diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool3d_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool3d_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d6acbd4972bdf06ae449c45678f44a7e16a4d6aa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool3d_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 & _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 compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_add_batch_dim_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_add_batch_dim_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..113a64fb012ff692787230590ad3fc4cd833d5db
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_add_batch_dim_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 _add_batch_dim {
+  using schema = at::Tensor (const at::Tensor &, int64_t, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_add_batch_dim";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_add_batch_dim(Tensor self, int batch_dim, int level) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t batch_dim, int64_t level);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t batch_dim, int64_t level);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_amp_update_scale_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_amp_update_scale_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..76f7c7d4476935e67660815bb9a22a6ced794863
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_amp_update_scale_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 _amp_update_scale_ {
+  using schema = at::Tensor & (at::Tensor &, at::Tensor &, const at::Tensor &, double, double, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_amp_update_scale_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_amp_update_scale_(Tensor(a!) self, Tensor(b!) growth_tracker, Tensor found_inf, float scale_growth_factor, float scale_backoff_factor, int growth_interval) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, at::Tensor & growth_tracker, const at::Tensor & found_inf, double scale_growth_factor, double scale_backoff_factor, int64_t growth_interval);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, at::Tensor & growth_tracker, const at::Tensor & found_inf, double scale_growth_factor, double scale_backoff_factor, int64_t growth_interval);
+};
+
+struct TORCH_API _amp_update_scale_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Tensor &, const at::Tensor &, double, double, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_amp_update_scale";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_amp_update_scale.out(Tensor self, Tensor(b!) growth_tracker, Tensor found_inf, float scale_growth_factor, float scale_backoff_factor, int growth_interval, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & growth_tracker, const at::Tensor & found_inf, double scale_growth_factor, double scale_backoff_factor, int64_t growth_interval, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & growth_tracker, const at::Tensor & found_inf, double scale_growth_factor, double scale_backoff_factor, int64_t growth_interval, at::Tensor & out);
+};
+
+struct TORCH_API _amp_update_scale {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, double, double, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_amp_update_scale";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_amp_update_scale(Tensor self, Tensor growth_tracker, Tensor found_inf, float scale_growth_factor, float scale_backoff_factor, int growth_interval) -> (Tensor, Tensor growth_tracker_out)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self, const at::Tensor & growth_tracker, const at::Tensor & found_inf, double scale_growth_factor, double scale_backoff_factor, int64_t growth_interval);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & growth_tracker, const at::Tensor & found_inf, double scale_growth_factor, double scale_backoff_factor, int64_t growth_interval);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_async_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_async_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0cdfd0b7dc8f080eecc8f0a5750676d812dac22c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_async_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 void _assert_async(const at::Tensor & self);
+TORCH_API void _assert_async(const at::Tensor & self, c10::string_view assert_msg);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_autocast_to_reduced_precision_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_autocast_to_reduced_precision_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..bb14e5b7deb4df6cc93d24772e3cb20579b2e808
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_autocast_to_reduced_precision_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 _autocast_to_reduced_precision(const at::Tensor & self, bool cuda_enabled, bool cpu_enabled, at::ScalarType cuda_dtype, at::ScalarType cpu_dtype);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index_backward_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index_backward_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d415ab531a7eb2e060cec1bdbb387ad0cbc523e1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index_backward_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,at::Tensor> _batch_norm_impl_index_backward(int64_t impl_index, const at::Tensor & input, const at::Tensor & grad_output, 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_var_transform, bool train, double eps, ::std::array<bool,3> output_mask, const at::Tensor & reservedSpace);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d6d0034c5e65e10679a919e274834c4caac5e193
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index_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,int64_t> _batch_norm_impl_index(const at::Tensor & input, const ::std::optional<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 momentum, double eps, bool cudnn_enabled);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Double_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Double_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c6552e25d1d461ca360d6a33879d7c2ece696e34
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Double_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_Double(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_Float_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Float_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a40f8edfc1397152c931fc33f076f8f658621373
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Float_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_Float(const at::Tensor & self, bool non_blocking=false);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Long.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Long.h
new file mode 100644
index 0000000000000000000000000000000000000000..ca6479cc3aef706160fdc7f63de87b911ac15f09
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Long.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/_cast_Long_ops.h>
+
+namespace at {
+
+
+// aten::_cast_Long(Tensor self, bool non_blocking=False) -> Tensor
+inline at::Tensor _cast_Long(const at::Tensor & self, bool non_blocking=false) {
+    return at::_ops::_cast_Long::call(self, non_blocking);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Short_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Short_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..96ca0518fd1a1121d805a374e59912ee84ad75f2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Short_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_Short(const at::Tensor & self, bool non_blocking=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_conj_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_conj_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3b0bacdad27ef87e0fe1075ba30d097d7ccc3209
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_conj_compositeexplicitautograd_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 compositeexplicitautograd {
+
+TORCH_API at::Tensor _conj(const at::Tensor & self);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_convolution_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_convolution_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..861977b688394d4114992a83b1e894eccd43ee90
