_ifftn_ops.h>
+
+namespace at {
+
+
+// aten::fft_ifftn(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None) -> Tensor
+inline at::Tensor fft_ifftn(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifftn::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor fft_ifftn(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifftn::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm);
+  }
+}
+
+// aten::fft_ifftn(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None) -> Tensor
+inline at::Tensor fft_ifftn_symint(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifftn::call(self, s, dim, norm);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor fft_ifftn(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifftn::call(self, s, dim, norm);
+  }
+}
+
+// aten::fft_ifftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_ifftn_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifftn_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & fft_ifftn_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifftn_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_ifftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_ifftn_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_ifftn_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & fft_ifftn_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_ifftn_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_ifftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_ifftn_symint_out(at::Tensor & out, const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifftn_out::call(self, s, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & fft_ifftn_out(at::Tensor & out, const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifftn_out::call(self, s, dim, norm, out);
+  }
+}
+
+// aten::fft_ifftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_ifftn_symint_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_ifftn_out::call(self, s, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & fft_ifftn_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_ifftn_out::call(self, s, dim, norm, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftn_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftn_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..cea5eaab86759c0039577126b82747fc2f04a846
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftn_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API fft_ifftn {
+  using schema = at::Tensor (const at::Tensor &, at::OptionalSymIntArrayRef, at::OptionalIntArrayRef, ::std::optional<c10::string_view>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fft_ifftn";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "fft_ifftn(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm);
+};
+
+struct TORCH_API fft_ifftn_out {
+  using schema = at::Tensor & (const at::Tensor &, at::OptionalSymIntArrayRef, at::OptionalIntArrayRef, ::std::optional<c10::string_view>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fft_ifftn";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "fft_ifftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftshift_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftshift_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fd94234a10263e427d0c20b57ea1d94d0db036ca
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftshift_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor fft_ifftshift(const at::Tensor & self, at::OptionalIntArrayRef dim=::std::nullopt);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/floor_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/floor_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..add8e67dd2cfc29ef7a6533c46619132417d4aac
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/floor_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor floor(const at::Tensor & self);
+TORCH_API at::Tensor & floor_(at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/floor_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/floor_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..058fdbb796fc12e1238b3a76ade4575504b0f0f0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/floor_meta_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor floor(const at::Tensor & self);
+TORCH_API at::Tensor & floor_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & floor_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & floor_(at::Tensor & self);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9b539c92c5ead61c9c46e99e6041436f7aa24eb8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor fmax(const at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ecc15bf7ab06a46aa2e290a333d93d0d41671a92
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor fmax(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & fmax_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & fmax_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fmin_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fmin_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c237d8d482305ff2bdf607e5a9e8af4da5ccd98b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fmin_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor fmin(const at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/frac_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/frac_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..af00a64acc1589b90a7835c95907d03f977bb7fa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/frac_native.h
@@ -0,0 +1,29 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/frac_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_frac_out : public at::meta::structured_frac {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+TORCH_API at::Tensor frac_sparse(const at::Tensor & self);
+TORCH_API at::Tensor & frac_sparse_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & frac_sparse_(at::Tensor & self);
+TORCH_API at::Tensor frac_sparse_csr(const at::Tensor & self);
+TORCH_API at::Tensor & frac_sparse_csr_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & frac_sparse_csr_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool2d_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool2d_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..72aa13a125e391972648f248c28c4c2c9a3e5da5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool2d_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> fractional_max_pool2d(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & random_samples);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> fractional_max_pool2d_out(at::Tensor & output, at::Tensor & indices, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & random_samples);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> fractional_max_pool2d_outf(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & random_samples, at::Tensor & output, at::Tensor & indices);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1282e3cfbddbe76d8cd749edb64cb28ed8a0a336
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_backward_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor fractional_max_pool3d_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & indices);
+TORCH_API at::Tensor & fractional_max_pool3d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & indices);
+TORCH_API at::Tensor & fractional_max_pool3d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & indices, at::Tensor & grad_input);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4181d812014b906272561a3b8d1ec0bfa46dc561
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_backward_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor fractional_max_pool3d_backward_cpu(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & indices);
+TORCH_API at::Tensor & fractional_max_pool3d_backward_out_cpu(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & indices, at::Tensor & grad_input);
+TORCH_API at::Tensor fractional_max_pool3d_backward_cuda(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & indices);
+TORCH_API at::Tensor & fractional_max_pool3d_backward_out_cuda(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & indices, at::Tensor & grad_input);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..aeb1de214d8f99f5a112e2357851093f1f7ee407
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> fractional_max_pool3d(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & random_samples);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> fractional_max_pool3d_out(at::Tensor & output, at::Tensor & indices, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & random_samples);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> fractional_max_pool3d_outf(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & random_samples, at::Tensor & output, at::Tensor & indices);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gather_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gather_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..807f2f4049014d4f7474a616a00d4cb6758981a6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gather_backward.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/gather_backward_ops.h>
+
+namespace at {
+
+
+// aten::gather_backward(Tensor grad, Tensor self, int dim, Tensor index, bool sparse_grad) -> Tensor
+inline at::Tensor gather_backward(const at::Tensor & grad, const at::Tensor & self, int64_t dim, const at::Tensor & index, bool sparse_grad) {
+    return at::_ops::gather_backward::call(grad, self, dim, index, sparse_grad);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gather_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gather_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..51c768ff7b71a6a2b2fa81f8980367e9b8190aa2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gather_native.h
@@ -0,0 +1,25 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/gather_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_gather_out : public at::meta::structured_gather {
+void impl(const at::Tensor & self, int64_t dim, const at::Tensor & index, bool sparse_grad, const at::Tensor & out);
+};
+TORCH_API at::Tensor gather(const at::Tensor & self, at::Dimname dim, const at::Tensor & index, bool sparse_grad=false);
+TORCH_API at::Tensor & gather_out(const at::Tensor & self, at::Dimname dim, const at::Tensor & index, bool sparse_grad, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/ge.h b/lib/python3.12/site-packages/torch/include/ATen/ops/ge.h
new file mode 100644
index 0000000000000000000000000000000000000000..34cc41531f0b2625a996cc8d9228ff8064c23444
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/ge.h
@@ -0,0 +1,53 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/ge_ops.h>
+
+namespace at {
+
+
+// aten::ge.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & ge_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::ge_Scalar_out::call(self, other, out);
+}
+// aten::ge.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & ge_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out) {
+    return at::_ops::ge_Scalar_out::call(self, other, out);
+}
+
+// aten::ge.Scalar(Tensor self, Scalar other) -> Tensor
+inline at::Tensor ge(const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::ge_Scalar::call(self, other);
+}
+
+// aten::ge.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & ge_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::ge_Tensor_out::call(self, other, out);
+}
+// aten::ge.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & ge_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::ge_Tensor_out::call(self, other, out);
+}
+
+// aten::ge.Tensor(Tensor self, Tensor other) -> Tensor
+inline at::Tensor ge(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::ge_Tensor::call(self, other);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gelu_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gelu_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cf3f261dcd2c9e65f308e42572ccfe7977b7dcdc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gelu_backward_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor gelu_backward(const at::Tensor & grad_output, const at::Tensor & self, c10::string_view approximate="none");
+TORCH_API at::Tensor & gelu_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, c10::string_view approximate="none");
+TORCH_API at::Tensor & gelu_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, c10::string_view approximate, at::Tensor & grad_input);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gelu_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gelu_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..e6e6df48c9a547695897ca91aedbc4ce1b77943b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gelu_native.h
@@ -0,0 +1,32 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/gelu_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_gelu_out_cpu : public at::meta::structured_gelu {
+void impl(const at::Tensor & self, c10::string_view approximate, const at::Tensor & out);
+};
+struct TORCH_API structured_gelu_out_cuda : public at::meta::structured_gelu {
+void impl(const at::Tensor & self, c10::string_view approximate, const at::Tensor & out);
+};
+TORCH_API at::Tensor NestedTensor_gelu(const at::Tensor & self, c10::string_view approximate="none");
+TORCH_API at::Tensor & NestedTensor_gelu_(at::Tensor & self, c10::string_view approximate="none");
+TORCH_API at::Tensor mkldnn_gelu(const at::Tensor & self, c10::string_view approximate="none");
+TORCH_API at::Tensor gelu_quantized_cpu(const at::Tensor & self, c10::string_view approximate="none");
+TORCH_API at::Tensor & gelu_quantized_cpu_(at::Tensor & self, c10::string_view approximate="none");
+TORCH_API at::Tensor gelu_quantized_cuda(const at::Tensor & self, c10::string_view approximate="none");
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_2d_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_2d_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6168dc34a416c3b96985903ab9777ed91c406fec
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_2d_backward_cpu_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> grid_sampler_2d_backward(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners, ::std::array<bool,2> output_mask);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d.h b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d.h
new file mode 100644
index 0000000000000000000000000000000000000000..cb7266ad38c30373a9612e6f20c75196da0060ce
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/grid_sampler_3d_ops.h>
+
+namespace at {
+
+
+// aten::grid_sampler_3d(Tensor input, Tensor grid, int interpolation_mode, int padding_mode, bool align_corners) -> Tensor
+inline at::Tensor grid_sampler_3d(const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners) {
+    return at::_ops::grid_sampler_3d::call(input, grid, interpolation_mode, padding_mode, align_corners);
+}
+
+// aten::grid_sampler_3d.out(Tensor input, Tensor grid, int interpolation_mode, int padding_mode, bool align_corners, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & grid_sampler_3d_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners) {
+    return at::_ops::grid_sampler_3d_out::call(input, grid, interpolation_mode, padding_mode, align_corners, out);
+}
+// aten::grid_sampler_3d.out(Tensor input, Tensor grid, int interpolation_mode, int padding_mode, bool align_corners, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & grid_sampler_3d_outf(const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners, at::Tensor & out) {
+    return at::_ops::grid_sampler_3d_out::call(input, grid, interpolation_mode, padding_mode, align_corners, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..1b079531ac1bf3d8550743047e2d4e36d3bb07e4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d_backward.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/grid_sampler_3d_backward_ops.h>
+
+namespace at {
+
+
+// aten::grid_sampler_3d_backward(Tensor grad_output, Tensor input, Tensor grid, int interpolation_mode, int padding_mode, bool align_corners, bool[2] output_mask) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> grid_sampler_3d_backward(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners, ::std::array<bool,2> output_mask) {
+    return at::_ops::grid_sampler_3d_backward::call(grad_output, input, grid, interpolation_mode, padding_mode, align_corners, output_mask);
+}
+
+// aten::grid_sampler_3d_backward.out(Tensor grad_output, Tensor input, Tensor grid, int interpolation_mode, int padding_mode, bool align_corners, bool[2] output_mask, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> grid_sampler_3d_backward_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners, ::std::array<bool,2> output_mask) {
+    return at::_ops::grid_sampler_3d_backward_out::call(grad_output, input, grid, interpolation_mode, padding_mode, align_corners, output_mask, out0, out1);
+}
+// aten::grid_sampler_3d_backward.out(Tensor grad_output, Tensor input, Tensor grid, int interpolation_mode, int padding_mode, bool align_corners, bool[2] output_mask, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> grid_sampler_3d_backward_outf(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners, ::std::array<bool,2> output_mask, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::grid_sampler_3d_backward_out::call(grad_output, input, grid, interpolation_mode, padding_mode, align_corners, output_mask, out0, out1);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..343023d64d5d21d7536eef5954d62fd9a0240a35
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d_backward_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> grid_sampler_3d_backward(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners, ::std::array<bool,2> output_mask);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gru_cell.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gru_cell.h
new file mode 100644
index 0000000000000000000000000000000000000000..0566f87fdc7f99a969093f3f74440d4806891949
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gru_cell.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/gru_cell_ops.h>
+
+namespace at {
+
+
+// aten::gru_cell(Tensor input, Tensor hx, Tensor w_ih, Tensor w_hh, Tensor? b_ih=None, Tensor? b_hh=None) -> Tensor
+inline at::Tensor gru_cell(const at::Tensor & input, const at::Tensor & hx, const at::Tensor & w_ih, const at::Tensor & w_hh, const ::std::optional<at::Tensor> & b_ih={}, const ::std::optional<at::Tensor> & b_hh={}) {
+    return at::_ops::gru_cell::call(input, hx, w_ih, w_hh, b_ih, b_hh);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gt_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gt_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..359ba036e171128210e518de814c615484bca24d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gt_meta_dispatch.h
@@ -0,0 +1,30 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor gt(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & gt_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & gt_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & gt_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor gt(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & gt_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & gt_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & gt_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..c7765a90e43a93fe516c5bcb608968ade1199504
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_hardshrink : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Scalar & lambd);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardswish.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardswish.h
new file mode 100644
index 0000000000000000000000000000000000000000..db48503ecafbe301ad41cdcac3846c38a74b9669
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardswish.h
@@ -0,0 +1,44 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/hardswish_ops.h>
+
+namespace at {
+
+
+// aten::hardswish.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hardswish_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::hardswish_out::call(self, out);
+}
+// aten::hardswish.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hardswish_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::hardswish_out::call(self, out);
+}
+
+// aten::hardswish(Tensor self) -> Tensor
+inline at::Tensor hardswish(const at::Tensor & self) {
+    return at::_ops::hardswish::call(self);
+}
+
+// aten::hardswish_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & hardswish_(at::Tensor & self) {
+    return at::_ops::hardswish_::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/heaviside.h b/lib/python3.12/site-packages/torch/include/ATen/ops/heaviside.h
new file mode 100644
index 0000000000000000000000000000000000000000..208478173add0bec5c3fa6f1798655d18550bacc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/heaviside.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/heaviside_ops.h>
+
+namespace at {
+
+
+// aten::heaviside.out(Tensor self, Tensor values, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & heaviside_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & values) {
+    return at::_ops::heaviside_out::call(self, values, out);
+}
+// aten::heaviside.out(Tensor self, Tensor values, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & heaviside_outf(const at::Tensor & self, const at::Tensor & values, at::Tensor & out) {
+    return at::_ops::heaviside_out::call(self, values, out);
+}
+
+// aten::heaviside(Tensor self, Tensor values) -> Tensor
+inline at::Tensor heaviside(const at::Tensor & self, const at::Tensor & values) {
+    return at::_ops::heaviside::call(self, values);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hypot_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hypot_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..36fe940f953a313521b36625dc28c3f5eaef62ca
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hypot_meta_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor hypot(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & hypot_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & hypot_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & hypot_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/i0_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/i0_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a4f3e136ebe261b908fc48dd390af63e641f922a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/i0_meta_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor i0(const at::Tensor & self);
+TORCH_API at::Tensor & i0_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & i0_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & i0_(at::Tensor & self);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/im2col_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/im2col_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1e4b438fbdf0cbc8427e29007b7de1c50e836dc2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/im2col_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API im2col_out {
+  using schema = at::Tensor & (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::im2col";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "im2col.out(Tensor self, int[2] kernel_size, int[2] dilation, int[2] padding, int[2] stride, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride, at::Tensor & out);
+};
+
+struct TORCH_API im2col {
+  using schema = at::Tensor (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::im2col";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "im2col(Tensor self, int[2] kernel_size, int[2] dilation, int[2] padding, int[2] stride) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_add_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_add_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c48adc8cf61afe59deed014552ecacfc2cbc1ee7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_add_meta_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor index_add(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & index_add_out(at::Tensor & out, const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & index_add_outf(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source, const at::Scalar & alpha, at::Tensor & out);
+TORCH_API at::Tensor & index_add_(at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source, const at::Scalar & alpha=1);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..ad9463bcb16ca692a8ade7bb56afe428b86b6e1d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_meta.h
@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_index_Tensor : public TensorIteratorBase {
+    
+                template <bool SIZES = false, bool STRIDES = false>
+                struct TORCH_API precompute_out {
+                    
+                    precompute_out<true, STRIDES> set_sizes(at::DimVector value) {
+                        static_assert(SIZES == false, "sizes already set");
+                        precompute_out<true, STRIDES> ret;
+ret.sizes = value;
+ret.strides = this->strides;
+return ret;
+                    }
+                
+
+                    precompute_out<SIZES, true> set_strides(at::DimVector value) {
+                        static_assert(STRIDES == false, "strides already set");
+                        precompute_out<SIZES, true> ret;
+ret.sizes = this->sizes;
+ret.strides = value;
+return ret;
+                    }
+                
+                    at::DimVector sizes;
+at::DimVector strides;
+            };
+    using meta_return_ty = precompute_out <true, true>;
+    meta_return_ty meta(const at::Tensor & self, at::IOptTensorListRef indices);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_put.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_put.h
new file mode 100644
index 0000000000000000000000000000000000000000..3dd5a658f81c52d84d985b98639abbc5ed4f17ab
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_put.h
@@ -0,0 +1,44 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/index_put_ops.h>
+
+namespace at {
+
+
+// aten::index_put_(Tensor(a!) self, Tensor?[] indices, Tensor values, bool accumulate=False) -> Tensor(a!)
+inline at::Tensor & index_put_(at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false) {
+    return at::_ops::index_put_::call(self, indices, values, accumulate);
+}
+
+// aten::index_put(Tensor self, Tensor?[] indices, Tensor values, bool accumulate=False) -> Tensor
+inline at::Tensor index_put(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false) {
+    return at::_ops::index_put::call(self, indices, values, accumulate);
+}
+
+// aten::index_put.out(Tensor self, Tensor?[] indices, Tensor values, bool accumulate=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & index_put_out(at::Tensor & out, const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false) {
+    return at::_ops::index_put_out::call(self, indices, values, accumulate, out);
+}
+// aten::index_put.out(Tensor self, Tensor?[] indices, Tensor values, bool accumulate=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & index_put_outf(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate, at::Tensor & out) {
+    return at::_ops::index_put_out::call(self, indices, values, accumulate, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a55bac9432763279dbe4bd8406c1bf11541e7d51
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_ops.h
@@ -0,0 +1,61 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API index_select_out {
+  using schema = at::Tensor & (const at::Tensor &, int64_t, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::index_select";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "index_select.out(Tensor self, int dim, Tensor index, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, int64_t dim, const at::Tensor & index, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, const at::Tensor & index, at::Tensor & out);
+};
+
+struct TORCH_API index_select {
+  using schema = at::Tensor (const at::Tensor &, int64_t, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::index_select";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "index_select(Tensor self, int dim, Tensor index) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t dim, const at::Tensor & index);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, const at::Tensor & index);
+};
+
+struct TORCH_API index_select_dimname_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Dimname, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::index_select";
+  static constexpr const char* overload_name = "dimname_out";
+  static constexpr const char* schema_str = "index_select.dimname_out(Tensor self, Dimname dim, Tensor index, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Dimname dim, const at::Tensor & index, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname dim, const at::Tensor & index, at::Tensor & out);
+};
+
+struct TORCH_API index_select_dimname {
+  using schema = at::Tensor (const at::Tensor &, at::Dimname, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::index_select";
+  static constexpr const char* overload_name = "dimname";
+  static constexpr const char* schema_str = "index_select.dimname(Tensor self, Dimname dim, Tensor index) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::Dimname dim, const at::Tensor & index);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname dim, const at::Tensor & index);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/infinitely_differentiable_gelu_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/infinitely_differentiable_gelu_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..6092c8e62d6ff35bcddbc9b39a93b5ce4e171883
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/infinitely_differentiable_gelu_backward.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/infinitely_differentiable_gelu_backward_ops.h>
+
+namespace at {
+
+
+// aten::infinitely_differentiable_gelu_backward(Tensor grad, Tensor self) -> Tensor
+inline at::Tensor infinitely_differentiable_gelu_backward(const at::Tensor & grad, const at::Tensor & self) {
+    return at::_ops::infinitely_differentiable_gelu_backward::call(grad, self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/int_repr.h b/lib/python3.12/site-packages/torch/include/ATen/ops/int_repr.h
new file mode 100644
index 0000000000000000000000000000000000000000..3d515e3aed167767d246dbe1694b086152b6d9ea
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/int_repr.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/int_repr_ops.h>
+
+namespace at {
+
+
+// aten::int_repr(Tensor self) -> Tensor
+inline at::Tensor int_repr(const at::Tensor & self) {
+    return at::_ops::int_repr::call(self);
+}
+
+// aten::int_repr.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & int_repr_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::int_repr_out::call(self, out);
+}
+// aten::int_repr.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & int_repr_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::int_repr_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_set_to_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_set_to_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e06a867a3f36a53486cbb1c50d2a9a72ab62924d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_set_to_cpu_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API bool is_set_to(const at::Tensor & self, const at::Tensor & tensor);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_signed.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_signed.h
new file mode 100644
index 0000000000000000000000000000000000000000..8bbc6acc8f6b79810dbc16e9de2a3b8d349b20ad
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_signed.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/is_signed_ops.h>
+
+namespace at {
+
+
+// aten::is_signed(Tensor self) -> bool
+inline bool __dispatch_is_signed(const at::Tensor & self) {
+    return at::_ops::is_signed::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_vulkan_available_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_vulkan_available_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..56fc6f811455c1a3ff745e9e5afa83f3ca2ae37f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_vulkan_available_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API is_vulkan_available {
+  using schema = bool ();
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::is_vulkan_available";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "is_vulkan_available() -> bool";
+  static bool call();
+  static bool redispatch(c10::DispatchKeySet dispatchKeySet);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/isin_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/isin_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..167b3745c943fde1372e7f45205ee26e87dcfd30
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/isin_ops.h
@@ -0,0 +1,83 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API isin_Tensor_Tensor_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, bool, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::isin";
+  static constexpr const char* overload_name = "Tensor_Tensor_out";
+  static constexpr const char* schema_str = "isin.Tensor_Tensor_out(Tensor elements, Tensor test_elements, *, bool assume_unique=False, bool invert=False, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & elements, const at::Tensor & test_elements, bool assume_unique, bool invert, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & elements, const at::Tensor & test_elements, bool assume_unique, bool invert, at::Tensor & out);
+};
+
+struct TORCH_API isin_Tensor_Tensor {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::isin";
+  static constexpr const char* overload_name = "Tensor_Tensor";
+  static constexpr const char* schema_str = "isin.Tensor_Tensor(Tensor elements, Tensor test_elements, *, bool assume_unique=False, bool invert=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & elements, const at::Tensor & test_elements, bool assume_unique, bool invert);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & elements, const at::Tensor & test_elements, bool assume_unique, bool invert);
+};
+
+struct TORCH_API isin_Tensor_Scalar_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, bool, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::isin";
+  static constexpr const char* overload_name = "Tensor_Scalar_out";
+  static constexpr const char* schema_str = "isin.Tensor_Scalar_out(Tensor elements, Scalar test_element, *, bool assume_unique=False, bool invert=False, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & elements, const at::Scalar & test_element, bool assume_unique, bool invert, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & elements, const at::Scalar & test_element, bool assume_unique, bool invert, at::Tensor & out);
+};
+
+struct TORCH_API isin_Tensor_Scalar {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::isin";
+  static constexpr const char* overload_name = "Tensor_Scalar";
+  static constexpr const char* schema_str = "isin.Tensor_Scalar(Tensor elements, Scalar test_element, *, bool assume_unique=False, bool invert=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & elements, const at::Scalar & test_element, bool assume_unique, bool invert);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & elements, const at::Scalar & test_element, bool assume_unique, bool invert);
+};
+
+struct TORCH_API isin_Scalar_Tensor_out {
+  using schema = at::Tensor & (const at::Scalar &, const at::Tensor &, bool, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::isin";
+  static constexpr const char* overload_name = "Scalar_Tensor_out";
+  static constexpr const char* schema_str = "isin.Scalar_Tensor_out(Scalar element, Tensor test_elements, *, bool assume_unique=False, bool invert=False, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Scalar & element, const at::Tensor & test_elements, bool assume_unique, bool invert, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & element, const at::Tensor & test_elements, bool assume_unique, bool invert, at::Tensor & out);
+};
+
+struct TORCH_API isin_Scalar_Tensor {
+  using schema = at::Tensor (const at::Scalar &, const at::Tensor &, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::isin";
+  static constexpr const char* overload_name = "Scalar_Tensor";
+  static constexpr const char* schema_str = "isin.Scalar_Tensor(Scalar element, Tensor test_elements, *, bool assume_unique=False, bool invert=False) -> Tensor";
+  static at::Tensor call(const at::Scalar & element, const at::Tensor & test_elements, bool assume_unique, bool invert);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & element, const at::Tensor & test_elements, bool assume_unique, bool invert);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/isposinf_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/isposinf_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5e8ad855dda3d589b9df3ec6d86da5d65220b07c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/isposinf_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor isposinf(const at::Tensor & self);
+TORCH_API at::Tensor & isposinf_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & isposinf_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/isposinf_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/isposinf_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..7b53e08400f0486f28de81071b9a54e75aedc58d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/isposinf_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_isposinf : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/item_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/item_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6648c6622ef814ee942cb359f17150185f8eac08
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/item_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Scalar item(const at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lgamma_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lgamma_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5895ca6f884253d6d07434b43e8276e7baf979a1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lgamma_cuda_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor lgamma(const at::Tensor & self);
+TORCH_API at::Tensor & lgamma_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & lgamma_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & lgamma_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_ex_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_ex_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3be116ba230f4f3b26f27741a436b5a163d6c03d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_ex_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> linalg_ldl_factor_ex(const at::Tensor & self, bool hermitian=false, bool check_errors=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linalg_ldl_factor_ex_out(at::Tensor & LD, at::Tensor & pivots, at::Tensor & info, const at::Tensor & self, bool hermitian=false, bool check_errors=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linalg_ldl_factor_ex_outf(const at::Tensor & self, bool hermitian, bool check_errors, at::Tensor & LD, at::Tensor & pivots, at::Tensor & info);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..64a0949f1f4ecee5f38aef449204a44c50ee785b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> linalg_lu(const at::Tensor & A, bool pivot=true);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linalg_lu_out(at::Tensor & P, at::Tensor & L, at::Tensor & U, const at::Tensor & A, bool pivot=true);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linalg_lu_outf(const at::Tensor & A, bool pivot, at::Tensor & P, at::Tensor & L, at::Tensor & U);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor_ex_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor_ex_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8089cc19412242c8c06e88b1ef36d425cf2c136c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor_ex_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> linalg_lu_factor_ex(const at::Tensor & A, bool pivot=true, bool check_errors=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linalg_lu_factor_ex_out(at::Tensor & LU, at::Tensor & pivots, at::Tensor & info, const at::Tensor & A, bool pivot=true, bool check_errors=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linalg_lu_factor_ex_outf(const at::Tensor & A, bool pivot, bool check_errors, at::Tensor & LU, at::Tensor & pivots, at::Tensor & info);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_power.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_power.h
new file mode 100644
index 0000000000000000000000000000000000000000..25c3e05a451f2f7d3b7f3d7c58c0fcb2f38a63e4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_power.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/linalg_matrix_power_ops.h>
+
+namespace at {
+
+
+// aten::linalg_matrix_power(Tensor self, int n) -> Tensor
+inline at::Tensor linalg_matrix_power(const at::Tensor & self, int64_t n) {
+    return at::_ops::linalg_matrix_power::call(self, n);
+}
+
+// aten::linalg_matrix_power.out(Tensor self, int n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matrix_power_out(at::Tensor & out, const at::Tensor & self, int64_t n) {
+    return at::_ops::linalg_matrix_power_out::call(self, n, out);
+}
+// aten::linalg_matrix_power.out(Tensor self, int n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matrix_power_outf(const at::Tensor & self, int64_t n, at::Tensor & out) {
+    return at::_ops::linalg_matrix_power_out::call(self, n, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_pinv_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_pinv_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ed8ce8dd259087d7ddbc9c89537db3558d48b11d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_pinv_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor linalg_pinv(const at::Tensor & self, const ::std::optional<at::Tensor> & atol={}, const ::std::optional<at::Tensor> & rtol={}, bool hermitian=false);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_solve.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_solve.h
new file mode 100644
index 0000000000000000000000000000000000000000..59da067edd1c1856ff5a14000e3aec89c7f1f3a4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_solve.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/linalg_solve_ops.h>
+
+namespace at {
+
+
+// aten::linalg_solve(Tensor A, Tensor B, *, bool left=True) -> Tensor
+inline at::Tensor linalg_solve(const at::Tensor & A, const at::Tensor & B, bool left=true) {
+    return at::_ops::linalg_solve::call(A, B, left);
+}
+
+// aten::linalg_solve.out(Tensor A, Tensor B, *, bool left=True, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_solve_out(at::Tensor & out, const at::Tensor & A, const at::Tensor & B, bool left=true) {
+    return at::_ops::linalg_solve_out::call(A, B, left, out);
+}
+// aten::linalg_solve.out(Tensor A, Tensor B, *, bool left=True, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_solve_outf(const at::Tensor & A, const at::Tensor & B, bool left, at::Tensor & out) {
+    return at::_ops::linalg_solve_out::call(A, B, left, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_solve_triangular_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_solve_triangular_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..79be6ba3e2f32dc84ffbfa214fe7736e6a4c9ed7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_solve_triangular_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor linalg_solve_triangular(const at::Tensor & self, const at::Tensor & B, bool upper, bool left=true, bool unitriangular=false);
+TORCH_API at::Tensor & linalg_solve_triangular_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & B, bool upper, bool left=true, bool unitriangular=false);
+TORCH_API at::Tensor & linalg_solve_triangular_outf(const at::Tensor & self, const at::Tensor & B, bool upper, bool left, bool unitriangular, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_svd_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_svd_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..eb04a7dbb91158794cd9a6729a6ee79f82205c6e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_svd_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API linalg_svd {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, bool, ::std::optional<c10::string_view>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_svd";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_svd(Tensor A, bool full_matrices=True, *, str? driver=None) -> (Tensor U, Tensor S, Tensor Vh)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & A, bool full_matrices, ::std::optional<c10::string_view> driver);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, bool full_matrices, ::std::optional<c10::string_view> driver);
+};
+
+struct TORCH_API linalg_svd_U {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> (const at::Tensor &, bool, ::std::optional<c10::string_view>, at::Tensor &, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_svd";
+  static constexpr const char* overload_name = "U";
+  static constexpr const char* schema_str = "linalg_svd.U(Tensor A, bool full_matrices=True, *, str? driver=None, Tensor(a!) U, Tensor(b!) S, Tensor(c!) Vh) -> (Tensor(a!) U, Tensor(b!) S, Tensor(c!) Vh)";
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> call(const at::Tensor & A, bool full_matrices, ::std::optional<c10::string_view> driver, at::Tensor & U, at::Tensor & S, at::Tensor & Vh);
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, bool full_matrices, ::std::optional<c10::string_view> driver, at::Tensor & U, at::Tensor & S, at::Tensor & Vh);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_svdvals.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_svdvals.h
new file mode 100644
index 0000000000000000000000000000000000000000..945a768e07b37f966f4d6e4094faa40ae9a83859
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_svdvals.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/linalg_svdvals_ops.h>
+
+namespace at {
+
+
+// aten::linalg_svdvals(Tensor A, *, str? driver=None) -> Tensor
+inline at::Tensor linalg_svdvals(const at::Tensor & A, ::std::optional<c10::string_view> driver=::std::nullopt) {
+    return at::_ops::linalg_svdvals::call(A, driver);
+}
+
+// aten::linalg_svdvals.out(Tensor A, *, str? driver=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_svdvals_out(at::Tensor & out, const at::Tensor & A, ::std::optional<c10::string_view> driver=::std::nullopt) {
+    return at::_ops::linalg_svdvals_out::call(A, driver, out);
+}
+// aten::linalg_svdvals.out(Tensor A, *, str? driver=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_svdvals_outf(const at::Tensor & A, ::std::optional<c10::string_view> driver, at::Tensor & out) {
+    return at::_ops::linalg_svdvals_out::call(A, driver, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_vector_norm_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_vector_norm_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..bae9edb4f3461d3c6f7763fd02c3dd957d7de719
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_vector_norm_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor linalg_vector_norm(const at::Tensor & self, const at::Scalar & ord=2, at::OptionalIntArrayRef dim=::std::nullopt, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & linalg_vector_norm_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & ord=2, at::OptionalIntArrayRef dim=::std::nullopt, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & linalg_vector_norm_outf(const at::Tensor & self, const at::Scalar & ord, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_vector_norm_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_vector_norm_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..25ead2d6e0e1adfafe12307243afd3b1e9351390
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_vector_norm_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API linalg_vector_norm {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &, at::OptionalIntArrayRef, bool, ::std::optional<at::ScalarType>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_vector_norm";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_vector_norm(Tensor self, Scalar ord=2, int[1]? dim=None, bool keepdim=False, *, ScalarType? dtype=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Scalar & ord, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & ord, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype);
+};
+
+struct TORCH_API linalg_vector_norm_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, at::OptionalIntArrayRef, bool, ::std::optional<at::ScalarType>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_vector_norm";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "linalg_vector_norm.out(Tensor self, Scalar ord=2, int[1]? dim=None, bool keepdim=False, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Scalar & ord, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & ord, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linear_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linear_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..08c2da14ad151a60c1d88aa87c8396ed012a3cf2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linear_backward.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/linear_backward_ops.h>
+
+namespace at {
+
+
+// aten::linear_backward(Tensor self, Tensor grad_output, Tensor weight, bool[3] output_mask) -> (Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> linear_backward(const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, ::std::array<bool,3> output_mask) {
+    return at::_ops::linear_backward::call(self, grad_output, weight, output_mask);
+}
+
+// aten::linear_backward.out(Tensor self, Tensor grad_output, Tensor weight, bool[3] output_mask, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linear_backward_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, ::std::array<bool,3> output_mask) {
+    return at::_ops::linear_backward_out::call(self, grad_output, weight, output_mask, out0, out1, out2);
+}
+// aten::linear_backward.out(Tensor self, Tensor grad_output, Tensor weight, bool[3] output_mask, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linear_backward_outf(const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2) {
+    return at::_ops::linear_backward_out::call(self, grad_output, weight, output_mask, out0, out1, out2);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linear_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linear_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ae96074b6a0bae11cc4b8ce4840dafbebffae665
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linear_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & linear_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={});
+TORCH_API at::Tensor & linear_outf(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linear_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linear_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b89506d8e1fce5f99e4079c2a1e08c54d10d2887
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linear_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor linear(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={});
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linspace_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linspace_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..db410916ccbaeed2ba64dfa188552ace59a35bd9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linspace_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,36 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor linspace(const at::Scalar & start, const at::Scalar & end, int64_t steps, at::TensorOptions options={});
+TORCH_API at::Tensor linspace(const at::Scalar & start, const at::Scalar & end, int64_t steps, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor linspace(const at::Tensor & start, const at::Tensor & end, int64_t steps, at::TensorOptions options={});
+TORCH_API at::Tensor linspace(const at::Tensor & start, const at::Tensor & end, int64_t steps, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor & linspace_out(at::Tensor & out, const at::Tensor & start, const at::Tensor & end, int64_t steps);
+TORCH_API at::Tensor & linspace_outf(const at::Tensor & start, const at::Tensor & end, int64_t steps, at::Tensor & out);
+TORCH_API at::Tensor linspace(const at::Tensor & start, const at::Scalar & end, int64_t steps, at::TensorOptions options={});
+TORCH_API at::Tensor linspace(const at::Tensor & start, const at::Scalar & end, int64_t steps, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor & linspace_out(at::Tensor & out, const at::Tensor & start, const at::Scalar & end, int64_t steps);
+TORCH_API at::Tensor & linspace_outf(const at::Tensor & start, const at::Scalar & end, int64_t steps, at::Tensor & out);
+TORCH_API at::Tensor linspace(const at::Scalar & start, const at::Tensor & end, int64_t steps, at::TensorOptions options={});
+TORCH_API at::Tensor linspace(const at::Scalar & start, const at::Tensor & end, int64_t steps, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor & linspace_out(at::Tensor & out, const at::Scalar & start, const at::Tensor & end, int64_t steps);
+TORCH_API at::Tensor & linspace_outf(const at::Scalar & start, const at::Tensor & end, int64_t steps, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linspace_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linspace_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cfaf717deb745b3b62bc8f2f8b88f32857b54cef
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linspace_cpu_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor & linspace_out(at::Tensor & out, const at::Scalar & start, const at::Scalar & end, int64_t steps);
+TORCH_API at::Tensor & linspace_outf(const at::Scalar & start, const at::Scalar & end, int64_t steps, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log10.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log10.h
new file mode 100644
index 0000000000000000000000000000000000000000..5929193ceacf67f7a12c6c28d6841dcfc6ea0a33
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log10.h
@@ -0,0 +1,44 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/log10_ops.h>
+
+namespace at {
+
+
+// aten::log10(Tensor self) -> Tensor
+inline at::Tensor log10(const at::Tensor & self) {
+    return at::_ops::log10::call(self);
+}
+
+// aten::log10_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & log10_(at::Tensor & self) {
+    return at::_ops::log10_::call(self);
+}
+
+// aten::log10.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & log10_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::log10_out::call(self, out);
+}
+// aten::log10.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & log10_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::log10_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log10_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log10_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..168b322c7be87bb2452d264e76a0d39c186478af
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log10_ops.h
@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API log10 {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::log10";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "log10(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+struct TORCH_API log10_ {