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_convolution_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 _convolution {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, bool, c10::SymIntArrayRef, c10::SymInt, bool, bool, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_convolution";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_convolution(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, SymInt[] output_padding, SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32) -> Tensor";
+  static at::Tensor call(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, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, 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, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32);
+};
+
+struct TORCH_API _convolution_deprecated {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, bool, at::IntArrayRef, c10::SymInt, bool, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_convolution";
+  static constexpr const char* overload_name = "deprecated";
+  static constexpr const char* schema_str = "_convolution.deprecated(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, int[] output_padding, SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled) -> Tensor";
+  static at::Tensor call(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, at::IntArrayRef output_padding, c10::SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, 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, at::IntArrayRef output_padding, c10::SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled);
+};
+
+struct TORCH_API _convolution_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, bool, c10::SymIntArrayRef, c10::SymInt, bool, bool, bool, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_convolution";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_convolution.out(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, SymInt[] output_padding, SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(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, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, 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, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_copy_from_and_resize_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_copy_from_and_resize_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e64f7f41ae5c27aeacb9cf86c33ecd8b7c5aded2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_copy_from_and_resize_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 & _copy_from_and_resize_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & dst);
+TORCH_API at::Tensor & _copy_from_and_resize_outf(const at::Tensor & self, const at::Tensor & dst, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cslt_sparse_mm_search_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cslt_sparse_mm_search_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..e8f6646733a36cc714b40f239c011b3c98efcbca
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cslt_sparse_mm_search_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 int64_t _cslt_sparse_mm_search(const at::Tensor & compressed_A, const at::Tensor & dense_B, const ::std::optional<at::Tensor> & bias={}, const ::std::optional<at::Tensor> & alpha={}, ::std::optional<at::ScalarType> out_dtype=::std::nullopt, bool transpose_result=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_ctc_loss.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_ctc_loss.h
new file mode 100644
index 0000000000000000000000000000000000000000..31edc522ed6a2b428be5b6a9ad11c3d13afff50e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_ctc_loss.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/_ctc_loss_ops.h>
+
+namespace at {
+
+
+// aten::_ctc_loss(Tensor log_probs, Tensor targets, int[] input_lengths, int[] target_lengths, int blank=0, bool zero_infinity=False) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _ctc_loss(const at::Tensor & log_probs, const at::Tensor & targets, at::IntArrayRef input_lengths, at::IntArrayRef target_lengths, int64_t blank=0, bool zero_infinity=false) {
+    return at::_ops::_ctc_loss::call(log_probs, targets, input_lengths, target_lengths, blank, zero_infinity);
+}
+
+// aten::_ctc_loss.Tensor(Tensor log_probs, Tensor targets, Tensor input_lengths, Tensor target_lengths, int blank=0, bool zero_infinity=False) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _ctc_loss(const at::Tensor & log_probs, const at::Tensor & targets, const at::Tensor & input_lengths, const at::Tensor & target_lengths, int64_t blank=0, bool zero_infinity=false) {
+    return at::_ops::_ctc_loss_Tensor::call(log_probs, targets, input_lengths, target_lengths, blank, zero_infinity);
+}
+
+// aten::_ctc_loss.out(Tensor log_probs, Tensor targets, int[] input_lengths, int[] target_lengths, int blank=0, bool zero_infinity=False, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _ctc_loss_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & log_probs, const at::Tensor & targets, at::IntArrayRef input_lengths, at::IntArrayRef target_lengths, int64_t blank=0, bool zero_infinity=false) {
+    return at::_ops::_ctc_loss_out::call(log_probs, targets, input_lengths, target_lengths, blank, zero_infinity, out0, out1);
+}
+// aten::_ctc_loss.out(Tensor log_probs, Tensor targets, int[] input_lengths, int[] target_lengths, int blank=0, bool zero_infinity=False, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _ctc_loss_outf(const at::Tensor & log_probs, const at::Tensor & targets, at::IntArrayRef input_lengths, at::IntArrayRef target_lengths, int64_t blank, bool zero_infinity, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::_ctc_loss_out::call(log_probs, targets, input_lengths, target_lengths, blank, zero_infinity, out0, out1);
+}
+
+// aten::_ctc_loss.Tensor_out(Tensor log_probs, Tensor targets, Tensor input_lengths, Tensor target_lengths, int blank=0, bool zero_infinity=False, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _ctc_loss_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & log_probs, const at::Tensor & targets, const at::Tensor & input_lengths, const at::Tensor & target_lengths, int64_t blank=0, bool zero_infinity=false) {
+    return at::_ops::_ctc_loss_Tensor_out::call(log_probs, targets, input_lengths, target_lengths, blank, zero_infinity, out0, out1);
+}
+// aten::_ctc_loss.Tensor_out(Tensor log_probs, Tensor targets, Tensor input_lengths, Tensor target_lengths, int blank=0, bool zero_infinity=False, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _ctc_loss_outf(const at::Tensor & log_probs, const at::Tensor & targets, const at::Tensor & input_lengths, const at::Tensor & target_lengths, int64_t blank, bool zero_infinity, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::_ctc_loss_Tensor_out::call(log_probs, targets, input_lengths, target_lengths, blank, zero_infinity, out0, out1);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_ctc_loss_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_ctc_loss_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d1ec555a6ab922737630fc9c9ee1aaf43758183b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_ctc_loss_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> _ctc_loss(const at::Tensor & log_probs, const at::Tensor & targets, at::IntArrayRef input_lengths, at::IntArrayRef target_lengths, int64_t blank=0, bool zero_infinity=false);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> _ctc_loss(const at::Tensor & log_probs, const at::Tensor & targets, const at::Tensor & input_lengths, const at::Tensor & target_lengths, int64_t blank=0, bool zero_infinity=false);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_flatten_weight_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_flatten_weight_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b0743c146921bcb04569b4242b015127397e38dd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_flatten_weight_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 & _cudnn_rnn_flatten_weight_out_symint(at::TensorList weight_arr, int64_t weight_stride0, c10::SymInt input_size, int64_t mode, c10::SymInt hidden_size, c10::SymInt proj_size, int64_t num_layers, bool batch_first, bool bidirectional, at::Tensor & out);
+TORCH_API at::Tensor _cudnn_rnn_flatten_weight(at::TensorList weight_arr, int64_t weight_stride0, int64_t input_size, int64_t mode, int64_t hidden_size, int64_t proj_size, int64_t num_layers, bool batch_first, bool bidirectional);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..c0430034c198f3ef01767a8359c7bbdd814f5a0a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_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_backward_ops.h>
+
+namespace at {
+
+
+// aten::_embedding_bag_backward(Tensor grad, Tensor indices, Tensor offsets, Tensor offset2bag, Tensor bag_size, Tensor maximum_indices, SymInt num_weights, bool scale_grad_by_freq, int mode, bool sparse, Tensor? per_sample_weights, int padding_idx=-1) -> Tensor
+inline 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) {
+    return at::_ops::_embedding_bag_backward::call(grad, indices, offsets, offset2bag, bag_size, maximum_indices, num_weights, scale_grad_by_freq, mode, sparse, 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_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) {
+    return at::_ops::_embedding_bag_backward::call(grad, indices, offsets, offset2bag, bag_size, maximum_indices, num_weights, scale_grad_by_freq, mode, sparse, per_sample_weights, padding_idx);
+  }
+}
+
+// aten::_embedding_bag_backward(Tensor grad, Tensor indices, Tensor offsets, Tensor offset2bag, Tensor bag_size, Tensor maximum_indices, SymInt num_weights, bool scale_grad_by_freq, int mode, bool sparse, Tensor? per_sample_weights, int padding_idx=-1) -> Tensor
+inline 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) {
+    return at::_ops::_embedding_bag_backward::call(grad, indices, offsets, offset2bag, bag_size, maximum_indices, num_weights, scale_grad_by_freq, mode, sparse, 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_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, 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) {
+    return at::_ops::_embedding_bag_backward::call(grad, indices, offsets, offset2bag, bag_size, maximum_indices, num_weights, scale_grad_by_freq, mode, sparse, per_sample_weights, padding_idx);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_euclidean_dist_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_euclidean_dist_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9519415fd7274b25d999527527379dcdfe9ea14f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_euclidean_dist_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 _euclidean_dist(const at::Tensor & x1, const at::Tensor & x2);
+TORCH_API at::Tensor & _euclidean_dist_out(at::Tensor & out, const at::Tensor & x1, const at::Tensor & x2);
+TORCH_API at::Tensor & _euclidean_dist_outf(const at::Tensor & x1, const at::Tensor & x2, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_tensor_affine_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_tensor_affine_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..19968879bfe5429b8e3a1bdbc2450519b0d110e4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_tensor_affine_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> _fake_quantize_learnable_per_tensor_affine_backward(const at::Tensor & grad, const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, 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/_fft_r2c_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fft_r2c_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6038230d73ce50265f1b16c36576095804735542
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fft_r2c_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 _fft_r2c(const at::Tensor & self, at::IntArrayRef dim, int64_t normalization, bool onesided);
+TORCH_API at::Tensor & _fft_r2c_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim, int64_t normalization, bool onesided);
+TORCH_API at::Tensor & _fft_r2c_outf(const at::Tensor & self, at::IntArrayRef dim, int64_t normalization, bool onesided, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_asin_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_asin_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..37c80dcd9fe43da8bba0c6f390186c2f6153cbf6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_asin_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_asin {
+  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_asin";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_asin(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_asin_ {
+  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_asin_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_asin_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_asin_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_asin";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_asin.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/_foreach_atan_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_atan_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..dc46ea4be51ae5ef3b843d45cd828f495f0753c4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_atan_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_atan_slow(at::TensorList self);
+TORCH_API void _foreach_atan_out(at::TensorList self, at::TensorList out);
+TORCH_API void foreach_tensor_atan_slow_(at::TensorList self);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_atan_cuda(at::TensorList self);
+TORCH_API void foreach_tensor_atan_cuda_(at::TensorList self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_copy_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_copy_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cab6e7c4e5f839bb1e1fa61235c3716c85e17adc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_copy_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_copy(at::TensorList self, at::TensorList src, bool non_blocking=false);
+TORCH_API void _foreach_copy_out(at::TensorList out, at::TensorList self, at::TensorList src, bool non_blocking=false);
+TORCH_API void _foreach_copy_outf(at::TensorList self, at::TensorList src, bool non_blocking, at::TensorList out);
+TORCH_API void _foreach_copy_(at::TensorList self, at::TensorList src, bool non_blocking=false);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_div_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_div_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..3cb8cc3928453faaab1a287e612c02b1e733fcad
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_div_ops.h
@@ -0,0 +1,149 @@
+#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_div_Scalar {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_div";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "_foreach_div.Scalar(Tensor[] self, Scalar scalar) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(at::TensorList self, const at::Scalar & scalar);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Scalar & scalar);
+};
+
+struct TORCH_API _foreach_div__Scalar {
+  using schema = void (at::TensorList, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_div_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "_foreach_div_.Scalar(Tensor(a!)[] self, Scalar scalar) -> ()";
+  static void call(at::TensorList self, const at::Scalar & scalar);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Scalar & scalar);
+};
+
+struct TORCH_API _foreach_div_List {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_div";
+  static constexpr const char* overload_name = "List";
+  static constexpr const char* schema_str = "_foreach_div.List(Tensor[] self, Tensor[] other) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(at::TensorList self, at::TensorList other);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::TensorList other);
+};
+
+struct TORCH_API _foreach_div__List {
+  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_div_";
+  static constexpr const char* overload_name = "List";
+  static constexpr const char* schema_str = "_foreach_div_.List(Tensor(a!)[] self, Tensor[] other) -> ()";
+  static void call(at::TensorList self, at::TensorList other);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::TensorList other);
+};
+