+  using schema = at::Tensor & (at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::log10_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "log10_(Tensor(a!) self) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self);
+};
+
+struct TORCH_API log10_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::log10";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "log10.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e6895456452f63e0a29d93c384a7ed6dff5e0b25
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log_cpu_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor log(const at::Tensor & self);
+TORCH_API at::Tensor & log_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & log_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & log_(at::Tensor & self);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_forward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_forward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0ad2fd16e07040a6387a04f43e0a8e92001fd7fa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_forward_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> log_sigmoid_forward(const at::Tensor & self);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> log_sigmoid_forward_out(at::Tensor & output, at::Tensor & buffer, const at::Tensor & self);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> log_sigmoid_forward_outf(const at::Tensor & self, at::Tensor & output, at::Tensor & buffer);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logcumsumexp_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logcumsumexp_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1f75cb31fecaa314e08a514e3f1bbc1ee258c761
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logcumsumexp_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor logcumsumexp(const at::Tensor & self, at::Dimname dim);
+TORCH_API at::Tensor & logcumsumexp_out(at::Tensor & out, const at::Tensor & self, at::Dimname dim);
+TORCH_API at::Tensor & logcumsumexp_outf(const at::Tensor & self, at::Dimname dim, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logit_backward_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logit_backward_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f111b9e02b26d573b3299fccc7ffc4fa6f0a2cbe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logit_backward_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor logit_backward(const at::Tensor & grad_output, const at::Tensor & self, ::std::optional<double> eps=::std::nullopt);
+TORCH_API at::Tensor & logit_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, ::std::optional<double> eps=::std::nullopt);
+TORCH_API at::Tensor & logit_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, ::std::optional<double> eps, at::Tensor & grad_input);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logspace_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logspace_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..604ad1e6f8814573094115038bd6af6cf7b6d71c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logspace_ops.h
@@ -0,0 +1,105 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API logspace {
+  using schema = at::Tensor (const at::Scalar &, const at::Scalar &, int64_t, double, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logspace";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "logspace(Scalar start, Scalar end, int steps, float base=10.0, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(const at::Scalar & start, const at::Scalar & end, int64_t steps, double base, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & start, const at::Scalar & end, int64_t steps, double base, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API logspace_Tensor_Tensor {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, int64_t, double, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logspace";
+  static constexpr const char* overload_name = "Tensor_Tensor";
+  static constexpr const char* schema_str = "logspace.Tensor_Tensor(Tensor start, Tensor end, int steps, float base=10.0, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & start, const at::Tensor & end, int64_t steps, double base, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & start, const at::Tensor & end, int64_t steps, double base, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API logspace_Tensor_Scalar {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &, int64_t, double, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logspace";
+  static constexpr const char* overload_name = "Tensor_Scalar";
+  static constexpr const char* schema_str = "logspace.Tensor_Scalar(Tensor start, Scalar end, int steps, float base=10.0, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & start, const at::Scalar & end, int64_t steps, double base, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & start, const at::Scalar & end, int64_t steps, double base, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API logspace_Scalar_Tensor {
+  using schema = at::Tensor (const at::Scalar &, const at::Tensor &, int64_t, double, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logspace";
+  static constexpr const char* overload_name = "Scalar_Tensor";
+  static constexpr const char* schema_str = "logspace.Scalar_Tensor(Scalar start, Tensor end, int steps, float base=10.0, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(const at::Scalar & start, const at::Tensor & end, int64_t steps, double base, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & start, const at::Tensor & end, int64_t steps, double base, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API logspace_out {
+  using schema = at::Tensor & (const at::Scalar &, const at::Scalar &, int64_t, double, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logspace";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "logspace.out(Scalar start, Scalar end, int steps, float base=10.0, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Scalar & start, const at::Scalar & end, int64_t steps, double base, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & start, const at::Scalar & end, int64_t steps, double base, at::Tensor & out);
+};
+
+struct TORCH_API logspace_Tensor_Tensor_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, int64_t, double, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logspace";
+  static constexpr const char* overload_name = "Tensor_Tensor_out";
+  static constexpr const char* schema_str = "logspace.Tensor_Tensor_out(Tensor start, Tensor end, int steps, float base=10.0, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & start, const at::Tensor & end, int64_t steps, double base, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & start, const at::Tensor & end, int64_t steps, double base, at::Tensor & out);
+};
+
+struct TORCH_API logspace_Tensor_Scalar_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, int64_t, double, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logspace";
+  static constexpr const char* overload_name = "Tensor_Scalar_out";
+  static constexpr const char* schema_str = "logspace.Tensor_Scalar_out(Tensor start, Scalar end, int steps, float base=10.0, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & start, const at::Scalar & end, int64_t steps, double base, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & start, const at::Scalar & end, int64_t steps, double base, at::Tensor & out);
+};
+
+struct TORCH_API logspace_Scalar_Tensor_out {
+  using schema = at::Tensor & (const at::Scalar &, const at::Tensor &, int64_t, double, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logspace";
+  static constexpr const char* overload_name = "Scalar_Tensor_out";
+  static constexpr const char* schema_str = "logspace.Scalar_Tensor_out(Scalar start, Tensor end, int steps, float base=10.0, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Scalar & start, const at::Tensor & end, int64_t steps, double base, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & start, const at::Tensor & end, int64_t steps, double base, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logsumexp_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logsumexp_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7610b25094236f45ea793b78587d68fe3f7de625
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logsumexp_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor & logsumexp_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false);
+TORCH_API at::Tensor & logsumexp_outf(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, at::Tensor & out);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logsumexp_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logsumexp_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..cbcbff20f71cd510fdff0ad7008058a6aa9df03c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logsumexp_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor logsumexp(const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false);
+TORCH_API at::Tensor & logsumexp_out(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, at::Tensor & out);
+TORCH_API at::Tensor logsumexp(const at::Tensor & self, at::DimnameList dim, bool keepdim=false);
+TORCH_API at::Tensor & logsumexp_out(const at::Tensor & self, at::DimnameList dim, bool keepdim, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lstm_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lstm_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..8a3a2ada61f8e5d0e1d79f63899415713b0a20a6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lstm_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API lstm_input {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, at::TensorList, at::TensorList, bool, int64_t, double, bool, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::lstm";
+  static constexpr const char* overload_name = "input";
+  static constexpr const char* schema_str = "lstm.input(Tensor input, Tensor[] hx, Tensor[] params, bool has_biases, int num_layers, float dropout, bool train, bool bidirectional, bool batch_first) -> (Tensor, Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & input, at::TensorList hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional, bool batch_first);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, at::TensorList hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional, bool batch_first);
+};
+
+struct TORCH_API lstm_data {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, at::TensorList, at::TensorList, bool, int64_t, double, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::lstm";
+  static constexpr const char* overload_name = "data";
+  static constexpr const char* schema_str = "lstm.data(Tensor data, Tensor batch_sizes, Tensor[] hx, Tensor[] params, bool has_biases, int num_layers, float dropout, bool train, bool bidirectional) -> (Tensor, Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & data, const at::Tensor & batch_sizes, at::TensorList hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & data, const at::Tensor & batch_sizes, at::TensorList hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lt_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lt_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2f39c1e160f1aab8eed435fc84ca45c2fb6878bf
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lt_cpu_dispatch.h
@@ -0,0 +1,30 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor lt(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & lt_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & lt_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & lt_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor lt(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & lt_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & lt_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & lt_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lu_solve_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lu_solve_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..949a0a64f1893d2e65cff291434650ee5f6b7f78
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lu_solve_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor lu_solve(const at::Tensor & self, const at::Tensor & LU_data, const at::Tensor & LU_pivots);
+TORCH_API at::Tensor & lu_solve_out(const at::Tensor & self, const at::Tensor & LU_data, const at::Tensor & LU_pivots, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lu_unpack_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lu_unpack_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..a233999fb384daa377c5ae0c6dd129974cf10b22
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lu_unpack_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_lu_unpack : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & LU_data, const at::Tensor & LU_pivots, bool unpack_data, bool unpack_pivots);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mT.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mT.h
new file mode 100644
index 0000000000000000000000000000000000000000..dcf67023b797514ea81c19333f461ddfd4348a7b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mT.h
@@ -0,0 +1,26 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/mT_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mT_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mT_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..72eac1fe8a81886f38f1e0089c8c06d6db47b372
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mT_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor mT(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..12164d10c8b54dd84e559f8c287d9189a2c3368a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select_backward.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/masked_select_backward_ops.h>
+
+namespace at {
+
+
+// aten::masked_select_backward(Tensor grad, Tensor input, Tensor mask) -> Tensor
+inline at::Tensor masked_select_backward(const at::Tensor & grad, const at::Tensor & input, const at::Tensor & mask) {
+    return at::_ops::masked_select_backward::call(grad, input, mask);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/matrix_exp_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/matrix_exp_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..9f8a4f024e6db417fb067abeb12e96d1cdc935a7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/matrix_exp_backward.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/matrix_exp_backward_ops.h>
+
+namespace at {
+
+
+// aten::matrix_exp_backward(Tensor self, Tensor grad) -> Tensor
+inline at::Tensor matrix_exp_backward(const at::Tensor & self, const at::Tensor & grad) {
+    return at::_ops::matrix_exp_backward::call(self, grad);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9680e33f3f4f259b6cf3f0cd76260b0791c55cbf
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> max(const at::Tensor & self, int64_t dim, bool keepdim=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> max_out(at::Tensor & max, at::Tensor & max_values, const at::Tensor & self, int64_t dim, bool keepdim=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> max_outf(const at::Tensor & self, int64_t dim, bool keepdim, at::Tensor & max, at::Tensor & max_values);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8fac3223e3eac29b0a94175a695094f16544cb7b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_native.h
@@ -0,0 +1,32 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/max_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_max_out : public at::meta::structured_max_dim {
+void impl(const at::Tensor & self, int64_t dim, bool keepdim, const at::Tensor & max, const at::Tensor & max_values);
+};
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> qmax(const at::Tensor & self, int64_t dim, bool keepdim=false);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> max(const at::Tensor & self, at::Dimname dim, bool keepdim=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> max_out(const at::Tensor & self, at::Dimname dim, bool keepdim, at::Tensor & max, at::Tensor & max_values);
+TORCH_API at::Tensor max(const at::Tensor & self);
+TORCH_API at::Tensor & max_unary_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor max_quantized_cpu(const at::Tensor & self);
+TORCH_API at::Tensor & max_quantized_unary_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor max(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & max_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool1d_with_indices_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool1d_with_indices_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9921979cc2be89f4db4f8a6cb9ac6b396db3b50c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool1d_with_indices_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> max_pool1d_with_indices(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, bool ceil_mode=false);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool1d_with_indices_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool1d_with_indices_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..97a6c9d1ef03b3dc38ca5a7e0e17362b5890bf6a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool1d_with_indices_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API max_pool1d_with_indices {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::max_pool1d_with_indices";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "max_pool1d_with_indices(Tensor self, int[1] kernel_size, int[1] stride=[], int[1] padding=0, int[1] dilation=1, bool ceil_mode=False) -> (Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_with_indices_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_with_indices_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6ea1ec241ee1d96df0745a5c868a2f1c07c264d2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_with_indices_backward_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor max_pool2d_with_indices_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices);
+TORCH_API at::Tensor & max_pool2d_with_indices_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices);
+TORCH_API at::Tensor & max_pool2d_with_indices_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices, at::Tensor & grad_input);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_with_indices_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_with_indices_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..dc43a4e2a308d6b3407e5f94fe4944a84834346b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_with_indices_backward_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API max_pool2d_with_indices_backward_grad_input {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::max_pool2d_with_indices_backward";
+  static constexpr const char* overload_name = "grad_input";
+  static constexpr const char* schema_str = "max_pool2d_with_indices_backward.grad_input(Tensor grad_output, Tensor self, int[2] kernel_size, int[2] stride, int[2] padding, int[2] dilation, bool ceil_mode, Tensor indices, *, Tensor(a!) grad_input) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices, at::Tensor & grad_input);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices, at::Tensor & grad_input);
+};
+
+struct TORCH_API max_pool2d_with_indices_backward {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::max_pool2d_with_indices_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "max_pool2d_with_indices_backward(Tensor grad_output, Tensor self, int[2] kernel_size, int[2] stride, int[2] padding, int[2] dilation, bool ceil_mode, Tensor indices) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c0d8ad619529da0ede5335542a72bb371da65c56
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_backward_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor max_pool3d_with_indices_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices);
+TORCH_API at::Tensor & max_pool3d_with_indices_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices);
+TORCH_API at::Tensor & max_pool3d_with_indices_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices, at::Tensor & grad_input);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..2e47cec5fd64e761e890163d95c9ba3dd467319d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_backward_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API max_pool3d_with_indices_backward_grad_input {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::max_pool3d_with_indices_backward";
+  static constexpr const char* overload_name = "grad_input";
+  static constexpr const char* schema_str = "max_pool3d_with_indices_backward.grad_input(Tensor grad_output, Tensor self, int[3] kernel_size, int[3] stride, int[3] padding, int[3] dilation, bool ceil_mode, Tensor indices, *, Tensor(a!) grad_input) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices, at::Tensor & grad_input);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices, at::Tensor & grad_input);
+};
+
+struct TORCH_API max_pool3d_with_indices_backward {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::max_pool3d_with_indices_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "max_pool3d_with_indices_backward(Tensor grad_output, Tensor self, int[3] kernel_size, int[3] stride, int[3] padding, int[3] dilation, bool ceil_mode, Tensor indices) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & indices);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mean_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mean_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..102bd60b4739956cf49c5d0b78b89f2002deb9ab
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mean_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor mean(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & mean_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & mean_outf(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_batch_norm_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_batch_norm_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..574c96d8de230d3dbacbb0d621688a266582a6ca
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_batch_norm_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API miopen_batch_norm {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, bool, double, double);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::miopen_batch_norm";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "miopen_batch_norm(Tensor input, Tensor weight, Tensor? bias, Tensor? running_mean, Tensor? running_var, bool training, float exponential_average_factor, float epsilon) -> (Tensor, Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, bool training, double exponential_average_factor, double epsilon);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, bool training, double exponential_average_factor, double epsilon);
+};
+
+struct TORCH_API miopen_batch_norm_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, bool, double, double, at::Tensor &, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::miopen_batch_norm";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "miopen_batch_norm.out(Tensor input, Tensor weight, Tensor? bias, Tensor? running_mean, Tensor? running_var, bool training, float exponential_average_factor, float epsilon, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> call(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, bool training, double exponential_average_factor, double epsilon, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, bool training, double exponential_average_factor, double epsilon, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c1d72ab2ea53e8f9001c672d5cbf418e41a6803b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & miopen_convolution_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic);
+TORCH_API at::Tensor & miopen_convolution_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, at::Tensor & out);
+TORCH_API at::Tensor & miopen_convolution_symint_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic);
+TORCH_API at::Tensor & miopen_convolution_symint_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_relu.h b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_relu.h
new file mode 100644
index 0000000000000000000000000000000000000000..b172055ea9926e436a14941a0893ad41bfb9c3f6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_relu.h
@@ -0,0 +1,47 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/miopen_convolution_relu_ops.h>
+
+namespace at {
+
+
+// aten::miopen_convolution_relu(Tensor self, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, SymInt groups) -> Tensor
+inline at::Tensor miopen_convolution_relu(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, int64_t groups) {
+    return at::_ops::miopen_convolution_relu::call(self, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), groups);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor miopen_convolution_relu(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, int64_t groups) {
+    return at::_ops::miopen_convolution_relu::call(self, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), groups);
+  }
+}
+
+// aten::miopen_convolution_relu(Tensor self, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, SymInt groups) -> Tensor
+inline at::Tensor miopen_convolution_relu_symint(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, c10::SymInt groups) {
+    return at::_ops::miopen_convolution_relu::call(self, weight, bias, stride, padding, dilation, groups);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor miopen_convolution_relu(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, c10::SymInt groups) {
+    return at::_ops::miopen_convolution_relu::call(self, weight, bias, stride, padding, dilation, groups);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..060c6e5e8a7705505b9e7ec8f87dd05009aedbde
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_backward_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API void miopen_rnn_backward_out(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, const at::Tensor & output, const ::std::optional<at::Tensor> & grad_output, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, int64_t mode, int64_t hidden_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, at::IntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::TensorList out3);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,::std::vector<at::Tensor>> miopen_rnn_backward(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, const at::Tensor & output, const ::std::optional<at::Tensor> & grad_output, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, int64_t mode, int64_t hidden_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, at::IntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mish_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mish_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..afb802fe15221216ec9775529cb62cb60e531d24
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mish_backward_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor mish_backward(const at::Tensor & grad_output, const at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mish_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mish_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cce89804eaebd1d8307c02ea6f4caa2d2651d47b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mish_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor mish(const at::Tensor & self);
+TORCH_API at::Tensor & mish_(at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mish_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mish_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9e655e690e0c9d2acf2815bd0347370c01506a29
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mish_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/mish_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_mish_out : public at::meta::structured_mish {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mish_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mish_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..283aabe3fa57456e1a1b8a04f55554698897a72a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mish_ops.h
@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API mish {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mish";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "mish(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+struct TORCH_API mish_ {
+  using schema = at::Tensor & (at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mish_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "mish_(Tensor(a!) self) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self);
+};
+
+struct TORCH_API mish_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mish";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "mish.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_backward_input_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_backward_input_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f66d6e9d012c31752dbae177c68a7d9fecb4d34b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_backward_input_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & mkldnn_linear_backward_input_out(at::Tensor & out, at::IntArrayRef input_size, const at::Tensor & grad_output, const at::Tensor & weight);
+TORCH_API at::Tensor & mkldnn_linear_backward_input_outf(at::IntArrayRef input_size, const at::Tensor & grad_output, const at::Tensor & weight, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..b2950d933e65739cbe88f8cdb6031390316adbee
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d_backward_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API mkldnn_max_pool2d_backward {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mkldnn_max_pool2d_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "mkldnn_max_pool2d_backward(Tensor grad_output, Tensor output, Tensor input, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad_output, const at::Tensor & output, const at::Tensor & input, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & output, const at::Tensor & input, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+};
+
+struct TORCH_API mkldnn_max_pool2d_backward_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mkldnn_max_pool2d_backward";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "mkldnn_max_pool2d_backward.out(Tensor grad_output, Tensor output, Tensor input, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & grad_output, const at::Tensor & output, const at::Tensor & input, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & output, const at::Tensor & input, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mode_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mode_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..53021626912e8a87044af031ffeaec02e6e69037
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mode_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> mode_out(at::Tensor & values, at::Tensor & indices, const at::Tensor & self, int64_t dim=-1, bool keepdim=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> mode_outf(const at::Tensor & self, int64_t dim, bool keepdim, at::Tensor & values, at::Tensor & indices);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/moveaxis_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/moveaxis_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..2639a91c4d9315c1b2ab5fc8ea413fb075dc4382
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/moveaxis_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor moveaxis(const at::Tensor & self, at::IntArrayRef source, at::IntArrayRef destination);
+TORCH_API at::Tensor moveaxis(const at::Tensor & self, int64_t source, int64_t destination);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mps_convolution_transpose_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mps_convolution_transpose_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..32252906698c331e962d28a3cc79ed8c574e823f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mps_convolution_transpose_backward.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/mps_convolution_transpose_backward_ops.h>
+
+namespace at {
+
+
+// aten::mps_convolution_transpose_backward(Tensor self, Tensor grad_output, Tensor weight, SymInt[] padding, SymInt[] output_padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool[2] output_mask) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> mps_convolution_transpose_backward(const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef output_padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, ::std::array<bool,2> output_mask) {
+    return at::_ops::mps_convolution_transpose_backward::call(self, grad_output, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(output_padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, output_mask);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  ::std::tuple<at::Tensor,at::Tensor> mps_convolution_transpose_backward(const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef output_padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, ::std::array<bool,2> output_mask) {
+    return at::_ops::mps_convolution_transpose_backward::call(self, grad_output, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(output_padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, output_mask);
+  }
+}
+
+// aten::mps_convolution_transpose_backward(Tensor self, Tensor grad_output, Tensor weight, SymInt[] padding, SymInt[] output_padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool[2] output_mask) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> mps_convolution_transpose_backward_symint(const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, ::std::array<bool,2> output_mask) {
+    return at::_ops::mps_convolution_transpose_backward::call(self, grad_output, weight, padding, output_padding, stride, dilation, groups, output_mask);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  ::std::tuple<at::Tensor,at::Tensor> mps_convolution_transpose_backward(const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, ::std::array<bool,2> output_mask) {
+    return at::_ops::mps_convolution_transpose_backward::call(self, grad_output, weight, padding, output_padding, stride, dilation, groups, output_mask);
+  }
+}
+
+// aten::mps_convolution_transpose_backward.out(Tensor self, Tensor grad_output, Tensor weight, SymInt[] padding, SymInt[] output_padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool[2] output_mask, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> mps_convolution_transpose_backward_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef output_padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, ::std::array<bool,2> output_mask) {
+    return at::_ops::mps_convolution_transpose_backward_out::call(self, grad_output, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(output_padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, output_mask, out0, out1);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  ::std::tuple<at::Tensor &,at::Tensor &> mps_convolution_transpose_backward_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef output_padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, ::std::array<bool,2> output_mask) {
+    return at::_ops::mps_convolution_transpose_backward_out::call(self, grad_output, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(output_padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, output_mask, out0, out1);
+  }
+}
+
+// aten::mps_convolution_transpose_backward.out(Tensor self, Tensor grad_output, Tensor weight, SymInt[] padding, SymInt[] output_padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool[2] output_mask, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> mps_convolution_transpose_backward_outf(const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef output_padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, ::std::array<bool,2> output_mask, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::mps_convolution_transpose_backward_out::call(self, grad_output, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(output_padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, output_mask, out0, out1);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  ::std::tuple<at::Tensor &,at::Tensor &> mps_convolution_transpose_backward_outf(const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef output_padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, ::std::array<bool,2> output_mask, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::mps_convolution_transpose_backward_out::call(self, grad_output, weight, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(output_padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, output_mask, out0, out1);
+  }
+}
+
+// aten::mps_convolution_transpose_backward.out(Tensor self, Tensor grad_output, Tensor weight, SymInt[] padding, SymInt[] output_padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool[2] output_mask, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> mps_convolution_transpose_backward_symint_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, ::std::array<bool,2> output_mask) {
+    return at::_ops::mps_convolution_transpose_backward_out::call(self, grad_output, weight, padding, output_padding, stride, dilation, groups, output_mask, out0, out1);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  ::std::tuple<at::Tensor &,at::Tensor &> mps_convolution_transpose_backward_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, ::std::array<bool,2> output_mask) {
+    return at::_ops::mps_convolution_transpose_backward_out::call(self, grad_output, weight, padding, output_padding, stride, dilation, groups, output_mask, out0, out1);
+  }
+}
+
+// aten::mps_convolution_transpose_backward.out(Tensor self, Tensor grad_output, Tensor weight, SymInt[] padding, SymInt[] output_padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool[2] output_mask, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> mps_convolution_transpose_backward_symint_outf(const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, ::std::array<bool,2> output_mask, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::mps_convolution_transpose_backward_out::call(self, grad_output, weight, padding, output_padding, stride, dilation, groups, output_mask, out0, out1);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  ::std::tuple<at::Tensor &,at::Tensor &> mps_convolution_transpose_backward_outf(const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, ::std::array<bool,2> output_mask, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::mps_convolution_transpose_backward_out::call(self, grad_output, weight, padding, output_padding, stride, dilation, groups, output_mask, out0, out1);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mse_loss_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mse_loss_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..576ec0ca90dfbd6fe0304aee06c84b6a061ae8c7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mse_loss_backward_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API mse_loss_backward_grad_input {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const at::Tensor &, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mse_loss_backward";
+  static constexpr const char* overload_name = "grad_input";
+  static constexpr const char* schema_str = "mse_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, int reduction, *, Tensor(a!) grad_input) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction, at::Tensor & grad_input);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction, at::Tensor & grad_input);
+};
+
+struct TORCH_API mse_loss_backward {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mse_loss_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "mse_loss_backward(Tensor grad_output, Tensor self, Tensor target, int reduction) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mul_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mul_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..737652d17bbf53ce63caf0f3028ec45df8d18664
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mul_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor mul(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & mul_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & mul_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & mul_(at::Tensor & self, const at::Scalar & other);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mul_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mul_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4239e5c710d6a811f2642f80123c99056860fc2c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mul_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor mul(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & mul_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/multi_margin_loss_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/multi_margin_loss_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..7dd6a486c7ca80aad6d5567b380113a2b6a64d3a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/multi_margin_loss_backward.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/multi_margin_loss_backward_ops.h>
+
+namespace at {
+
+
+// aten::multi_margin_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Scalar p, Scalar margin, Tensor? weight=None, int reduction=Mean, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & multi_margin_loss_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const at::Scalar & p, const at::Scalar & margin, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean) {
+    return at::_ops::multi_margin_loss_backward_grad_input::call(grad_output, self, target, p, margin, weight, reduction, grad_input);
+}
+// aten::multi_margin_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Scalar p, Scalar margin, Tensor? weight=None, int reduction=Mean, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & multi_margin_loss_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const at::Scalar & p, const at::Scalar & margin, const ::std::optional<at::Tensor> & weight, int64_t reduction, at::Tensor & grad_input) {
+    return at::_ops::multi_margin_loss_backward_grad_input::call(grad_output, self, target, p, margin, weight, reduction, grad_input);
+}
+
+// aten::multi_margin_loss_backward(Tensor grad_output, Tensor self, Tensor target, Scalar p, Scalar margin, Tensor? weight=None, int reduction=Mean) -> Tensor
+inline at::Tensor multi_margin_loss_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const at::Scalar & p, const at::Scalar & margin, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean) {
+    return at::_ops::multi_margin_loss_backward::call(grad_output, self, target, p, margin, weight, reduction);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_loss.h b/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_loss.h
new file mode 100644
index 0000000000000000000000000000000000000000..5a3e5fefe0b042225db84850be7e9ba325b3282b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_loss.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/multilabel_margin_loss_ops.h>
+
+namespace at {
+
+
+// aten::multilabel_margin_loss.out(Tensor self, Tensor target, int reduction=Mean, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & multilabel_margin_loss_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & target, int64_t reduction=at::Reduction::Mean) {
+    return at::_ops::multilabel_margin_loss_out::call(self, target, reduction, out);
+}
+// aten::multilabel_margin_loss.out(Tensor self, Tensor target, int reduction=Mean, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & multilabel_margin_loss_outf(const at::Tensor & self, const at::Tensor & target, int64_t reduction, at::Tensor & out) {
+    return at::_ops::multilabel_margin_loss_out::call(self, target, reduction, out);
+}
+
+// aten::multilabel_margin_loss(Tensor self, Tensor target, int reduction=Mean) -> Tensor
+inline at::Tensor multilabel_margin_loss(const at::Tensor & self, const at::Tensor & target, int64_t reduction=at::Reduction::Mean) {
+    return at::_ops::multilabel_margin_loss::call(self, target, reduction);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mv_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mv_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a32cd8b5a8c3a5334fc95ff76380855cdb27c5c1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mv_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API mv {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mv";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "mv(Tensor self, Tensor vec) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & vec);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & vec);
+};
+
+struct TORCH_API mv_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mv";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "mv.out(Tensor self, Tensor vec, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & vec, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & vec, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nan_to_num_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nan_to_num_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..945c0b52230ba148b4afc1dd8c8341048c075fc8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nan_to_num_native.h
@@ -0,0 +1,26 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor nan_to_num(const at::Tensor & self, ::std::optional<double> nan=::std::nullopt, ::std::optional<double> posinf=::std::nullopt, ::std::optional<double> neginf=::std::nullopt);
+TORCH_API at::Tensor & nan_to_num_(at::Tensor & self, ::std::optional<double> nan=::std::nullopt, ::std::optional<double> posinf=::std::nullopt, ::std::optional<double> neginf=::std::nullopt);
+TORCH_API at::Tensor & nan_to_num_out(const at::Tensor & self, ::std::optional<double> nan, ::std::optional<double> posinf, ::std::optional<double> neginf, at::Tensor & out);
+TORCH_API at::Tensor nan_to_num_sparse(const at::Tensor & self, ::std::optional<double> nan=::std::nullopt, ::std::optional<double> posinf=::std::nullopt, ::std::optional<double> neginf=::std::nullopt);
+TORCH_API at::Tensor & nan_to_num_sparse_out(const at::Tensor & self, ::std::optional<double> nan, ::std::optional<double> posinf, ::std::optional<double> neginf, at::Tensor & out);
+TORCH_API at::Tensor & nan_to_num_sparse_(at::Tensor & self, ::std::optional<double> nan=::std::nullopt, ::std::optional<double> posinf=::std::nullopt, ::std::optional<double> neginf=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nanquantile_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nanquantile_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..b7161f4c767cb4da64f08338cfe8acbd8de7501a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nanquantile_ops.h
@@ -0,0 +1,61 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API nanquantile {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, ::std::optional<int64_t>, bool, c10::string_view);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::nanquantile";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "nanquantile(Tensor self, Tensor q, int? dim=None, bool keepdim=False, *, str interpolation='linear') -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation);
+};
+
+struct TORCH_API nanquantile_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, ::std::optional<int64_t>, bool, c10::string_view, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::nanquantile";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "nanquantile.out(Tensor self, Tensor q, int? dim=None, bool keepdim=False, *, str interpolation='linear', Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation, at::Tensor & out);
+};
+
+struct TORCH_API nanquantile_scalar {
+  using schema = at::Tensor (const at::Tensor &, double, ::std::optional<int64_t>, bool, c10::string_view);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::nanquantile";
+  static constexpr const char* overload_name = "scalar";
+  static constexpr const char* schema_str = "nanquantile.scalar(Tensor self, float q, int? dim=None, bool keepdim=False, *, str interpolation='linear') -> Tensor";
+  static at::Tensor call(const at::Tensor & self, double q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, double q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation);
+};
+
+struct TORCH_API nanquantile_scalar_out {
+  using schema = at::Tensor & (const at::Tensor &, double, ::std::optional<int64_t>, bool, c10::string_view, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::nanquantile";
+  static constexpr const char* overload_name = "scalar_out";
+  static constexpr const char* schema_str = "nanquantile.scalar_out(Tensor self, float q, int? dim=None, bool keepdim=False, *, str interpolation='linear', Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, double q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, double q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_batch_norm_backward_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_batch_norm_backward_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d5106188afa5820e538f9f1849ffa33bc6753800
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_batch_norm_backward_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_batch_norm_backward_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_batch_norm_backward_outf(const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_batch_norm_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_batch_norm_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..0565ced0be7dbde6751de13d1c4862b77dae2526
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_batch_norm_backward_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API native_batch_norm_backward {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, bool, double, ::std::array<bool,3>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::native_batch_norm_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "native_batch_norm_backward(Tensor grad_out, Tensor input, Tensor? weight, Tensor? running_mean, Tensor? running_var, Tensor? save_mean, Tensor? save_invstd, bool train, float eps, bool[3] output_mask) -> (Tensor, Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask);
+};
+
+struct TORCH_API native_batch_norm_backward_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, bool, double, ::std::array<bool,3>, at::Tensor &, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::native_batch_norm_backward";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "native_batch_norm_backward.out(Tensor grad_out, Tensor input, Tensor? weight, Tensor? running_mean, Tensor? running_var, Tensor? save_mean, Tensor? save_invstd, bool train, float eps, bool[3] output_mask, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> call(const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_layer_norm.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_layer_norm.h
new file mode 100644
index 0000000000000000000000000000000000000000..9b5ad1cffb703ab131640383504a4c8a999547fb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_layer_norm.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/native_layer_norm_ops.h>
+
+namespace at {
+
+
+// aten::native_layer_norm(Tensor input, SymInt[] normalized_shape, Tensor? weight, Tensor? bias, float eps) -> (Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> native_layer_norm(const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps) {
+    return at::_ops::native_layer_norm::call(input, c10::fromIntArrayRefSlow(normalized_shape), weight, bias, eps);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  ::std::tuple<at::Tensor,at::Tensor,at::Tensor> native_layer_norm(const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps) {
+    return at::_ops::native_layer_norm::call(input, c10::fromIntArrayRefSlow(normalized_shape), weight, bias, eps);
+  }
+}
+
+// aten::native_layer_norm(Tensor input, SymInt[] normalized_shape, Tensor? weight, Tensor? bias, float eps) -> (Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> native_layer_norm_symint(const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps) {
+    return at::_ops::native_layer_norm::call(input, normalized_shape, weight, bias, eps);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  ::std::tuple<at::Tensor,at::Tensor,at::Tensor> native_layer_norm(const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps) {
+    return at::_ops::native_layer_norm::call(input, normalized_shape, weight, bias, eps);
+  }
+}
+
+// aten::native_layer_norm.out(Tensor input, SymInt[] normalized_shape, Tensor? weight, Tensor? bias, float eps, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps) {
+    return at::_ops::native_layer_norm_out::call(input, c10::fromIntArrayRefSlow(normalized_shape), weight, bias, eps, out0, out1, out2);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps) {
+    return at::_ops::native_layer_norm_out::call(input, c10::fromIntArrayRefSlow(normalized_shape), weight, bias, eps, out0, out1, out2);
+  }
+}
+
+// aten::native_layer_norm.out(Tensor input, SymInt[] normalized_shape, Tensor? weight, Tensor? bias, float eps, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_outf(const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2) {
+    return at::_ops::native_layer_norm_out::call(input, c10::fromIntArrayRefSlow(normalized_shape), weight, bias, eps, out0, out1, out2);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_outf(const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2) {
+    return at::_ops::native_layer_norm_out::call(input, c10::fromIntArrayRefSlow(normalized_shape), weight, bias, eps, out0, out1, out2);
+  }
+}
+
+// aten::native_layer_norm.out(Tensor input, SymInt[] normalized_shape, Tensor? weight, Tensor? bias, float eps, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_symint_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps) {
+    return at::_ops::native_layer_norm_out::call(input, normalized_shape, weight, bias, eps, out0, out1, out2);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps) {
+    return at::_ops::native_layer_norm_out::call(input, normalized_shape, weight, bias, eps, out0, out1, out2);
+  }
+}
+
+// aten::native_layer_norm.out(Tensor input, SymInt[] normalized_shape, Tensor? weight, Tensor? bias, float eps, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_symint_outf(const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2) {
+    return at::_ops::native_layer_norm_out::call(input, normalized_shape, weight, bias, eps, out0, out1, out2);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_outf(const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2) {
+    return at::_ops::native_layer_norm_out::call(input, normalized_shape, weight, bias, eps, out0, out1, out2);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_layer_norm_backward_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_layer_norm_backward_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7cf97947c314780a7535f6acc05406040781334c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_layer_norm_backward_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_backward_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & grad_out, const at::Tensor & input, at::IntArrayRef normalized_shape, const at::Tensor & mean, const at::Tensor & rstd, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_backward_outf(const at::Tensor & grad_out, const at::Tensor & input, at::IntArrayRef normalized_shape, const at::Tensor & mean, const at::Tensor & rstd, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_backward_symint_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & grad_out, const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const at::Tensor & mean, const at::Tensor & rstd, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_backward_symint_outf(const at::Tensor & grad_out, const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const at::Tensor & mean, const at::Tensor & rstd, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_norm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_norm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..86f8cb25a1d4c6aa1a52ef21e6ffedcc246ad190
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_norm_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & native_norm_out(const at::Tensor & self, const at::Scalar & p, at::Tensor & out);
+TORCH_API at::Tensor norm_sparse(const at::Tensor & self, const at::Scalar & p=2);
+TORCH_API at::Tensor & native_norm_ScalarOpt_dim_dtype_out(const at::Tensor & self, const ::std::optional<at::Scalar> & p, at::IntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+TORCH_API at::Tensor norm_sparse(const at::Tensor & self, const ::std::optional<at::Scalar> & p, at::IntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/ne_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/ne_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..174579fb764599d2d273770cf9ef28254272c2c7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/ne_ops.h
@@ -0,0 +1,83 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API ne_Scalar_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::ne";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "ne.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+};
+
+struct TORCH_API ne_Scalar {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::ne";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "ne.Scalar(Tensor self, Scalar other) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Scalar & other);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & other);
+};
+
+struct TORCH_API ne_Tensor_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::ne";
+  static constexpr const char* overload_name = "Tensor_out";
+  static constexpr const char* schema_str = "ne.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+};
+
+struct TORCH_API ne_Tensor {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::ne";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "ne.Tensor(Tensor self, Tensor other) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & other);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other);
+};
+
+struct TORCH_API ne__Scalar {
+  using schema = at::Tensor & (at::Tensor &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::ne_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "ne_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Scalar & other);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Scalar & other);
+};
+
+struct TORCH_API ne__Tensor {