+struct TORCH_API _foreach_div_ScalarList {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, at::ArrayRef<at::Scalar>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_div";
+  static constexpr const char* overload_name = "ScalarList";
+  static constexpr const char* schema_str = "_foreach_div.ScalarList(Tensor[] self, Scalar[] scalars) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+};
+
+struct TORCH_API _foreach_div__ScalarList {
+  using schema = void (at::TensorList, at::ArrayRef<at::Scalar>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_div_";
+  static constexpr const char* overload_name = "ScalarList";
+  static constexpr const char* schema_str = "_foreach_div_.ScalarList(Tensor(a!)[] self, Scalar[] scalars) -> ()";
+  static void call(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+};
+
+struct TORCH_API _foreach_div_Tensor {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_div";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "_foreach_div.Tensor(Tensor[] self, Tensor other) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(at::TensorList self, const at::Tensor & other);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Tensor & other);
+};
+
+struct TORCH_API _foreach_div__Tensor {
+  using schema = void (at::TensorList, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_div_";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "_foreach_div_.Tensor(Tensor(a!)[] self, Tensor other) -> ()";
+  static void call(at::TensorList self, const at::Tensor & other);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Tensor & other);
+};
+
+struct TORCH_API _foreach_div_Scalar_out {
+  using schema = void (at::TensorList, const at::Scalar &, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_div";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "_foreach_div.Scalar_out(Tensor[] self, Scalar scalar, *, Tensor(a!)[] out) -> ()";
+  static void call(at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+};
+
+struct TORCH_API _foreach_div_List_out {
+  using schema = void (at::TensorList, at::TensorList, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_div";
+  static constexpr const char* overload_name = "List_out";
+  static constexpr const char* schema_str = "_foreach_div.List_out(Tensor[] self, Tensor[] other, *, Tensor(a!)[] out) -> ()";
+  static void call(at::TensorList self, at::TensorList other, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::TensorList other, at::TensorList out);
+};
+
+struct TORCH_API _foreach_div_ScalarList_out {
+  using schema = void (at::TensorList, at::ArrayRef<at::Scalar>, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_div";
+  static constexpr const char* overload_name = "ScalarList_out";
+  static constexpr const char* schema_str = "_foreach_div.ScalarList_out(Tensor[] self, Scalar[] scalars, *, Tensor(a!)[] out) -> ()";
+  static void call(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+};
+
+struct TORCH_API _foreach_div_Tensor_out {
+  using schema = void (at::TensorList, const at::Tensor &, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_div";
+  static constexpr const char* overload_name = "Tensor_out";
+  static constexpr const char* schema_str = "_foreach_div.Tensor_out(Tensor[] self, Tensor other, *, Tensor(a!)[] out) -> ()";
+  static void call(at::TensorList self, const at::Tensor & other, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Tensor & other, at::TensorList out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log.h
new file mode 100644
index 0000000000000000000000000000000000000000..9365a4b10244c175ddb63d6b6d882009013b8acb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log.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/_foreach_log_ops.h>
+
+namespace at {
+
+
+// aten::_foreach_log(Tensor[] self) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_log(at::TensorList self) {
+    return at::_ops::_foreach_log::call(self);
+}
+
+// aten::_foreach_log_(Tensor(a!)[] self) -> ()
+inline void _foreach_log_(at::TensorList self) {
+    return at::_ops::_foreach_log_::call(self);
+}
+
+// aten::_foreach_log.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_log_out(at::TensorList out, at::TensorList self) {
+    return at::_ops::_foreach_log_out::call(self, out);
+}
+// aten::_foreach_log.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_log_outf(at::TensorList self, at::TensorList out) {
+    return at::_ops::_foreach_log_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log1p.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log1p.h
new file mode 100644
index 0000000000000000000000000000000000000000..dc912733db67903fdf1924af79bfbdc70afe4667
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log1p.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/_foreach_log1p_ops.h>
+
+namespace at {
+
+
+// aten::_foreach_log1p(Tensor[] self) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_log1p(at::TensorList self) {
+    return at::_ops::_foreach_log1p::call(self);
+}
+
+// aten::_foreach_log1p_(Tensor(a!)[] self) -> ()
+inline void _foreach_log1p_(at::TensorList self) {
+    return at::_ops::_foreach_log1p_::call(self);
+}
+
+// aten::_foreach_log1p.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_log1p_out(at::TensorList out, at::TensorList self) {
+    return at::_ops::_foreach_log1p_out::call(self, out);
+}
+// aten::_foreach_log1p.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_log1p_outf(at::TensorList self, at::TensorList out) {
+    return at::_ops::_foreach_log1p_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log1p_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log1p_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ce50d61345e05742017d1a337c68bf56dcaffd22
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log1p_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_log1p(at::TensorList self);
+TORCH_API void _foreach_log1p_(at::TensorList self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..f706208653debd601014c15eb88ecea9b2ccac54
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log_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_log_slow(at::TensorList self);
+TORCH_API void _foreach_log_out(at::TensorList self, at::TensorList out);
+TORCH_API void foreach_tensor_log_slow_(at::TensorList self);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_log_cuda(at::TensorList self);
+TORCH_API void foreach_tensor_log_cuda_(at::TensorList self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_max_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_max_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b81aed856bca4b094cf09afd1307faf7d3e3bb0f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_max_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::vector<at::Tensor> _foreach_max(at::TensorList self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_max_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_max_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c58bb0a96d283bdd2256c4c250eb55b956d06e60
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_max_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 ::std::vector<at::Tensor> foreach_tensor_max_slow(at::TensorList self);
+TORCH_API void _foreach_max_out(at::TensorList self, at::TensorList out);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_max_cuda(at::TensorList self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_neg_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_neg_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..44c384ee34f94e4aa9a1e50bf2b9e565369f0d8d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_neg_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_neg(at::TensorList self);
+TORCH_API void _foreach_neg_out(at::TensorList out, at::TensorList self);
+TORCH_API void _foreach_neg_outf(at::TensorList self, at::TensorList out);