+  using schema = at::Tensor & (at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::ne_";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "ne_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & other);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & other);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/neg_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/neg_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5218e0f2a31f6bb178f98d074cb6f42795056b77
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/neg_cpu_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor neg(const at::Tensor & self);
+TORCH_API at::Tensor & neg_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & neg_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & neg_(at::Tensor & self);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/negative_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/negative_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..b12fb2037c047009276a2a36e8687896f72a71a4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/negative_ops.h
@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API negative {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::negative";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "negative(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+struct TORCH_API negative_ {
+  using schema = at::Tensor & (at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::negative_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "negative_(Tensor(a!) self) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self);
+};
+
+struct TORCH_API negative_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::negative";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "negative.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nested_to_padded_tensor_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nested_to_padded_tensor_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fdd0e93c65010abc7fccca2eb4865c560a615a7c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nested_to_padded_tensor_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor nested_to_padded_tensor(const at::Tensor & self, double padding, at::OptionalIntArrayRef output_size=::std::nullopt);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/new_empty_strided.h b/lib/python3.12/site-packages/torch/include/ATen/ops/new_empty_strided.h
new file mode 100644
index 0000000000000000000000000000000000000000..defe79f70132d212a49ebbafe543ef21995a252e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/new_empty_strided.h
@@ -0,0 +1,97 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/new_empty_strided_ops.h>
+
+namespace at {
+
+
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor new_empty_strided(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride, at::TensorOptions options={}) {
+    return at::_ops::new_empty_strided::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride), c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+  }
+}
+
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor new_empty_strided(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::new_empty_strided::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride), dtype, layout, device, pin_memory);
+  }
+}
+
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor new_empty_strided(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, at::TensorOptions options={}) {
+    return at::_ops::new_empty_strided::call(self, size, stride, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+  }
+}
+
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor new_empty_strided(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::new_empty_strided::call(self, size, stride, dtype, layout, device, pin_memory);
+  }
+}
+
+// aten::new_empty_strided.out(Tensor self, SymInt[] size, SymInt[] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & new_empty_strided_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride) {
+    return at::_ops::new_empty_strided_out::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride), out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & new_empty_strided_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride) {
+    return at::_ops::new_empty_strided_out::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride), out);
+  }
+}
+
+// aten::new_empty_strided.out(Tensor self, SymInt[] size, SymInt[] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & new_empty_strided_outf(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride, at::Tensor & out) {
+    return at::_ops::new_empty_strided_out::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride), out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & new_empty_strided_outf(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride, at::Tensor & out) {
+    return at::_ops::new_empty_strided_out::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride), out);
+  }
+}
+
+// aten::new_empty_strided.out(Tensor self, SymInt[] size, SymInt[] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & new_empty_strided_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride) {
+    return at::_ops::new_empty_strided_out::call(self, size, stride, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & new_empty_strided_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride) {
+    return at::_ops::new_empty_strided_out::call(self, size, stride, out);
+  }
+}
+
+// aten::new_empty_strided.out(Tensor self, SymInt[] size, SymInt[] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & new_empty_strided_symint_outf(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, at::Tensor & out) {
+    return at::_ops::new_empty_strided_out::call(self, size, stride, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & new_empty_strided_outf(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, at::Tensor & out) {
+    return at::_ops::new_empty_strided_out::call(self, size, stride, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss2d_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss2d_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..625ed9ea194c86e72b63af0fbd59cfd7abdef73d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss2d_backward.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/nll_loss2d_backward_ops.h>
+
+namespace at {
+
+
+// aten::nll_loss2d_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & nll_loss2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss2d_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & nll_loss2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss2d_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+  }
+}
+
+// aten::nll_loss2d_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & nll_loss2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight, at::Tensor & grad_input) {
+    return at::_ops::nll_loss2d_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & nll_loss2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight, at::Tensor & grad_input) {
+    return at::_ops::nll_loss2d_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+  }
+}
+
+// aten::nll_loss2d_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & nll_loss2d_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss2d_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & nll_loss2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss2d_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+  }
+}
+
+// aten::nll_loss2d_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & nll_loss2d_backward_symint_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight, at::Tensor & grad_input) {
+    return at::_ops::nll_loss2d_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & nll_loss2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight, at::Tensor & grad_input) {
+    return at::_ops::nll_loss2d_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+  }
+}
+
+// aten::nll_loss2d_backward(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight) -> Tensor
+inline at::Tensor nll_loss2d_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss2d_backward::call(grad_output, self, target, weight, reduction, ignore_index, total_weight);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor nll_loss2d_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss2d_backward::call(grad_output, self, target, weight, reduction, ignore_index, total_weight);
+  }
+}
+
+// aten::nll_loss2d_backward(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight) -> Tensor
+inline at::Tensor nll_loss2d_backward_symint(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss2d_backward::call(grad_output, self, target, weight, reduction, ignore_index, total_weight);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor nll_loss2d_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss2d_backward::call(grad_output, self, target, weight, reduction, ignore_index, total_weight);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_backward_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_backward_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..9d7568b8985f07f201eb0e6a090e302c60e5c458
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_backward_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_nll_loss_backward : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, at::OptionalTensorRef weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..a6c070f1aa8c4494865486e521f2eca7f3e695ae
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor nonzero_cpu(const at::Tensor & self);
+TORCH_API at::Tensor & nonzero_out_cpu(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor nonzero_cuda(const at::Tensor & self);
+TORCH_API at::Tensor & nonzero_out_cuda(const at::Tensor & self, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_static_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_static_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b9cd79be9e342ea43ccc8e1d992b33616d3394d4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_static_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor nonzero_static(const at::Tensor & self, int64_t size, int64_t fill_value=-1);
+TORCH_API at::Tensor & nonzero_static_out(at::Tensor & out, const at::Tensor & self, int64_t size, int64_t fill_value=-1);
+TORCH_API at::Tensor & nonzero_static_outf(const at::Tensor & self, int64_t size, int64_t fill_value, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_static_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_static_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..de248a8f52b1641d89deeb82fd0e9bb32f755375
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_static_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor nonzero_static_cpu(const at::Tensor & self, int64_t size, int64_t fill_value=-1);
+TORCH_API at::Tensor & nonzero_static_out_cpu(const at::Tensor & self, int64_t size, int64_t fill_value, at::Tensor & out);
+TORCH_API at::Tensor nonzero_static_cuda(const at::Tensor & self, int64_t size, int64_t fill_value=-1);
+TORCH_API at::Tensor & nonzero_static_out_cuda(const at::Tensor & self, int64_t size, int64_t fill_value, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/numpy_T_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/numpy_T_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..957070b01fd2b1d97dd3f736ec5b2472ed96a9c1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/numpy_T_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor numpy_T(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/orgqr_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/orgqr_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..416db33f4ae71f407a78e0c2630dcebcd878677e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/orgqr_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor orgqr(const at::Tensor & self, const at::Tensor & input2);
+TORCH_API at::Tensor & orgqr_out(const at::Tensor & self, const at::Tensor & input2, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/ormqr_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/ormqr_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1f9498221406fe9e29bb8ebd9fef9b2fc8a0d64b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/ormqr_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor ormqr(const at::Tensor & self, const at::Tensor & input2, const at::Tensor & input3, bool left=true, bool transpose=false);
+TORCH_API at::Tensor & ormqr_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & input2, const at::Tensor & input3, bool left=true, bool transpose=false);
+TORCH_API at::Tensor & ormqr_outf(const at::Tensor & self, const at::Tensor & input2, const at::Tensor & input3, bool left, bool transpose, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/pdist_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/pdist_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..66e1a0f74c99445ff59e1f00af048a8e1f4e6e70
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/pdist_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor pdist(const at::Tensor & self, double p=2);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/polar.h b/lib/python3.12/site-packages/torch/include/ATen/ops/polar.h
new file mode 100644
index 0000000000000000000000000000000000000000..643a6b11eb4550243ccd696a78b9008127980f42
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/polar.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/polar_ops.h>
+
+namespace at {
+
+
+// aten::polar(Tensor abs, Tensor angle) -> Tensor
+inline at::Tensor polar(const at::Tensor & abs, const at::Tensor & angle) {
+    return at::_ops::polar::call(abs, angle);
+}
+
+// aten::polar.out(Tensor abs, Tensor angle, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & polar_out(at::Tensor & out, const at::Tensor & abs, const at::Tensor & angle) {
+    return at::_ops::polar_out::call(abs, angle, out);
+}
+// aten::polar.out(Tensor abs, Tensor angle, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & polar_outf(const at::Tensor & abs, const at::Tensor & angle, at::Tensor & out) {
+    return at::_ops::polar_out::call(abs, angle, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/polygamma.h b/lib/python3.12/site-packages/torch/include/ATen/ops/polygamma.h
new file mode 100644
index 0000000000000000000000000000000000000000..94f409788e70e0159680b985fd412cdd6e395598
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/polygamma.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/polygamma_ops.h>
+
+namespace at {
+
+
+// aten::polygamma.out(int n, Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & polygamma_out(at::Tensor & out, int64_t n, const at::Tensor & self) {
+    return at::_ops::polygamma_out::call(n, self, out);
+}
+// aten::polygamma.out(int n, Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & polygamma_outf(int64_t n, const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::polygamma_out::call(n, self, out);
+}
+
+// aten::polygamma(int n, Tensor self) -> Tensor
+inline at::Tensor polygamma(int64_t n, const at::Tensor & self) {
+    return at::_ops::polygamma::call(n, self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/polygamma_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/polygamma_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0243e48e27c4f8d6cc2fe0f4619498d28892b5c9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/polygamma_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor polygamma(int64_t n, const at::Tensor & self);
+TORCH_API at::Tensor & polygamma_out(at::Tensor & out, int64_t n, const at::Tensor & self);
+TORCH_API at::Tensor & polygamma_outf(int64_t n, const at::Tensor & self, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/prod_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/prod_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..61d170c1db469d9ac1bfa5f68ca24b32b1f7303e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/prod_cuda_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor prod(const at::Tensor & self, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor prod(const at::Tensor & self, int64_t dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & prod_out(at::Tensor & out, const at::Tensor & self, int64_t dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & prod_outf(const at::Tensor & self, int64_t dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/prod_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/prod_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..085d66fbe5c4716e0fa2150352c7b26f07d9f441
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/prod_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_prod_dim_int : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, int64_t dim, bool keepdim, ::std::optional<at::ScalarType> dtype);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/prod_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/prod_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..f6a9338449de883d2de4d7f0314dbed679d62747
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/prod_native.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/prod_meta.h>
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & prod_out(const at::Tensor & self, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+TORCH_API at::Tensor prod(const at::Tensor & self, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+struct TORCH_API structured_prod_out : public at::meta::structured_prod_dim_int {
+void impl(const at::Tensor & self, int64_t dim, bool keepdim, ::std::optional<at::ScalarType> dtype, const at::Tensor & out);
+};
+TORCH_API at::Tensor prod(const at::Tensor & self, at::Dimname dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & prod_out(const at::Tensor & self, at::Dimname dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/q_zero_point_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/q_zero_point_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..41903dbce4c1f505cfa49ca820b1381dbf3301ae
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/q_zero_point_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API q_zero_point {
+  using schema = int64_t (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::q_zero_point";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "q_zero_point(Tensor self) -> int";
+  static int64_t call(const at::Tensor & self);
+  static int64_t redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..56fe1407a84f61ee0a9eeb9098488c7a866766cd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & quantize_per_tensor_out(at::Tensor & out, const at::Tensor & self, double scale, int64_t zero_point, at::ScalarType dtype);
+TORCH_API at::Tensor & quantize_per_tensor_outf(const at::Tensor & self, double scale, int64_t zero_point, at::ScalarType dtype, at::Tensor & out);
+TORCH_API at::Tensor & quantize_per_tensor_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, at::ScalarType dtype);
+TORCH_API at::Tensor & quantize_per_tensor_outf(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, at::ScalarType dtype, at::Tensor & out);
+TORCH_API void quantize_per_tensor_out(at::TensorList out, at::TensorList tensors, const at::Tensor & scales, const at::Tensor & zero_points, at::ScalarType dtype);
+TORCH_API void quantize_per_tensor_outf(at::TensorList tensors, const at::Tensor & scales, const at::Tensor & zero_points, at::ScalarType dtype, at::TensorList out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_batch_norm_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_batch_norm_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d4e2f514adecfd3684f12edcbc65845ed52b545e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_batch_norm_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & quantized_batch_norm_out(at::Tensor & out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & mean, const at::Tensor & var, double eps, double output_scale, int64_t output_zero_point);
+TORCH_API at::Tensor & quantized_batch_norm_outf(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & mean, const at::Tensor & var, double eps, double output_scale, int64_t output_zero_point, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_gru_cell.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_gru_cell.h
new file mode 100644
index 0000000000000000000000000000000000000000..3344edc7e693f1d12af422fd0027f98a402f2ae9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_gru_cell.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/quantized_gru_cell_ops.h>
+
+namespace at {
+
+
+// aten::quantized_gru_cell(Tensor input, Tensor hx, Tensor w_ih, Tensor w_hh, Tensor b_ih, Tensor b_hh, Tensor packed_ih, Tensor packed_hh, Tensor col_offsets_ih, Tensor col_offsets_hh, Scalar scale_ih, Scalar scale_hh, Scalar zero_point_ih, Scalar zero_point_hh) -> Tensor
+inline at::Tensor quantized_gru_cell(const at::Tensor & input, const at::Tensor & hx, const at::Tensor & w_ih, const at::Tensor & w_hh, const at::Tensor & b_ih, const at::Tensor & b_hh, const at::Tensor & packed_ih, const at::Tensor & packed_hh, const at::Tensor & col_offsets_ih, const at::Tensor & col_offsets_hh, const at::Scalar & scale_ih, const at::Scalar & scale_hh, const at::Scalar & zero_point_ih, const at::Scalar & zero_point_hh) {
+    return at::_ops::quantized_gru_cell::call(input, hx, w_ih, w_hh, b_ih, b_hh, packed_ih, packed_hh, col_offsets_ih, col_offsets_hh, scale_ih, scale_hh, zero_point_ih, zero_point_hh);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_max_pool3d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_max_pool3d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..70b4a840ccf0603b9c4a8ceae579d501f7bce3df
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_max_pool3d_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API quantized_max_pool3d {
+  using schema = at::Tensor (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::quantized_max_pool3d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "quantized_max_pool3d(Tensor self, int[3] kernel_size, int[3] stride=[], int[3] padding=0, int[3] dilation=1, bool ceil_mode=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+};
+
+struct TORCH_API quantized_max_pool3d_out {
+  using schema = at::Tensor & (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::quantized_max_pool3d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "quantized_max_pool3d.out(Tensor self, int[3] kernel_size, int[3] stride=[], int[3] padding=0, int[3] dilation=1, bool ceil_mode=False, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/randint_like_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/randint_like_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fa2997d1088fe907a66d9ced31fffee5f4f27697
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/randint_like_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,38 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor randint_like(const at::Tensor & self, int64_t high, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor randint_like(const at::Tensor & self, int64_t high, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+TORCH_API at::Tensor randint_like_symint(const at::Tensor & self, c10::SymInt high, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor randint_like_symint(const at::Tensor & self, c10::SymInt high, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+TORCH_API at::Tensor & randint_like_out(at::Tensor & out, const at::Tensor & self, int64_t high, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor & randint_like_outf(const at::Tensor & self, int64_t high, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+TORCH_API at::Tensor & randint_like_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymInt high, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor & randint_like_symint_outf(const at::Tensor & self, c10::SymInt high, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+TORCH_API at::Tensor randint_like(const at::Tensor & self, int64_t low, int64_t high, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor randint_like(const at::Tensor & self, int64_t low, int64_t high, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+TORCH_API at::Tensor randint_like_symint(const at::Tensor & self, c10::SymInt low, c10::SymInt high, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor randint_like_symint(const at::Tensor & self, c10::SymInt low, c10::SymInt high, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+TORCH_API at::Tensor & randint_like_out(at::Tensor & out, const at::Tensor & self, int64_t low, int64_t high, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor & randint_like_outf(const at::Tensor & self, int64_t low, int64_t high, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+TORCH_API at::Tensor & randint_like_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymInt low, c10::SymInt high, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor & randint_like_symint_outf(const at::Tensor & self, c10::SymInt low, c10::SymInt high, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/randperm.h b/lib/python3.12/site-packages/torch/include/ATen/ops/randperm.h
new file mode 100644
index 0000000000000000000000000000000000000000..cb0c385c40c9b0dc8da8d5e1b6605b402b5b1e83
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/randperm.h
@@ -0,0 +1,201 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/randperm_ops.h>
+
+namespace at {
+
+
+// aten::randperm(SymInt n, *, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor randperm(int64_t n, at::TensorOptions options=at::kLong) {
+    return at::_ops::randperm::call(n, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor randperm(int64_t n, at::TensorOptions options=at::kLong) {
+    return at::_ops::randperm::call(n, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+  }
+}
+
+// aten::randperm(SymInt n, *, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor randperm(int64_t n, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::randperm::call(n, dtype, layout, device, pin_memory);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor randperm(int64_t n, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::randperm::call(n, dtype, layout, device, pin_memory);
+  }
+}
+
+// aten::randperm(SymInt n, *, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor randperm_symint(c10::SymInt n, at::TensorOptions options=at::kLong) {
+    return at::_ops::randperm::call(n, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor randperm(c10::SymInt n, at::TensorOptions options=at::kLong) {
+    return at::_ops::randperm::call(n, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+  }
+}
+
+// aten::randperm(SymInt n, *, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor randperm_symint(c10::SymInt n, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::randperm::call(n, dtype, layout, device, pin_memory);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor randperm(c10::SymInt n, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::randperm::call(n, dtype, layout, device, pin_memory);
+  }
+}
+
+// aten::randperm.generator(SymInt n, *, Generator? generator, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor randperm(int64_t n, ::std::optional<at::Generator> generator, at::TensorOptions options=at::kLong) {
+    return at::_ops::randperm_generator::call(n, generator, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor randperm(int64_t n, ::std::optional<at::Generator> generator, at::TensorOptions options=at::kLong) {
+    return at::_ops::randperm_generator::call(n, generator, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+  }
+}
+
+// aten::randperm.generator(SymInt n, *, Generator? generator, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor randperm(int64_t n, ::std::optional<at::Generator> generator, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::randperm_generator::call(n, generator, dtype, layout, device, pin_memory);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor randperm(int64_t n, ::std::optional<at::Generator> generator, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::randperm_generator::call(n, generator, dtype, layout, device, pin_memory);
+  }
+}
+
+// aten::randperm.generator(SymInt n, *, Generator? generator, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor randperm_symint(c10::SymInt n, ::std::optional<at::Generator> generator, at::TensorOptions options=at::kLong) {
+    return at::_ops::randperm_generator::call(n, generator, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor randperm(c10::SymInt n, ::std::optional<at::Generator> generator, at::TensorOptions options=at::kLong) {
+    return at::_ops::randperm_generator::call(n, generator, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+  }
+}
+
+// aten::randperm.generator(SymInt n, *, Generator? generator, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor randperm_symint(c10::SymInt n, ::std::optional<at::Generator> generator, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::randperm_generator::call(n, generator, dtype, layout, device, pin_memory);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor randperm(c10::SymInt n, ::std::optional<at::Generator> generator, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::randperm_generator::call(n, generator, dtype, layout, device, pin_memory);
+  }
+}
+
+// aten::randperm.out(SymInt n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & randperm_out(at::Tensor & out, int64_t n) {
+    return at::_ops::randperm_out::call(n, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & randperm_out(at::Tensor & out, int64_t n) {
+    return at::_ops::randperm_out::call(n, out);
+  }
+}
+
+// aten::randperm.out(SymInt n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & randperm_outf(int64_t n, at::Tensor & out) {
+    return at::_ops::randperm_out::call(n, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & randperm_outf(int64_t n, at::Tensor & out) {
+    return at::_ops::randperm_out::call(n, out);
+  }
+}
+
+// aten::randperm.out(SymInt n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & randperm_symint_out(at::Tensor & out, c10::SymInt n) {
+    return at::_ops::randperm_out::call(n, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & randperm_out(at::Tensor & out, c10::SymInt n) {
+    return at::_ops::randperm_out::call(n, out);
+  }
+}
+
+// aten::randperm.out(SymInt n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & randperm_symint_outf(c10::SymInt n, at::Tensor & out) {
+    return at::_ops::randperm_out::call(n, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & randperm_outf(c10::SymInt n, at::Tensor & out) {
+    return at::_ops::randperm_out::call(n, out);
+  }
+}
+
+// aten::randperm.generator_out(SymInt n, *, Generator? generator, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & randperm_out(at::Tensor & out, int64_t n, ::std::optional<at::Generator> generator) {
+    return at::_ops::randperm_generator_out::call(n, generator, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & randperm_out(at::Tensor & out, int64_t n, ::std::optional<at::Generator> generator) {
+    return at::_ops::randperm_generator_out::call(n, generator, out);
+  }
+}
+
+// aten::randperm.generator_out(SymInt n, *, Generator? generator, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & randperm_outf(int64_t n, ::std::optional<at::Generator> generator, at::Tensor & out) {
+    return at::_ops::randperm_generator_out::call(n, generator, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & randperm_outf(int64_t n, ::std::optional<at::Generator> generator, at::Tensor & out) {
+    return at::_ops::randperm_generator_out::call(n, generator, out);
+  }
+}
+
+// aten::randperm.generator_out(SymInt n, *, Generator? generator, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & randperm_symint_out(at::Tensor & out, c10::SymInt n, ::std::optional<at::Generator> generator) {
+    return at::_ops::randperm_generator_out::call(n, generator, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & randperm_out(at::Tensor & out, c10::SymInt n, ::std::optional<at::Generator> generator) {
+    return at::_ops::randperm_generator_out::call(n, generator, out);
+  }
+}
+
+// aten::randperm.generator_out(SymInt n, *, Generator? generator, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & randperm_symint_outf(c10::SymInt n, ::std::optional<at::Generator> generator, at::Tensor & out) {
+    return at::_ops::randperm_generator_out::call(n, generator, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & randperm_outf(c10::SymInt n, ::std::optional<at::Generator> generator, at::Tensor & out) {
+    return at::_ops::randperm_generator_out::call(n, generator, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad3d_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad3d_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..030dc1628813c20df08b22d355187fa147f5042f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad3d_backward_cpu_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor reflection_pad3d_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor reflection_pad3d_backward_symint(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & reflection_pad3d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor & reflection_pad3d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding, at::Tensor & grad_input);
+TORCH_API at::Tensor & reflection_pad3d_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & reflection_pad3d_backward_symint_outf(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & grad_input);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/relu.h b/lib/python3.12/site-packages/torch/include/ATen/ops/relu.h
new file mode 100644
index 0000000000000000000000000000000000000000..deb7264c4ac23feb5bef55b9888bbd843669a7bc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/relu.h
@@ -0,0 +1,44 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/relu_ops.h>
+
+namespace at {
+
+
+// aten::relu(Tensor self) -> Tensor
+inline at::Tensor relu(const at::Tensor & self) {
+    return at::_ops::relu::call(self);
+}
+
+// aten::relu_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & relu_(at::Tensor & self) {
+    return at::_ops::relu_::call(self);
+}
+
+// aten::relu.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & relu_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::relu_out::call(self, out);
+}
+// aten::relu.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & relu_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::relu_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/remainder_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/remainder_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3c6bcbfc97fb839cb850562c37df29668341c78c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/remainder_cpu_dispatch.h
@@ -0,0 +1,27 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor remainder(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & remainder_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & remainder_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & remainder_(at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor remainder(const at::Scalar & self, const at::Tensor & other);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/remainder_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/remainder_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..590350e9cf15c4b92ab38b72ee1baaa3d61937e7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/remainder_native.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/remainder_meta.h>
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor remainder(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & remainder_out(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & remainder_(at::Tensor & self, const at::Scalar & other);
+struct TORCH_API structured_remainder_out : public at::meta::structured_remainder_Tensor {
+void impl(const at::Tensor & self, const at::Tensor & other, const at::Tensor & out);
+};
+TORCH_API at::Tensor & remainder_Scalar_Tensor_out(const at::Scalar & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor remainder(const at::Scalar & self, const at::Tensor & other);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad2d_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad2d_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a43cb2d9bf18f98bcfc9068fe76099759e4b232c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad2d_cpu_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor replication_pad2d(const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor replication_pad2d_symint(const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & replication_pad2d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor & replication_pad2d_outf(const at::Tensor & self, at::IntArrayRef padding, at::Tensor & out);
+TORCH_API at::Tensor & replication_pad2d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & replication_pad2d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad2d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad2d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..ae3956b2e8ad92c2b529159f3882645d7a1ea6af
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad2d_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API replication_pad2d_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::replication_pad2d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "replication_pad2d.out(Tensor self, SymInt[4] padding, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & out);
+};
+
+struct TORCH_API replication_pad2d {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::replication_pad2d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "replication_pad2d(Tensor self, SymInt[4] padding) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef padding);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef padding);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/resolve_neg.h b/lib/python3.12/site-packages/torch/include/ATen/ops/resolve_neg.h
new file mode 100644
index 0000000000000000000000000000000000000000..90f45b4260f7bab5a4ddd6903f306608eb68a479
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/resolve_neg.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/resolve_neg_ops.h>
+
+namespace at {
+
+
+// aten::resolve_neg(Tensor(a) self) -> Tensor(a)
+inline at::Tensor resolve_neg(const at::Tensor & self) {
+    return at::_ops::resolve_neg::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rnn_relu_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rnn_relu_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..07e64d539f183dbb59ff3b8333fc0114043854e6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rnn_relu_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> rnn_relu(const at::Tensor & input, const at::Tensor & hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional, bool batch_first);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> rnn_relu(const at::Tensor & data, const at::Tensor & batch_sizes, const at::Tensor & hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rnn_tanh_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rnn_tanh_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9a47cd5c51b295b99711242b02cfc5006bcf5ff9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rnn_tanh_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> rnn_tanh(const at::Tensor & input, const at::Tensor & hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional, bool batch_first);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> rnn_tanh(const at::Tensor & data, const at::Tensor & batch_sizes, const at::Tensor & hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rot90_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rot90_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..56228fe257c6204106610b4ae8242c3db76e98d0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rot90_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor rot90(const at::Tensor & self, int64_t k=1, at::IntArrayRef dims={0,1});
+TORCH_API at::Tensor & rot90_out(at::Tensor & out, const at::Tensor & self, int64_t k=1, at::IntArrayRef dims={0,1});
+TORCH_API at::Tensor & rot90_outf(const at::Tensor & self, int64_t k, at::IntArrayRef dims, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/round_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/round_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..dd43946208ce0e60a09444ea4bb78c4b3ccb21dc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/round_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor round(const at::Tensor & self);
+TORCH_API at::Tensor & round_(at::Tensor & self);
+TORCH_API at::Tensor round(const at::Tensor & self, int64_t decimals);
+TORCH_API at::Tensor & round_(at::Tensor & self, int64_t decimals);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rshift.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rshift.h
new file mode 100644
index 0000000000000000000000000000000000000000..5e457a2885253bb6041bdd49ae34a4b1731452a2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rshift.h
@@ -0,0 +1,53 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/rshift_ops.h>
+
+namespace at {
+
+
+// aten::__rshift__.Scalar(Tensor self, Scalar other) -> Tensor
+inline at::Tensor __rshift__(const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::__rshift___Scalar::call(self, other);
+}
+
+// aten::__rshift__.Tensor(Tensor self, Tensor other) -> Tensor
+inline at::Tensor __rshift__(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::__rshift___Tensor::call(self, other);
+}
+
+// aten::__rshift__.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & __rshift___out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::__rshift___Scalar_out::call(self, other, out);
+}
+// aten::__rshift__.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & __rshift___outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out) {
+    return at::_ops::__rshift___Scalar_out::call(self, other, out);
+}
+
+// aten::__rshift__.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & __rshift___out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::__rshift___Tensor_out::call(self, other, out);
+}
+// aten::__rshift__.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & __rshift___outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::__rshift___Tensor_out::call(self, other, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/scalar_tensor_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/scalar_tensor_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..61572582bc11b4f3cac1d45c5265c2e5e1decbd9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/scalar_tensor_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API scalar_tensor {
+  using schema = at::Tensor (const at::Scalar &, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::scalar_tensor";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "scalar_tensor(Scalar s, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(const at::Scalar & s, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & s, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API scalar_tensor_out {
+  using schema = at::Tensor & (const at::Scalar &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::scalar_tensor";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "scalar_tensor.out(Scalar s, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Scalar & s, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & s, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/searchsorted_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/searchsorted_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..49e23d83dc4ee801ee8912ac5c40dcac60433e33
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/searchsorted_ops.h
@@ -0,0 +1,61 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API searchsorted_Tensor {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, bool, bool, ::std::optional<c10::string_view>, const ::std::optional<at::Tensor> &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::searchsorted";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "searchsorted.Tensor(Tensor sorted_sequence, Tensor self, *, bool out_int32=False, bool right=False, str? side=None, Tensor? sorter=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & sorted_sequence, const at::Tensor & self, bool out_int32, bool right, ::std::optional<c10::string_view> side, const ::std::optional<at::Tensor> & sorter);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & sorted_sequence, const at::Tensor & self, bool out_int32, bool right, ::std::optional<c10::string_view> side, const ::std::optional<at::Tensor> & sorter);
+};
+
+struct TORCH_API searchsorted_Tensor_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, bool, bool, ::std::optional<c10::string_view>, const ::std::optional<at::Tensor> &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::searchsorted";
+  static constexpr const char* overload_name = "Tensor_out";
+  static constexpr const char* schema_str = "searchsorted.Tensor_out(Tensor sorted_sequence, Tensor self, *, bool out_int32=False, bool right=False, str? side=None, Tensor? sorter=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & sorted_sequence, const at::Tensor & self, bool out_int32, bool right, ::std::optional<c10::string_view> side, const ::std::optional<at::Tensor> & sorter, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & sorted_sequence, const at::Tensor & self, bool out_int32, bool right, ::std::optional<c10::string_view> side, const ::std::optional<at::Tensor> & sorter, at::Tensor & out);
+};
+
+struct TORCH_API searchsorted_Scalar {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &, bool, bool, ::std::optional<c10::string_view>, const ::std::optional<at::Tensor> &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::searchsorted";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "searchsorted.Scalar(Tensor sorted_sequence, Scalar self, *, bool out_int32=False, bool right=False, str? side=None, Tensor? sorter=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & sorted_sequence, const at::Scalar & self, bool out_int32, bool right, ::std::optional<c10::string_view> side, const ::std::optional<at::Tensor> & sorter);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & sorted_sequence, const at::Scalar & self, bool out_int32, bool right, ::std::optional<c10::string_view> side, const ::std::optional<at::Tensor> & sorter);
+};
+
+struct TORCH_API searchsorted_Scalar_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, bool, bool, ::std::optional<c10::string_view>, const ::std::optional<at::Tensor> &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::searchsorted";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "searchsorted.Scalar_out(Tensor sorted_sequence, Scalar self, *, bool out_int32=False, bool right=False, str? side=None, Tensor? sorter=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & sorted_sequence, const at::Scalar & self, bool out_int32, bool right, ::std::optional<c10::string_view> side, const ::std::optional<at::Tensor> & sorter, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & sorted_sequence, const at::Scalar & self, bool out_int32, bool right, ::std::optional<c10::string_view> side, const ::std::optional<at::Tensor> & sorter, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/segment_reduce_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/segment_reduce_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..01effadc9cecdf13642e480ff0444426db4a08fd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/segment_reduce_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor segment_reduce(const at::Tensor & data, c10::string_view reduce, const ::std::optional<at::Tensor> & lengths={}, const ::std::optional<at::Tensor> & indices={}, const ::std::optional<at::Tensor> & offsets={}, int64_t axis=0, bool unsafe=false, const ::std::optional<at::Scalar> & initial=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/segment_reduce_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/segment_reduce_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..5a4fd8fa7698b9aa227bd0b7dea741a7d740daa1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/segment_reduce_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API segment_reduce {
+  using schema = at::Tensor (const at::Tensor &, c10::string_view, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, int64_t, bool, const ::std::optional<at::Scalar> &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::segment_reduce";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "segment_reduce(Tensor data, str reduce, *, Tensor? lengths=None, Tensor? indices=None, Tensor? offsets=None, int axis=0, bool unsafe=False, Scalar? initial=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & data, c10::string_view reduce, const ::std::optional<at::Tensor> & lengths, const ::std::optional<at::Tensor> & indices, const ::std::optional<at::Tensor> & offsets, int64_t axis, bool unsafe, const ::std::optional<at::Scalar> & initial);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & data, c10::string_view reduce, const ::std::optional<at::Tensor> & lengths, const ::std::optional<at::Tensor> & indices, const ::std::optional<at::Tensor> & offsets, int64_t axis, bool unsafe, const ::std::optional<at::Scalar> & initial);
+};
+
+struct TORCH_API segment_reduce_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::string_view, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, int64_t, bool, const ::std::optional<at::Scalar> &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::segment_reduce";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "segment_reduce.out(Tensor data, str reduce, *, Tensor? lengths=None, Tensor? indices=None, Tensor? offsets=None, int axis=0, bool unsafe=False, Scalar? initial=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & data, c10::string_view reduce, const ::std::optional<at::Tensor> & lengths, const ::std::optional<at::Tensor> & indices, const ::std::optional<at::Tensor> & offsets, int64_t axis, bool unsafe, const ::std::optional<at::Scalar> & initial, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & data, c10::string_view reduce, const ::std::optional<at::Tensor> & lengths, const ::std::optional<at::Tensor> & indices, const ::std::optional<at::Tensor> & offsets, int64_t axis, bool unsafe, const ::std::optional<at::Scalar> & initial, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/select_scatter_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/select_scatter_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..7ca329bc060e6780a3e020d3bf61c94972880fcc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/select_scatter_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API select_scatter {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, int64_t, c10::SymInt);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::select_scatter";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "select_scatter(Tensor self, Tensor src, int dim, SymInt index) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & src, int64_t dim, c10::SymInt index);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & src, int64_t dim, c10::SymInt index);
+};
+
+struct TORCH_API select_scatter_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, int64_t, c10::SymInt, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::select_scatter";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "select_scatter.out(Tensor self, Tensor src, int dim, SymInt index, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & src, int64_t dim, c10::SymInt index, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & src, int64_t dim, c10::SymInt index, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..7cc3d056dd4a43cbaa84b1766f7feeca81968e81
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_backward_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/sigmoid_backward_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_sigmoid_backward_out : public at::meta::structured_sigmoid_backward {
+void impl(const at::Tensor & grad_output, const at::Tensor & output, const at::Tensor & grad_input);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slice_backward_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_backward_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a00669c0a330447fca405ce9743c8ad161077cc2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_backward_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor slice_backward(const at::Tensor & grad_output, at::IntArrayRef input_sizes, int64_t dim, int64_t start, int64_t end, int64_t step);
+TORCH_API at::Tensor slice_backward_symint(const at::Tensor & grad_output, c10::SymIntArrayRef input_sizes, int64_t dim, c10::SymInt start, c10::SymInt end, c10::SymInt step);
+TORCH_API at::Tensor & slice_backward_out(at::Tensor & out, const at::Tensor & grad_output, at::IntArrayRef input_sizes, int64_t dim, int64_t start, int64_t end, int64_t step);
+TORCH_API at::Tensor & slice_backward_outf(const at::Tensor & grad_output, at::IntArrayRef input_sizes, int64_t dim, int64_t start, int64_t end, int64_t step, at::Tensor & out);
+TORCH_API at::Tensor & slice_backward_symint_out(at::Tensor & out, const at::Tensor & grad_output, c10::SymIntArrayRef input_sizes, int64_t dim, c10::SymInt start, c10::SymInt end, c10::SymInt step);
+TORCH_API at::Tensor & slice_backward_symint_outf(const at::Tensor & grad_output, c10::SymIntArrayRef input_sizes, int64_t dim, c10::SymInt start, c10::SymInt end, c10::SymInt step, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_dilated2d_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_dilated2d_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..399d1e36bafdbdfbc0f79d5f4efb273e4a7c298d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_dilated2d_cuda_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor slow_conv_dilated2d(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias={}, at::IntArrayRef stride=1, at::IntArrayRef padding=0, at::IntArrayRef dilation=1);
+TORCH_API at::Tensor slow_conv_dilated2d_symint(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef padding=c10::SymInt(0), c10::SymIntArrayRef dilation=c10::SymInt(1));
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_dilated2d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_dilated2d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..06afe03ffe51a4130cf4f40888aaa25ac5ad92ff
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_dilated2d_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API slow_conv_dilated2d {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, c10::SymIntArrayRef, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::slow_conv_dilated2d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "slow_conv_dilated2d(Tensor self, Tensor weight, SymInt[2] kernel_size, Tensor? bias=None, SymInt[2] stride=1, SymInt[2] padding=0, SymInt[2] dilation=1) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation);
+};
+
+struct TORCH_API slow_conv_dilated2d_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, c10::SymIntArrayRef, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::slow_conv_dilated2d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "slow_conv_dilated2d.out(Tensor self, Tensor weight, SymInt[2] kernel_size, Tensor? bias=None, SymInt[2] stride=1, SymInt[2] padding=0, SymInt[2] dilation=1, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/smooth_l1_loss_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/smooth_l1_loss_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..13b3b9b45fff4f24a3a6a62fe137e3f6cbe1272b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/smooth_l1_loss_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor smooth_l1_loss(const at::Tensor & self, const at::Tensor & target, int64_t reduction=at::Reduction::Mean, double beta=1.0);
+TORCH_API at::Tensor & smooth_l1_loss_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & target, int64_t reduction=at::Reduction::Mean, double beta=1.0);
+TORCH_API at::Tensor & smooth_l1_loss_outf(const at::Tensor & self, const at::Tensor & target, int64_t reduction, double beta, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/softplus_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/softplus_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ad4380868908401b3764570aefeb27d1bf7e8cee
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/softplus_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor softplus(const at::Tensor & self, const at::Scalar & beta=1, const at::Scalar & threshold=20);
+TORCH_API at::Tensor & softplus_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & beta=1, const at::Scalar & threshold=20);
+TORCH_API at::Tensor & softplus_outf(const at::Tensor & self, const at::Scalar & beta, const at::Scalar & threshold, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/softshrink_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/softshrink_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..caebd9a5cee5115a09a8fc5da811f14d691f24c0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/softshrink_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor softshrink(const at::Tensor & self, const at::Scalar & lambd=0.5);
+TORCH_API at::Tensor & softshrink_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & lambd=0.5);
+TORCH_API at::Tensor & softshrink_outf(const at::Tensor & self, const at::Scalar & lambd, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_bsr_tensor.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_bsr_tensor.h
new file mode 100644
index 0000000000000000000000000000000000000000..246e303d33f07741efdde2f8aa04bc9b4ca04fb4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_bsr_tensor.h
@@ -0,0 +1,43 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/sparse_bsr_tensor_ops.h>
+
+namespace at {
+
+
+// aten::sparse_bsr_tensor.crow_col_value_size(Tensor crow_indices, Tensor col_indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_bsr_tensor(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, at::IntArrayRef size, at::TensorOptions options) {
+    return at::_ops::sparse_bsr_tensor_crow_col_value_size::call(crow_indices, col_indices, values, size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::sparse_bsr_tensor.crow_col_value_size(Tensor crow_indices, Tensor col_indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_bsr_tensor(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::sparse_bsr_tensor_crow_col_value_size::call(crow_indices, col_indices, values, size, dtype, layout, device, pin_memory);