+TORCH_API void _foreach_neg_(at::TensorList self);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_norm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_norm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..04656cca9e7c24a50f855e602e74c738ae426bb5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_norm_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 ::std::vector<at::Tensor> foreach_tensor_norm_slow(at::TensorList self, const at::Scalar & ord=2, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API void _foreach_norm_Scalar_out(at::TensorList self, const at::Scalar & ord, ::std::optional<at::ScalarType> dtype, at::TensorList out);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_norm_cuda(at::TensorList self, const at::Scalar & ord=2, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_pow_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_pow_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8700d50d6fc515c8ef89f1c4052ed81f661ccfd0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_pow_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,35 @@
+#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_pow(at::TensorList self, at::TensorList exponent);
+TORCH_API void _foreach_pow_out(at::TensorList out, at::TensorList self, at::TensorList exponent);
+TORCH_API void _foreach_pow_outf(at::TensorList self, at::TensorList exponent, at::TensorList out);
+TORCH_API void _foreach_pow_(at::TensorList self, at::TensorList exponent);
+TORCH_API ::std::vector<at::Tensor> _foreach_pow(at::TensorList self, const at::Scalar & exponent);
+TORCH_API void _foreach_pow_out(at::TensorList out, at::TensorList self, const at::Scalar & exponent);
+TORCH_API void _foreach_pow_outf(at::TensorList self, const at::Scalar & exponent, at::TensorList out);
+TORCH_API void _foreach_pow_(at::TensorList self, const at::Scalar & exponent);
+TORCH_API ::std::vector<at::Tensor> _foreach_pow(at::TensorList self, at::ArrayRef<at::Scalar> exponent);
+TORCH_API void _foreach_pow_out(at::TensorList out, at::TensorList self, at::ArrayRef<at::Scalar> exponent);
+TORCH_API void _foreach_pow_outf(at::TensorList self, at::ArrayRef<at::Scalar> exponent, at::TensorList out);
+TORCH_API void _foreach_pow_(at::TensorList self, at::ArrayRef<at::Scalar> exponent);
+TORCH_API ::std::vector<at::Tensor> _foreach_pow(const at::Scalar & self, at::TensorList exponent);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_pow_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_pow_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2d2b73f82dc7a1ac13c58e9ed6a617810fd0ed1f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_pow_cuda_dispatch.h
@@ -0,0 +1,29 @@
+#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_pow(at::TensorList self, at::TensorList exponent);
+TORCH_API void _foreach_pow_(at::TensorList self, at::TensorList exponent);
+TORCH_API ::std::vector<at::Tensor> _foreach_pow(at::TensorList self, const at::Scalar & exponent);
+TORCH_API void _foreach_pow_(at::TensorList self, const at::Scalar & exponent);
+TORCH_API ::std::vector<at::Tensor> _foreach_pow(at::TensorList self, at::ArrayRef<at::Scalar> exponent);
+TORCH_API void _foreach_pow_(at::TensorList self, at::ArrayRef<at::Scalar> exponent);
+TORCH_API ::std::vector<at::Tensor> _foreach_pow(const at::Scalar & self, at::TensorList exponent);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sin.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sin.h
new file mode 100644
index 0000000000000000000000000000000000000000..6de1076a058eb2c58fddfee6eaaddecbfe4d8dab
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sin.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/_foreach_sin_ops.h>
+
+namespace at {
+
+
+// aten::_foreach_sin(Tensor[] self) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_sin(at::TensorList self) {
+    return at::_ops::_foreach_sin::call(self);
+}
+
+// aten::_foreach_sin_(Tensor(a!)[] self) -> ()
+inline void _foreach_sin_(at::TensorList self) {
+    return at::_ops::_foreach_sin_::call(self);
+}
+
+// aten::_foreach_sin.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_sin_out(at::TensorList out, at::TensorList self) {
+    return at::_ops::_foreach_sin_out::call(self, out);
+}
+// aten::_foreach_sin.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_sin_outf(at::TensorList self, at::TensorList out) {
+    return at::_ops::_foreach_sin_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sinh_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sinh_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..140a0e09d28348be8bb2de1e4d8d37d6c97abe1b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sinh_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_sinh {
+  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_sinh";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_sinh(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_sinh_ {
+  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_sinh_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_sinh_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_sinh_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_sinh";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_sinh.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/_foreach_sqrt_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sqrt_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..6e9a82f5197cead81d4e46d41ea111c11e299e57
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sqrt_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_sqrt_slow(at::TensorList self);
+TORCH_API void _foreach_sqrt_out(at::TensorList self, at::TensorList out);
+TORCH_API void foreach_tensor_sqrt_slow_(at::TensorList self);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_sqrt_cuda(at::TensorList self);
+TORCH_API void foreach_tensor_sqrt_cuda_(at::TensorList self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_zero_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_zero_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..15c894ec32d8e09850ea08156ed1f69a5501dff8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_zero_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_zero_ {
+  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_zero_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_zero_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_zero_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_zero";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_zero.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);
+};
+
+struct TORCH_API _foreach_zero {
+  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_zero";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_zero(Tensor[] self) -> Tensor[] self_out";
+  static ::std::vector<at::Tensor> call(at::TensorList self);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_assert_scalar_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_assert_scalar_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..52c067c1c32751aebd7a0c895c6ea5feb148b849
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_assert_scalar_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 _functional_assert_scalar {
+  using schema = at::Tensor (const at::Scalar &, c10::string_view, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_functional_assert_scalar";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_functional_assert_scalar(Scalar self, str assert_msg, Tensor dep_token) -> Tensor";
+  static at::Tensor call(const at::Scalar & self, c10::string_view assert_msg, const at::Tensor & dep_token);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & self, c10::string_view assert_msg, const at::Tensor & dep_token);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_dropout_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_dropout_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2993f6a6283355a4dce8620963f08bcfd249093d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_dropout_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> _fused_dropout(const at::Tensor & self, double p, ::std::optional<at::Generator> generator=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f1452d8c6c8dc79bb1fe9b98aebae782098038f8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_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> _fused_moving_avg_obs_fq_helper(const at::Tensor & self, const at::Tensor & observer_on, const at::Tensor & fake_quant_on, at::Tensor & running_min, at::Tensor & running_max, at::Tensor & scale, at::Tensor & zero_point, double averaging_const, int64_t quant_min, int64_t quant_max, int64_t ch_axis, bool per_row_fake_quant=false, bool symmetric_quant=false);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_sdp_choice_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_sdp_choice_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..7bd28ba073bc89b13f43e9a354dfdb407313f9b3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_sdp_choice_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 int64_t _fused_sdp_choice_cpp(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_mask={}, double dropout_p=0.0, bool is_causal=false, ::std::optional<double> scale=::std::nullopt, bool enable_gqa=false);