+}
+
+// aten::sparse_bsr_tensor.crow_col_value(Tensor crow_indices, Tensor col_indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_bsr_tensor(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, at::TensorOptions options) {
+    return at::_ops::sparse_bsr_tensor_crow_col_value::call(crow_indices, col_indices, values, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::sparse_bsr_tensor.crow_col_value(Tensor crow_indices, Tensor col_indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_bsr_tensor(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::sparse_bsr_tensor_crow_col_value::call(crow_indices, col_indices, values, dtype, layout, device, pin_memory);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_coo_tensor_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_coo_tensor_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f16fea9499bc905f03540c3fad63b4d14549cacd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_coo_tensor_ops.h
@@ -0,0 +1,61 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API sparse_coo_tensor_size {
+  using schema = at::Tensor (at::IntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sparse_coo_tensor";
+  static constexpr const char* overload_name = "size";
+  static constexpr const char* schema_str = "sparse_coo_tensor.size(int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor";
+  static at::Tensor call(at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API sparse_coo_tensor_indices {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sparse_coo_tensor";
+  static constexpr const char* overload_name = "indices";
+  static constexpr const char* schema_str = "sparse_coo_tensor.indices(Tensor indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, bool? is_coalesced=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & indices, const at::Tensor & values, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<bool> is_coalesced);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & indices, const at::Tensor & values, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<bool> is_coalesced);
+};
+
+struct TORCH_API sparse_coo_tensor_indices_size {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, at::IntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sparse_coo_tensor";
+  static constexpr const char* overload_name = "indices_size";
+  static constexpr const char* schema_str = "sparse_coo_tensor.indices_size(Tensor indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, bool? is_coalesced=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<bool> is_coalesced);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<bool> is_coalesced);
+};
+
+struct TORCH_API sparse_coo_tensor_size_out {
+  using schema = at::Tensor & (at::IntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sparse_coo_tensor";
+  static constexpr const char* overload_name = "size_out";
+  static constexpr const char* schema_str = "sparse_coo_tensor.size_out(int[] size, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(at::IntArrayRef size, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::IntArrayRef size, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csc_tensor_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csc_tensor_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..993184371c3e2a5ce7939c8cd7d55d2e4ad43bb6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csc_tensor_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor sparse_csc_tensor(const at::Tensor & ccol_indices, const at::Tensor & row_indices, const at::Tensor & values, at::IntArrayRef size, at::TensorOptions options);
+TORCH_API at::Tensor sparse_csc_tensor(const at::Tensor & ccol_indices, const at::Tensor & row_indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor sparse_csc_tensor(const at::Tensor & ccol_indices, const at::Tensor & row_indices, const at::Tensor & values, at::TensorOptions options);
+TORCH_API at::Tensor sparse_csc_tensor(const at::Tensor & ccol_indices, const at::Tensor & row_indices, const at::Tensor & values, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csr_tensor_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csr_tensor_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..27f2f1a18fd032a758af78a388b86f978e640ec9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csr_tensor_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API sparse_csr_tensor_crow_col_value_size {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, at::IntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sparse_csr_tensor";
+  static constexpr const char* overload_name = "crow_col_value_size";
+  static constexpr const char* schema_str = "sparse_csr_tensor.crow_col_value_size(Tensor crow_indices, Tensor col_indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API sparse_csr_tensor_crow_col_value {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sparse_csr_tensor";
+  static constexpr const char* overload_name = "crow_col_value";
+  static constexpr const char* schema_str = "sparse_csr_tensor.crow_col_value(Tensor crow_indices, Tensor col_indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_resize_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_resize_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fbdaa39b990e2d4aecadda81a29656cb32032394
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_resize_meta_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API const at::Tensor & sparse_resize_(const at::Tensor & self, at::IntArrayRef size, int64_t sparse_dim, int64_t dense_dim);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_bessel_j1_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_bessel_j1_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d1ef0be50368e58adb9a41baa8d1148cdf47cddc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_bessel_j1_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/special_bessel_j1_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_bessel_j1_out : public at::meta::structured_special_bessel_j1 {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_u_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_u_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..48da479f327d6256731f1e1a009871d981172221
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_u_ops.h
@@ -0,0 +1,83 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API special_chebyshev_polynomial_u {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_chebyshev_polynomial_u";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_chebyshev_polynomial_u(Tensor x, Tensor n) -> Tensor";
+  static at::Tensor call(const at::Tensor & x, const at::Tensor & n);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & x, const at::Tensor & n);
+};
+
+struct TORCH_API special_chebyshev_polynomial_u_x_scalar {
+  using schema = at::Tensor (const at::Scalar &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_chebyshev_polynomial_u";
+  static constexpr const char* overload_name = "x_scalar";
+  static constexpr const char* schema_str = "special_chebyshev_polynomial_u.x_scalar(Scalar x, Tensor n) -> Tensor";
+  static at::Tensor call(const at::Scalar & x, const at::Tensor & n);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & x, const at::Tensor & n);
+};
+
+struct TORCH_API special_chebyshev_polynomial_u_n_scalar {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_chebyshev_polynomial_u";
+  static constexpr const char* overload_name = "n_scalar";
+  static constexpr const char* schema_str = "special_chebyshev_polynomial_u.n_scalar(Tensor x, Scalar n) -> Tensor";
+  static at::Tensor call(const at::Tensor & x, const at::Scalar & n);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & x, const at::Scalar & n);
+};
+
+struct TORCH_API special_chebyshev_polynomial_u_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_chebyshev_polynomial_u";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_chebyshev_polynomial_u.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & x, const at::Tensor & n, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & x, const at::Tensor & n, at::Tensor & out);
+};
+
+struct TORCH_API special_chebyshev_polynomial_u_x_scalar_out {
+  using schema = at::Tensor & (const at::Scalar &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_chebyshev_polynomial_u";
+  static constexpr const char* overload_name = "x_scalar_out";
+  static constexpr const char* schema_str = "special_chebyshev_polynomial_u.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Scalar & x, const at::Tensor & n, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & x, const at::Tensor & n, at::Tensor & out);
+};
+
+struct TORCH_API special_chebyshev_polynomial_u_n_scalar_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_chebyshev_polynomial_u";
+  static constexpr const char* overload_name = "n_scalar_out";
+  static constexpr const char* schema_str = "special_chebyshev_polynomial_u.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & x, const at::Scalar & n, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & x, const at::Scalar & n, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_erfinv_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_erfinv_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f013747c084c3db2399f9e117554b5319bebca87
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_erfinv_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor special_erfinv(const at::Tensor & self);
+TORCH_API at::Tensor & special_erfinv_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & special_erfinv_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_gammaln_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_gammaln_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..35c76732e1546eedc14f3928d35f01258e07cb41
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_gammaln_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API special_gammaln {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_gammaln";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_gammaln(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+struct TORCH_API special_gammaln_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_gammaln";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_gammaln.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_hermite_polynomial_he_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_hermite_polynomial_he_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..80e76c21f6bb7e72aea22d5b50aee1b4a50787fe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_hermite_polynomial_he_ops.h
@@ -0,0 +1,83 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API special_hermite_polynomial_he {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_hermite_polynomial_he";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_hermite_polynomial_he(Tensor x, Tensor n) -> Tensor";
+  static at::Tensor call(const at::Tensor & x, const at::Tensor & n);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & x, const at::Tensor & n);
+};
+
+struct TORCH_API special_hermite_polynomial_he_x_scalar {
+  using schema = at::Tensor (const at::Scalar &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_hermite_polynomial_he";
+  static constexpr const char* overload_name = "x_scalar";
+  static constexpr const char* schema_str = "special_hermite_polynomial_he.x_scalar(Scalar x, Tensor n) -> Tensor";
+  static at::Tensor call(const at::Scalar & x, const at::Tensor & n);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & x, const at::Tensor & n);
+};
+
+struct TORCH_API special_hermite_polynomial_he_n_scalar {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_hermite_polynomial_he";
+  static constexpr const char* overload_name = "n_scalar";
+  static constexpr const char* schema_str = "special_hermite_polynomial_he.n_scalar(Tensor x, Scalar n) -> Tensor";
+  static at::Tensor call(const at::Tensor & x, const at::Scalar & n);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & x, const at::Scalar & n);
+};
+
+struct TORCH_API special_hermite_polynomial_he_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_hermite_polynomial_he";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_hermite_polynomial_he.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & x, const at::Tensor & n, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & x, const at::Tensor & n, at::Tensor & out);
+};
+
+struct TORCH_API special_hermite_polynomial_he_x_scalar_out {
+  using schema = at::Tensor & (const at::Scalar &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_hermite_polynomial_he";
+  static constexpr const char* overload_name = "x_scalar_out";
+  static constexpr const char* schema_str = "special_hermite_polynomial_he.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Scalar & x, const at::Tensor & n, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & x, const at::Tensor & n, at::Tensor & out);
+};
+
+struct TORCH_API special_hermite_polynomial_he_n_scalar_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_hermite_polynomial_he";
+  static constexpr const char* overload_name = "n_scalar_out";
+  static constexpr const char* schema_str = "special_hermite_polynomial_he.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & x, const at::Scalar & n, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & x, const at::Scalar & n, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_i1_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i1_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8cb59a021a937603b3f311fef6060131decd0afe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i1_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor special_i1(const at::Tensor & self);
+TORCH_API at::Tensor & special_i1_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & special_i1_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_laguerre_polynomial_l_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_laguerre_polynomial_l_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..875b1dd39a9e278a96ffe7087852fffbb09c1923
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_laguerre_polynomial_l_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor special_laguerre_polynomial_l(const at::Tensor & x, const at::Tensor & n);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_legendre_polynomial_p_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_legendre_polynomial_p_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..79db3964d4cc831551a4f44a00a07493b2a66881
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_legendre_polynomial_p_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor special_legendre_polynomial_p(const at::Tensor & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_legendre_polynomial_p_out(at::Tensor & out, const at::Tensor & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_legendre_polynomial_p_outf(const at::Tensor & x, const at::Tensor & n, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_logsumexp.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_logsumexp.h
new file mode 100644
index 0000000000000000000000000000000000000000..35db8df77d7b04a882deb17ce65af58ef8dc38d4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_logsumexp.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/special_logsumexp_ops.h>
+
+namespace at {
+
+
+// aten::special_logsumexp(Tensor self, int[1] dim, bool keepdim=False) -> Tensor
+inline at::Tensor special_logsumexp(const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false) {
+    return at::_ops::special_logsumexp::call(self, dim, keepdim);
+}
+
+// aten::special_logsumexp.out(Tensor self, int[1] dim, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_logsumexp_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false) {
+    return at::_ops::special_logsumexp_out::call(self, dim, keepdim, out);
+}
+// aten::special_logsumexp.out(Tensor self, int[1] dim, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_logsumexp_outf(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, at::Tensor & out) {
+    return at::_ops::special_logsumexp_out::call(self, dim, keepdim, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_i0_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_i0_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7146ca6a736b1bd7c5df0541272b1135786ede8b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_i0_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor special_modified_bessel_i0(const at::Tensor & self);
+TORCH_API at::Tensor & special_modified_bessel_i0_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & special_modified_bessel_i0_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_i1_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_i1_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..3e47a58cd693667194f5e024df2dc3ceb38a08e7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_i1_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/special_modified_bessel_i1_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_modified_bessel_i1_out : public at::meta::structured_special_modified_bessel_i1 {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtri_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtri_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8126de2b73de55c42f2faf42dfb5924e18513522
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtri_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/special_ndtri_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_ndtri_out : public at::meta::structured_special_ndtri {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k1_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k1_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0789c669cda9f72630a662514e2f247ad9457457
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k1_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor special_scaled_modified_bessel_k1(const at::Tensor & x);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_u_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_u_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..6082131de9cf787f8140a1ae18383fb0441de9ea
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_u_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_special_shifted_chebyshev_polynomial_u : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & x, const at::Tensor & n);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_w_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_w_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..37cb67f3f134e4f709d9725baf751f262dba44b7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_w_native.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/special_shifted_chebyshev_polynomial_w_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_shifted_chebyshev_polynomial_w_out : public at::meta::structured_special_shifted_chebyshev_polynomial_w {
+void impl(const at::Tensor & x, const at::Tensor & n, const at::Tensor & out);
+};
+TORCH_API at::Tensor special_shifted_chebyshev_polynomial_w(const at::Scalar & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_shifted_chebyshev_polynomial_w_out(const at::Scalar & x, const at::Tensor & n, at::Tensor & out);
+TORCH_API at::Tensor special_shifted_chebyshev_polynomial_w(const at::Tensor & x, const at::Scalar & n);
+TORCH_API at::Tensor & special_shifted_chebyshev_polynomial_w_out(const at::Tensor & x, const at::Scalar & n, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_spherical_bessel_j0_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_spherical_bessel_j0_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b4264836072be955d962661e47f684bce7a8ef49
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_spherical_bessel_j0_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor special_spherical_bessel_j0(const at::Tensor & x);
+TORCH_API at::Tensor & special_spherical_bessel_j0_out(at::Tensor & out, const at::Tensor & x);
+TORCH_API at::Tensor & special_spherical_bessel_j0_outf(const at::Tensor & x, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_spherical_bessel_j0_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_spherical_bessel_j0_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..31ef5d6fb71f772eadf1f2ef76b7b45859af8a06
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_spherical_bessel_j0_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/special_spherical_bessel_j0_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_spherical_bessel_j0_out : public at::meta::structured_special_spherical_bessel_j0 {
+void impl(const at::Tensor & x, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_zeta_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_zeta_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..dbaa2ea7751a49ac5fe43b690948b675d02bdadc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_zeta_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor special_zeta(const at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_zeta_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_zeta_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..047a25d5eeed6446e96653f7ab461f18c8bee9d2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_zeta_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor special_zeta(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & special_zeta_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & special_zeta_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/split_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/split_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..e70bed95ab68f62979b745aad259d0e2f48166cb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/split_copy.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/split_copy_ops.h>
+
+namespace at {
+
+
+// aten::split_copy.Tensor(Tensor self, SymInt split_size, int dim=0) -> Tensor[]
+inline ::std::vector<at::Tensor> split_copy(const at::Tensor & self, int64_t split_size, int64_t dim=0) {
+    return at::_ops::split_copy_Tensor::call(self, split_size, dim);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  ::std::vector<at::Tensor> split_copy(const at::Tensor & self, int64_t split_size, int64_t dim=0) {
+    return at::_ops::split_copy_Tensor::call(self, split_size, dim);
+  }
+}
+
+// aten::split_copy.Tensor(Tensor self, SymInt split_size, int dim=0) -> Tensor[]
+inline ::std::vector<at::Tensor> split_copy_symint(const at::Tensor & self, c10::SymInt split_size, int64_t dim=0) {
+    return at::_ops::split_copy_Tensor::call(self, split_size, dim);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  ::std::vector<at::Tensor> split_copy(const at::Tensor & self, c10::SymInt split_size, int64_t dim=0) {
+    return at::_ops::split_copy_Tensor::call(self, split_size, dim);
+  }
+}
+
+// aten::split_copy.Tensor_out(Tensor self, SymInt split_size, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void split_copy_out(at::TensorList out, const at::Tensor & self, int64_t split_size, int64_t dim=0) {
+    return at::_ops::split_copy_Tensor_out::call(self, split_size, dim, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  void split_copy_out(at::TensorList out, const at::Tensor & self, int64_t split_size, int64_t dim=0) {
+    return at::_ops::split_copy_Tensor_out::call(self, split_size, dim, out);
+  }
+}
+
+// aten::split_copy.Tensor_out(Tensor self, SymInt split_size, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void split_copy_outf(const at::Tensor & self, int64_t split_size, int64_t dim, at::TensorList out) {
+    return at::_ops::split_copy_Tensor_out::call(self, split_size, dim, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  void split_copy_outf(const at::Tensor & self, int64_t split_size, int64_t dim, at::TensorList out) {
+    return at::_ops::split_copy_Tensor_out::call(self, split_size, dim, out);
+  }
+}
+
+// aten::split_copy.Tensor_out(Tensor self, SymInt split_size, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void split_copy_symint_out(at::TensorList out, const at::Tensor & self, c10::SymInt split_size, int64_t dim=0) {
+    return at::_ops::split_copy_Tensor_out::call(self, split_size, dim, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  void split_copy_out(at::TensorList out, const at::Tensor & self, c10::SymInt split_size, int64_t dim=0) {
+    return at::_ops::split_copy_Tensor_out::call(self, split_size, dim, out);
+  }
+}
+
+// aten::split_copy.Tensor_out(Tensor self, SymInt split_size, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void split_copy_symint_outf(const at::Tensor & self, c10::SymInt split_size, int64_t dim, at::TensorList out) {
+    return at::_ops::split_copy_Tensor_out::call(self, split_size, dim, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  void split_copy_outf(const at::Tensor & self, c10::SymInt split_size, int64_t dim, at::TensorList out) {
+    return at::_ops::split_copy_Tensor_out::call(self, split_size, dim, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/split_with_sizes_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/split_with_sizes_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..1d5c50a78125b15b3388490d0fd4e5c546928f12
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/split_with_sizes_copy.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/split_with_sizes_copy_ops.h>
+
+namespace at {
+
+
+// aten::split_with_sizes_copy(Tensor self, SymInt[] split_sizes, int dim=0) -> Tensor[]
+inline ::std::vector<at::Tensor> split_with_sizes_copy(const at::Tensor & self, at::IntArrayRef split_sizes, int64_t dim=0) {
+    return at::_ops::split_with_sizes_copy::call(self, c10::fromIntArrayRefSlow(split_sizes), dim);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  ::std::vector<at::Tensor> split_with_sizes_copy(const at::Tensor & self, at::IntArrayRef split_sizes, int64_t dim=0) {
+    return at::_ops::split_with_sizes_copy::call(self, c10::fromIntArrayRefSlow(split_sizes), dim);
+  }
+}
+
+// aten::split_with_sizes_copy(Tensor self, SymInt[] split_sizes, int dim=0) -> Tensor[]
+inline ::std::vector<at::Tensor> split_with_sizes_copy_symint(const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim=0) {
+    return at::_ops::split_with_sizes_copy::call(self, split_sizes, dim);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  ::std::vector<at::Tensor> split_with_sizes_copy(const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim=0) {
+    return at::_ops::split_with_sizes_copy::call(self, split_sizes, dim);
+  }
+}
+
+// aten::split_with_sizes_copy.out(Tensor self, SymInt[] split_sizes, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void split_with_sizes_copy_out(at::TensorList out, const at::Tensor & self, at::IntArrayRef split_sizes, int64_t dim=0) {
+    return at::_ops::split_with_sizes_copy_out::call(self, c10::fromIntArrayRefSlow(split_sizes), dim, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  void split_with_sizes_copy_out(at::TensorList out, const at::Tensor & self, at::IntArrayRef split_sizes, int64_t dim=0) {
+    return at::_ops::split_with_sizes_copy_out::call(self, c10::fromIntArrayRefSlow(split_sizes), dim, out);
+  }
+}
+
+// aten::split_with_sizes_copy.out(Tensor self, SymInt[] split_sizes, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void split_with_sizes_copy_outf(const at::Tensor & self, at::IntArrayRef split_sizes, int64_t dim, at::TensorList out) {
+    return at::_ops::split_with_sizes_copy_out::call(self, c10::fromIntArrayRefSlow(split_sizes), dim, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  void split_with_sizes_copy_outf(const at::Tensor & self, at::IntArrayRef split_sizes, int64_t dim, at::TensorList out) {
+    return at::_ops::split_with_sizes_copy_out::call(self, c10::fromIntArrayRefSlow(split_sizes), dim, out);
+  }
+}
+
+// aten::split_with_sizes_copy.out(Tensor self, SymInt[] split_sizes, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void split_with_sizes_copy_symint_out(at::TensorList out, const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim=0) {
+    return at::_ops::split_with_sizes_copy_out::call(self, split_sizes, dim, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  void split_with_sizes_copy_out(at::TensorList out, const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim=0) {
+    return at::_ops::split_with_sizes_copy_out::call(self, split_sizes, dim, out);
+  }
+}
+
+// aten::split_with_sizes_copy.out(Tensor self, SymInt[] split_sizes, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void split_with_sizes_copy_symint_outf(const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim, at::TensorList out) {
+    return at::_ops::split_with_sizes_copy_out::call(self, split_sizes, dim, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  void split_with_sizes_copy_outf(const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim, at::TensorList out) {
+    return at::_ops::split_with_sizes_copy_out::call(self, split_sizes, dim, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/std_mean_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/std_mean_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..674693ac41fdb5605adb1de851a302e3b07f0a36
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/std_mean_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> std_mean_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & self, at::OptionalIntArrayRef dim=::std::nullopt, const ::std::optional<at::Scalar> & correction=::std::nullopt, bool keepdim=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> std_mean_outf(const at::Tensor & self, at::OptionalIntArrayRef dim, const ::std::optional<at::Scalar> & correction, bool keepdim, at::Tensor & out0, at::Tensor & out1);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/std_mean_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/std_mean_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..94862248084b3008aafcbc0b4c9d3c41affa7a0b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/std_mean_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> std_mean(const at::Tensor & self, at::OptionalIntArrayRef dim=::std::nullopt, const ::std::optional<at::Scalar> & correction=::std::nullopt, bool keepdim=false);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sub_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sub_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..295e9658d8ffe3b812227696874143f853593190
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sub_cuda_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor sub(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & sub_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & sub_outf(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha, at::Tensor & out);
+TORCH_API at::Tensor & sub_(at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sub_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sub_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f4cefe3391ab62e6b83f937fcccfb90e421ad9f5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sub_ops.h
@@ -0,0 +1,83 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API sub_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const at::Scalar &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sub";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "sub.out(Tensor self, Tensor other, *, Scalar alpha=1, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha, at::Tensor & out);
+};
+
+struct TORCH_API sub_Tensor {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sub";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "sub.Tensor(Tensor self, Tensor other, *, Scalar alpha=1) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha);
+};
+
+struct TORCH_API sub__Tensor {
+  using schema = at::Tensor & (at::Tensor &, const at::Tensor &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sub_";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "sub_.Tensor(Tensor(a!) self, Tensor other, *, Scalar alpha=1) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha);
+};
+
+struct TORCH_API sub_Scalar {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sub";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "sub.Scalar(Tensor self, Scalar other, Scalar alpha=1) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha);
+};
+
+struct TORCH_API sub__Scalar {
+  using schema = at::Tensor & (at::Tensor &, const at::Scalar &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sub_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "sub_.Scalar(Tensor(a!) self, Scalar other, Scalar alpha=1) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha);
+};
+
+struct TORCH_API sub_Scalar_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, const at::Scalar &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sub";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "sub.Scalar_out(Tensor self, Scalar other, Scalar alpha=1, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sym_size_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sym_size_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..7a9b32bac1ed713cd4149d1cbebe181d6a5f8cb4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sym_size_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API c10::SymInt sym_size(const at::Tensor & self, int64_t dim);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..1f504edb2a97eacd9bfbae61854092080181dc3e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_tanh : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..5513a15f68e33d327e4647fef91c5f67a3d83e7c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_ops.h
@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API tanh {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::tanh";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "tanh(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+struct TORCH_API tanh_ {
+  using schema = at::Tensor & (at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::tanh_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "tanh_(Tensor(a!) self) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self);
+};
+
+struct TORCH_API tanh_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::tanh";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "tanh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/thnn_conv2d_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/thnn_conv2d_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ff20733445d05424ae492bab0c23ee84bcb095f8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/thnn_conv2d_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor thnn_conv2d(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias={}, at::IntArrayRef stride=1, at::IntArrayRef padding=0);
+TORCH_API at::Tensor thnn_conv2d_symint(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef padding=c10::SymInt(0));
+TORCH_API at::Tensor & thnn_conv2d_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias={}, at::IntArrayRef stride=1, at::IntArrayRef padding=0);
+TORCH_API at::Tensor & thnn_conv2d_outf(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::Tensor & out);
+TORCH_API at::Tensor & thnn_conv2d_symint_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef padding=c10::SymInt(0));
+TORCH_API at::Tensor & thnn_conv2d_symint_outf(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_backward_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_backward_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0c7c028abef7cfa46f034f769207f833e4f48562
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_backward_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor threshold_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold);
+TORCH_API at::Tensor & threshold_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold);
+TORCH_API at::Tensor & threshold_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold, at::Tensor & grad_input);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/to_dense.h b/lib/python3.12/site-packages/torch/include/ATen/ops/to_dense.h
new file mode 100644
index 0000000000000000000000000000000000000000..679356784436e1cea2a80bedb0c6ebddb6ef9f38
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/to_dense.h
@@ -0,0 +1,26 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/to_dense_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/to_dense_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/to_dense_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..0e423e6ab65ec60ca9b63846812e1a5011c3be76
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/to_dense_backward.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/to_dense_backward_ops.h>
+
+namespace at {
+
+
+// aten::to_dense_backward(Tensor grad, Tensor input, bool? masked_grad=None) -> Tensor
+inline at::Tensor to_dense_backward(const at::Tensor & grad, const at::Tensor & input, ::std::optional<bool> masked_grad=::std::nullopt) {
+    return at::_ops::to_dense_backward::call(grad, input, masked_grad);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/to_mkldnn_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/to_mkldnn_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0fbe8446e56d16acb7f3e34dd2213e601a6a6f33
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/to_mkldnn_cpu_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor to_mkldnn(const at::Tensor & self, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_bsc_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_bsc_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..2b442b28cd44422cba57a2f28673153a660878e0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_bsc_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API to_sparse_bsc {
+  using schema = at::Tensor (const at::Tensor &, at::IntArrayRef, ::std::optional<int64_t>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::to_sparse_bsc";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "to_sparse_bsc(Tensor self, int[2] blocksize, int? dense_dim=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::IntArrayRef blocksize, ::std::optional<int64_t> dense_dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef blocksize, ::std::optional<int64_t> dense_dim);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_csr_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_csr_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4be401666a1d610ea7b20ff0f5e4108c1ac28448
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_csr_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor to_sparse_csr(const at::Tensor & self, ::std::optional<int64_t> dense_dim=::std::nullopt);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/topk_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/topk_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..f808898ae57d9f374dcd7820c0eddea3b2cb4f8d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/topk_native.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/topk_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_topk_out_cpu : public at::meta::structured_topk {
+void impl(const at::Tensor & self, int64_t k, int64_t dim, bool largest, bool sorted, const at::Tensor & values, const at::Tensor & indices);
+};
+struct TORCH_API structured_topk_out_cuda : public at::meta::structured_topk {
+void impl(const at::Tensor & self, int64_t k, int64_t dim, bool largest, bool sorted, const at::Tensor & values, const at::Tensor & indices);
+};
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> topk_quantized_cpu(const at::Tensor & self, int64_t k, int64_t dim=-1, bool largest=true, bool sorted=true);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/transpose_copy_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/transpose_copy_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..973e68ce6c1f7b0de1d60b7ee7aa7aba025e215f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/transpose_copy_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & transpose_copy_out(at::Tensor & out, const at::Tensor & self, int64_t dim0, int64_t dim1);
+TORCH_API at::Tensor & transpose_copy_outf(const at::Tensor & self, int64_t dim0, int64_t dim1, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/trapz_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/trapz_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..64ba658f8d79955836043c18c99fc927f9d24426
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/trapz_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API trapz_x {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::trapz";
+  static constexpr const char* overload_name = "x";
+  static constexpr const char* schema_str = "trapz.x(Tensor y, Tensor x, *, int dim=-1) -> Tensor";
+  static at::Tensor call(const at::Tensor & y, const at::Tensor & x, int64_t dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & y, const at::Tensor & x, int64_t dim);
+};
+
+struct TORCH_API trapz_dx {
+  using schema = at::Tensor (const at::Tensor &, double, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::trapz";
+  static constexpr const char* overload_name = "dx";
+  static constexpr const char* schema_str = "trapz.dx(Tensor y, *, float dx=1, int dim=-1) -> Tensor";
+  static at::Tensor call(const at::Tensor & y, double dx, int64_t dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & y, double dx, int64_t dim);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/tril_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/tril_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1b9e1fe036a0b8c9e09aa1050bf4c192cb9c7742
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/tril_cpu_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor tril(const at::Tensor & self, int64_t diagonal=0);
+TORCH_API at::Tensor & tril_out(at::Tensor & out, const at::Tensor & self, int64_t diagonal=0);
+TORCH_API at::Tensor & tril_outf(const at::Tensor & self, int64_t diagonal, at::Tensor & out);
+TORCH_API at::Tensor & tril_(at::Tensor & self, int64_t diagonal=0);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/tril_indices_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/tril_indices_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9a46e9bb9856e20879034a78364f466718d45b8d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/tril_indices_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & tril_indices_out(int64_t row, int64_t col, int64_t offset, at::Tensor & out);
+TORCH_API at::Tensor tril_indices_cpu(int64_t row, int64_t col, int64_t offset=0, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={});
+TORCH_API at::Tensor tril_indices_cuda(int64_t row, int64_t col, int64_t offset=0, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={});
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/triu_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/triu_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..630d3b0de6ba95926b77a7037d2509424afe2084
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/triu_native.h
@@ -0,0 +1,26 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/triu_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_triu_cpu : public at::meta::structured_triu {
+void impl(const at::Tensor & self, int64_t diagonal, const at::Tensor & out);
+};
+struct TORCH_API structured_triu_cuda : public at::meta::structured_triu {
+void impl(const at::Tensor & self, int64_t diagonal, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unfold_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unfold_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b6940f845cd2ba9d0300001907e23ab542bf56d3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unfold_cpu_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor unfold(const at::Tensor & self, int64_t dimension, int64_t size, int64_t step);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unique_dim_consecutive.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unique_dim_consecutive.h
new file mode 100644
index 0000000000000000000000000000000000000000..fda1d7c1aa3218e0db5a1ecb71845d52a8d8adbe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unique_dim_consecutive.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/unique_dim_consecutive_ops.h>
+
+namespace at {
+
+
+// aten::unique_dim_consecutive(Tensor self, int dim, bool return_inverse=False, bool return_counts=False) -> (Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> unique_dim_consecutive(const at::Tensor & self, int64_t dim, bool return_inverse=false, bool return_counts=false) {
+    return at::_ops::unique_dim_consecutive::call(self, dim, return_inverse, return_counts);
+}
+
+// aten::unique_dim_consecutive.out(Tensor self, int dim, bool return_inverse=False, bool return_counts=False, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> unique_dim_consecutive_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & self, int64_t dim, bool return_inverse=false, bool return_counts=false) {
+    return at::_ops::unique_dim_consecutive_out::call(self, dim, return_inverse, return_counts, out0, out1, out2);
+}
+// aten::unique_dim_consecutive.out(Tensor self, int dim, bool return_inverse=False, bool return_counts=False, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> unique_dim_consecutive_outf(const at::Tensor & self, int64_t dim, bool return_inverse, bool return_counts, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2) {
+    return at::_ops::unique_dim_consecutive_out::call(self, dim, return_inverse, return_counts, out0, out1, out2);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_split_with_sizes_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_split_with_sizes_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6f7c9db51ef65c0c349c512fa6c77e2c74d68ed1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_split_with_sizes_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::vector<at::Tensor> unsafe_split_with_sizes(const at::Tensor & self, at::IntArrayRef split_sizes, int64_t dim=0);
+TORCH_API ::std::vector<at::Tensor> unsafe_split_with_sizes_symint(const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim=0);
+TORCH_API void unsafe_split_with_sizes_out(at::TensorList out, const at::Tensor & self, at::IntArrayRef split_sizes, int64_t dim=0);
+TORCH_API void unsafe_split_with_sizes_outf(const at::Tensor & self, at::IntArrayRef split_sizes, int64_t dim, at::TensorList out);
+TORCH_API void unsafe_split_with_sizes_symint_out(at::TensorList out, const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim=0);
+TORCH_API void unsafe_split_with_sizes_symint_outf(const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim, at::TensorList out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bilinear2d_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bilinear2d_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4e153303d59c0276bec51570be7e9d74869d71d8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bilinear2d_backward_cpu_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor upsample_bilinear2d_backward(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor upsample_bilinear2d_backward_symint(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_bilinear2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_bilinear2d_backward_outf(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input);
+TORCH_API at::Tensor & upsample_bilinear2d_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_bilinear2d_backward_symint_outf(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_backward_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_backward_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cba02f4ef660188d87be3c9e1e0b851db31af365
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_backward_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor upsample_linear1d_backward(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor upsample_linear1d_backward_symint(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales=::std::nullopt);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d190c903ba8e22cf2984ce43d954c7b741bddf67
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor upsample_linear1d(const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor upsample_linear1d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales=::std::nullopt);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest1d_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest1d_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..47f7ce375d97755dcd05bf29504e5d8a9e4b4715
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest1d_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor upsample_nearest1d(const at::Tensor & input, at::OptionalIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors);
+TORCH_API at::Tensor upsample_nearest1d_symint(const at::Tensor & input, at::OptionalSymIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d.h
new file mode 100644
index 0000000000000000000000000000000000000000..6066df9922aa35c04410bdb99b74873da7caf73a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d.h
@@ -0,0 +1,113 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/upsample_nearest3d_ops.h>
+
+namespace at {
+
+
+// aten::upsample_nearest3d.vec(Tensor input, SymInt[]? output_size, float[]? scale_factors) -> Tensor
+inline at::Tensor upsample_nearest3d(const at::Tensor & input, at::OptionalIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest3d_vec::call(input, output_size.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*output_size)) : ::std::nullopt, scale_factors);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor upsample_nearest3d(const at::Tensor & input, at::OptionalIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest3d_vec::call(input, output_size.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*output_size)) : ::std::nullopt, scale_factors);
+  }
+}
+
+// aten::upsample_nearest3d.vec(Tensor input, SymInt[]? output_size, float[]? scale_factors) -> Tensor
+inline at::Tensor upsample_nearest3d_symint(const at::Tensor & input, at::OptionalSymIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest3d_vec::call(input, output_size, scale_factors);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor upsample_nearest3d(const at::Tensor & input, at::OptionalSymIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest3d_vec::call(input, output_size, scale_factors);
+  }
+}
+
+// aten::upsample_nearest3d.out(Tensor self, SymInt[3] output_size, float? scales_d=None, float? scales_h=None, float? scales_w=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest3d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::upsample_nearest3d_out::call(self, c10::fromIntArrayRefSlow(output_size), scales_d, scales_h, scales_w, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & upsample_nearest3d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::upsample_nearest3d_out::call(self, c10::fromIntArrayRefSlow(output_size), scales_d, scales_h, scales_w, out);
+  }
+}
+
+// aten::upsample_nearest3d.out(Tensor self, SymInt[3] output_size, float? scales_d=None, float? scales_h=None, float? scales_w=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest3d_outf(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out) {
+    return at::_ops::upsample_nearest3d_out::call(self, c10::fromIntArrayRefSlow(output_size), scales_d, scales_h, scales_w, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & upsample_nearest3d_outf(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out) {
+    return at::_ops::upsample_nearest3d_out::call(self, c10::fromIntArrayRefSlow(output_size), scales_d, scales_h, scales_w, out);
+  }
+}
+
+// aten::upsample_nearest3d.out(Tensor self, SymInt[3] output_size, float? scales_d=None, float? scales_h=None, float? scales_w=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest3d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::upsample_nearest3d_out::call(self, output_size, scales_d, scales_h, scales_w, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & upsample_nearest3d_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::upsample_nearest3d_out::call(self, output_size, scales_d, scales_h, scales_w, out);
+  }
+}
+
+// aten::upsample_nearest3d.out(Tensor self, SymInt[3] output_size, float? scales_d=None, float? scales_h=None, float? scales_w=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest3d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out) {
+    return at::_ops::upsample_nearest3d_out::call(self, output_size, scales_d, scales_h, scales_w, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & upsample_nearest3d_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out) {
+    return at::_ops::upsample_nearest3d_out::call(self, output_size, scales_d, scales_h, scales_w, out);
+  }
+}
+
+// aten::upsample_nearest3d(Tensor self, SymInt[3] output_size, float? scales_d=None, float? scales_h=None, float? scales_w=None) -> Tensor
+inline at::Tensor upsample_nearest3d(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::upsample_nearest3d::call(self, c10::fromIntArrayRefSlow(output_size), scales_d, scales_h, scales_w);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor upsample_nearest3d(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::upsample_nearest3d::call(self, c10::fromIntArrayRefSlow(output_size), scales_d, scales_h, scales_w);
+  }
+}
+
+// aten::upsample_nearest3d(Tensor self, SymInt[3] output_size, float? scales_d=None, float? scales_h=None, float? scales_w=None) -> Tensor
+inline at::Tensor upsample_nearest3d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::upsample_nearest3d::call(self, output_size, scales_d, scales_h, scales_w);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor upsample_nearest3d(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::upsample_nearest3d::call(self, output_size, scales_d, scales_h, scales_w);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..721a3e904dff21209b5a099df3e1740e0d30408a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_cuda_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor upsample_nearest3d(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor upsample_nearest3d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_nearest3d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_nearest3d_outf(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+TORCH_API at::Tensor & upsample_nearest3d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_nearest3d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/value_selecting_reduction_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/value_selecting_reduction_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a46d939951a1560a7114a6afa98b342de8019543
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/value_selecting_reduction_backward_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API value_selecting_reduction_backward {
+  using schema = at::Tensor (const at::Tensor &, int64_t, const at::Tensor &, c10::SymIntArrayRef, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::value_selecting_reduction_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "value_selecting_reduction_backward(Tensor grad, int dim, Tensor indices, SymInt[] sizes, bool keepdim) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad, int64_t dim, const at::Tensor & indices, c10::SymIntArrayRef sizes, bool keepdim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad, int64_t dim, const at::Tensor & indices, c10::SymIntArrayRef sizes, bool keepdim);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/vdot_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/vdot_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..cd6265f94a7dd8ac1cd79fd5d47c74d545a83ae4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/vdot_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API vdot {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::vdot";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "vdot(Tensor self, Tensor other) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & other);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other);
+};
+
+struct TORCH_API vdot_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::vdot";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "vdot.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..bd200e821a75af658957bb5614bb442b02302065
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_cuda_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor xlogy(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & xlogy_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & xlogy_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & xlogy_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/zeros_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/zeros_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1de9f9a876df0df3af0db2d1f101656dfc5a5984
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/zeros_ops.h
@@ -0,0 +1,61 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API zeros_names {
+  using schema = at::Tensor (at::IntArrayRef, ::std::optional<at::DimnameList>, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::zeros";
+  static constexpr const char* overload_name = "names";
+  static constexpr const char* schema_str = "zeros.names(int[] size, *, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(at::IntArrayRef size, ::std::optional<at::DimnameList> names, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, at::IntArrayRef size, ::std::optional<at::DimnameList> names, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API zeros {
+  using schema = at::Tensor (c10::SymIntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::zeros";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "zeros(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API zeros_out {
+  using schema = at::Tensor & (c10::SymIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::zeros";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "zeros.out(SymInt[] size, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymIntArrayRef size, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, at::Tensor & out);
+};
+
+struct TORCH_API zeros_names_out {
+  using schema = at::Tensor & (at::IntArrayRef, ::std::optional<at::DimnameList>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::zeros";
+  static constexpr const char* overload_name = "names_out";
+  static constexpr const char* schema_str = "zeros.names_out(int[] size, *, Dimname[]? names, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(at::IntArrayRef size, ::std::optional<at::DimnameList> names, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::IntArrayRef size, ::std::optional<at::DimnameList> names, at::Tensor & out);
+};
+
+}} // namespace at::_ops