+TORCH_API int64_t _fused_sdp_choice_cuda(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_mask={}, double dropout_p=0.0, bool is_causal=false, ::std::optional<double> scale=::std::nullopt, bool enable_gqa=false);
+TORCH_API int64_t _fused_sdp_choice_meta(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_mask={}, double dropout_p=0.0, bool is_causal=false, ::std::optional<double> scale=::std::nullopt, bool enable_gqa=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fw_primal_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fw_primal_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..aa91848397008a42ce309455a1625c7c52110e89
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fw_primal_compositeexplicitautograd_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 compositeexplicitautograd {
+
+TORCH_API at::Tensor _fw_primal(const at::Tensor & self, int64_t level);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_cts_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_cts_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..27c359fdf4fea923de98d7bb3d93973dfacf5cf3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_cts_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 _histogramdd_from_bin_cts {
+  using schema = at::Tensor (const at::Tensor &, at::IntArrayRef, ::std::optional<at::ArrayRef<double>>, const ::std::optional<at::Tensor> &, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_histogramdd_from_bin_cts";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_histogramdd_from_bin_cts(Tensor self, int[] bins, *, float[]? range=None, Tensor? weight=None, bool density=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range, const ::std::optional<at::Tensor> & weight, bool density);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range, const ::std::optional<at::Tensor> & weight, bool density);
+};
+
+struct TORCH_API _histogramdd_from_bin_cts_out {
+  using schema = at::Tensor & (const at::Tensor &, at::IntArrayRef, ::std::optional<at::ArrayRef<double>>, const ::std::optional<at::Tensor> &, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_histogramdd_from_bin_cts";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_histogramdd_from_bin_cts.out(Tensor self, int[] bins, *, float[]? range=None, Tensor? weight=None, bool density=False, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range, const ::std::optional<at::Tensor> & weight, bool density, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, 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 at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_tensors_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_tensors_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..da9f60d9e7ae5a86bc209126b1cee5af216b95a8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_tensors_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 _histogramdd_from_bin_tensors(const at::Tensor & self, at::TensorList bins, const ::std::optional<at::Tensor> & weight={}, bool density=false);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_int_mm_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_int_mm_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2a312a781ba387eaba12f616e903957674981fd3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_int_mm_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 _int_mm(const at::Tensor & self, const at::Tensor & mat2);
+TORCH_API at::Tensor & _int_mm_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & mat2);
+TORCH_API at::Tensor & _int_mm_outf(const at::Tensor & self, const at::Tensor & mat2, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_check_errors_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_check_errors_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9cddc0da5b905c46c3b25cb943e70d82df446bfc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_check_errors_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 _linalg_check_errors {
+  using schema = void (const at::Tensor &, c10::string_view, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_linalg_check_errors";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_linalg_check_errors(Tensor info, str api_name, *, bool is_matrix) -> ()";
+  static void call(const at::Tensor & info, c10::string_view api_name, bool is_matrix);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & info, c10::string_view api_name, bool is_matrix);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_backward_data_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_backward_data_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..26d3139bc6afb0cfba06d5123e409cd293a83a3b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_backward_data_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__log_softmax_backward_data : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, at::ScalarType input_dtype);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_logcumsumexp.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_logcumsumexp.h
new file mode 100644
index 0000000000000000000000000000000000000000..fc686560eb59c41003c558f697c73589ab1c49a7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_logcumsumexp.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/_logcumsumexp_ops.h>
+
+namespace at {
+
+
+// aten::_logcumsumexp(Tensor self, int dim) -> Tensor
+inline at::Tensor _logcumsumexp(const at::Tensor & self, int64_t dim) {
+    return at::_ops::_logcumsumexp::call(self, dim);
+}
+
+// aten::_logcumsumexp.out(Tensor self, int dim, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _logcumsumexp_out(at::Tensor & out, const at::Tensor & self, int64_t dim) {
+    return at::_ops::_logcumsumexp_out::call(self, dim, out);
+}
+// aten::_logcumsumexp.out(Tensor self, int dim, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _logcumsumexp_outf(const at::Tensor & self, int64_t dim, at::Tensor & out) {
+    return at::_ops::_logcumsumexp_out::call(self, dim, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_masked_softmax_backward_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_masked_softmax_backward_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..21994ce860095118b3bff267dbf33b56e35eb28f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_masked_softmax_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 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);
+TORCH_API 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);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_mkldnn_transpose_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_mkldnn_transpose_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b0c8ac78e6442e19ae12e655bf04c900a5ad43d5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_mkldnn_transpose_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 & _mkldnn_transpose_(at::Tensor & self, int64_t dim0, int64_t dim1);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ec6e5481eae2f8a2ef8cb37dbec58f22669767d8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_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 ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _native_batch_norm_legit_out(at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, bool training, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _native_batch_norm_legit_outf(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, bool training, double momentum, double eps, at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _native_batch_norm_legit(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, bool training, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _native_batch_norm_legit(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, bool training, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _native_batch_norm_legit_out(at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, bool training, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _native_batch_norm_legit_outf(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, bool training, double momentum, double eps, at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_no_training_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_no_training_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..c9429ab6a1764b423d5e4e09dcd7c460fb48ba05
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_no_training_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_legit_no_training {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const at::Tensor &, const at::Tensor &, double, double);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_native_batch_norm_legit_no_training";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_native_batch_norm_legit_no_training(Tensor input, Tensor? weight, Tensor? bias, Tensor running_mean, Tensor running_var, float momentum, float eps) -> (Tensor, Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, double momentum, double eps);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, double momentum, double eps);
+};
+
+struct TORCH_API _native_batch_norm_legit_no_training_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> (const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const at::Tensor &, const at::Tensor &, 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::_native_batch_norm_legit_no_training";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_native_batch_norm_legit_no_training.out(Tensor input, Tensor? weight, Tensor? bias, Tensor running_mean, Tensor running_var, float momentum, float eps, *, 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 ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, double momentum, double eps, 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 ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, double momentum, double eps, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_ragged_idx_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_ragged_idx_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..e9d1361bfc39384baa19fb73a4598f777de03c8f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_ragged_idx_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_ragged_idx {
+  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::_nested_get_ragged_idx";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_nested_get_ragged_idx(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/_nested_tensor_from_mask_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_from_mask_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..886edb0166a9b6b6c79ed27da1069be53ff416ad
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_from_mask_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 & _nested_tensor_from_mask_out(const at::Tensor & t, const at::Tensor & mask, bool mask_check, at::Tensor & out);
+TORCH_API at::Tensor NestedTensor_nested_tensor_from_mask(const at::Tensor & t, const at::Tensor & mask, bool mask_check=true);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_buffer_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_buffer_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..8d0efd095f30d0666f0e6ce01cf857ac0c2388b8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_buffer_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_view_from_buffer {
+  using schema = at::Tensor (const at::Tensor &, const 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::_nested_view_from_buffer";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_nested_view_from_buffer(Tensor(a) self, Tensor nested_size, Tensor nested_strides, Tensor offsets) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & nested_size, const at::Tensor & nested_strides, const at::Tensor & offsets);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & nested_size, const at::Tensor & nested_strides, const at::Tensor & offsets);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_jagged_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_jagged_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b9d2e5fd5be542077158871f57ae9c937e4f7895
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_jagged_native.h
@@ -0,0 +1,20 @@
+#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 {
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_new_zeros_with_same_feature_meta_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_new_zeros_with_same_feature_meta_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b54d055631ae64d33e7c17b0fc9efa043c7ec26c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_new_zeros_with_same_feature_meta_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 _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(const at::Tensor & self, const at::Tensor & other, int64_t self_num_batch_dims, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_pad_packed_sequence.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_pad_packed_sequence.h
new file mode 100644
index 0000000000000000000000000000000000000000..b1f3fa880d9ec2e54c8917608b9e34c540106c99
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_pad_packed_sequence.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/_pad_packed_sequence_ops.h>
+
+namespace at {
+
+
+// aten::_pad_packed_sequence(Tensor data, Tensor batch_sizes, bool batch_first, Scalar padding_value, int total_length) -> (Tensor, Tensor)
+inline ::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) {
+    return at::_ops::_pad_packed_sequence::call(data, batch_sizes, batch_first, padding_value, total_length);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_prelu_kernel.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_prelu_kernel.h
new file mode 100644
index 0000000000000000000000000000000000000000..d774253dba49b9bb18d75999fc0fb70c6e6d673f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_prelu_kernel.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/_prelu_kernel_ops.h>
+
+namespace at {
+
+
+// aten::_prelu_kernel(Tensor self, Tensor weight) -> Tensor
+inline at::Tensor _prelu_kernel(const at::Tensor & self, const at::Tensor & weight) {
+    return at::_ops::_prelu_kernel::call(self, weight);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_propagate_xla_data.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_propagate_xla_data.h
new file mode 100644
index 0000000000000000000000000000000000000000..99d26a7aceda106231d6389d143ae9817c95490f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_propagate_xla_data.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/_propagate_xla_data_ops.h>
+
+namespace at {
+
+
+// aten::_propagate_xla_data(Tensor input, Tensor output) -> ()
+inline void _propagate_xla_data(const at::Tensor & input, const at::Tensor & output) {
+    return at::_ops::_propagate_xla_data::call(input, output);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_reshape_alias_copy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_reshape_alias_copy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2ccc18128777ccbe0344035298607910a9fef55e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_reshape_alias_copy_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 _reshape_alias_copy(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride);
+TORCH_API at::Tensor _reshape_alias_copy_symint(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_safe_softmax.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_safe_softmax.h
new file mode 100644
index 0000000000000000000000000000000000000000..35c4df01c99270388bb18d4534c419a3d433f6e7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_safe_softmax.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/_safe_softmax_ops.h>
+
+namespace at {
+
+
+// aten::_safe_softmax(Tensor self, int dim, ScalarType? dtype=None) -> Tensor
+inline at::Tensor _safe_softmax(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype=::std::nullopt) {
+    return at::_ops::_safe_softmax::call(self, dim, dtype);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_for_mps_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_for_mps_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b9d2e5fd5be542077158871f57ae9c937e4f7895
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_for_mps_native.h
@@ -0,0 +1,20 @@
+#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 {
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_cudnn_attention_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_cudnn_attention_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..375349ff4f05591f578d964c370cb10605b1dd08
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_cudnn_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_cudnn_attention_backward_ops.h>
+
+namespace at {
+
+
+// aten::_scaled_dot_product_cudnn_attention_backward(Tensor grad_out, Tensor query, Tensor key, Tensor value, Tensor out, Tensor logsumexp, Tensor philox_seed, Tensor philox_offset, Tensor attn_bias, Tensor cum_seq_q, Tensor cum_seq_k, SymInt max_q, SymInt max_k, float dropout_p, bool is_causal, *, float? scale=None) -> (Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _scaled_dot_product_cudnn_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 & philox_seed, const at::Tensor & philox_offset, const at::Tensor & attn_bias, 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, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_cudnn_attention_backward::call(grad_out, query, key, value, out, logsumexp, philox_seed, philox_offset, attn_bias, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, 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_cudnn_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 & philox_seed, const at::Tensor & philox_offset, const at::Tensor & attn_bias, 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, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_cudnn_attention_backward::call(grad_out, query, key, value, out, logsumexp, philox_seed, philox_offset, attn_bias, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, scale);
+  }
+}
+
+// aten::_scaled_dot_product_cudnn_attention_backward(Tensor grad_out, Tensor query, Tensor key, Tensor value, Tensor out, Tensor logsumexp, Tensor philox_seed, Tensor philox_offset, Tensor attn_bias, Tensor cum_seq_q, Tensor cum_seq_k, SymInt max_q, SymInt max_k, float dropout_p, bool is_causal, *, float? scale=None) -> (Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _scaled_dot_product_cudnn_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 & philox_seed, const at::Tensor & philox_offset, const at::Tensor & attn_bias, 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, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_cudnn_attention_backward::call(grad_out, query, key, value, out, logsumexp, philox_seed, philox_offset, attn_bias, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, 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_cudnn_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 & philox_seed, const at::Tensor & philox_offset, const at::Tensor & attn_bias, 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, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_cudnn_attention_backward::call(grad_out, query, key, value, out, logsumexp, philox_seed, philox_offset, attn_bias, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, scale);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_efficient_attention_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_efficient_attention_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..b78d84e0ba46eab0ea7ad14a0416d98d079f6cbd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_efficient_attention_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/_scaled_dot_product_efficient_attention_backward_ops.h>
+
+namespace at {
+
+
+// aten::_scaled_dot_product_efficient_attention_backward(Tensor grad_out_, Tensor query, Tensor key, Tensor value, Tensor attn_bias, Tensor out, Tensor logsumexp, Tensor philox_seed, Tensor philox_offset, float dropout_p, bool[4] grad_input_mask, bool is_causal=False, *, float? scale=None) -> (Tensor, Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> _scaled_dot_product_efficient_attention_backward(const at::Tensor & grad_out_, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const at::Tensor & attn_bias, const at::Tensor & out, const at::Tensor & logsumexp, const at::Tensor & philox_seed, const at::Tensor & philox_offset, double dropout_p, ::std::array<bool,4> grad_input_mask, bool is_causal=false, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_efficient_attention_backward::call(grad_out_, query, key, value, attn_bias, out, logsumexp, philox_seed, philox_offset, dropout_p, grad_input_mask, is_causal, scale);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_mm.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_mm.h
new file mode 100644
index 0000000000000000000000000000000000000000..ba192b617eb6a7fae58c684f5286659147ff87ec
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_mm.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/_scaled_mm_ops.h>
+
+namespace at {
+
+
+// aten::_scaled_mm(Tensor self, Tensor mat2, Tensor scale_a, Tensor scale_b, Tensor? bias=None, Tensor? scale_result=None, ScalarType? out_dtype=None, bool use_fast_accum=False) -> Tensor
+inline at::Tensor _scaled_mm(const at::Tensor & self, const at::Tensor & mat2, const at::Tensor & scale_a, const at::Tensor & scale_b, const ::std::optional<at::Tensor> & bias={}, const ::std::optional<at::Tensor> & scale_result={}, ::std::optional<at::ScalarType> out_dtype=::std::nullopt, bool use_fast_accum=false) {
+    return at::_ops::_scaled_mm::call(self, mat2, scale_a, scale_b, bias, scale_result, out_dtype, use_fast_accum);
+}
+
+// aten::_scaled_mm.out(Tensor self, Tensor mat2, Tensor scale_a, Tensor scale_b, Tensor? bias=None, Tensor? scale_result=None, ScalarType? out_dtype=None, bool use_fast_accum=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _scaled_mm_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & mat2, const at::Tensor & scale_a, const at::Tensor & scale_b, const ::std::optional<at::Tensor> & bias={}, const ::std::optional<at::Tensor> & scale_result={}, ::std::optional<at::ScalarType> out_dtype=::std::nullopt, bool use_fast_accum=false) {
+    return at::_ops::_scaled_mm_out::call(self, mat2, scale_a, scale_b, bias, scale_result, out_dtype, use_fast_accum, out);
+}
+// aten::_scaled_mm.out(Tensor self, Tensor mat2, Tensor scale_a, Tensor scale_b, Tensor? bias=None, Tensor? scale_result=None, ScalarType? out_dtype=None, bool use_fast_accum=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _scaled_mm_outf(const at::Tensor & self, const at::Tensor & mat2, const at::Tensor & scale_a, const at::Tensor & scale_b, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & scale_result, ::std::optional<at::ScalarType> out_dtype, bool use_fast_accum, at::Tensor & out) {
+    return at::_ops::_scaled_mm_out::call(self, mat2, scale_a, scale_b, bias, scale_result, out_dtype, use_fast_accum, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_ff_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_ff_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..44304158c3bb87821b17ecf28862031942be9944
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_ff_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_ff_(at::Tensor & self, int64_t n, const at::Tensor & sobolstate, int64_t dimension, int64_t num_generated);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_initialize_state_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_initialize_state_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..84a9959339e022599e14a7b0ba2936f05ecc63e1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_initialize_state_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 _sobol_engine_initialize_state_ {
+  using schema = at::Tensor & (at::Tensor &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sobol_engine_initialize_state_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_sobol_engine_initialize_state_(Tensor(a!) self, int dimension) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, int64_t dimension);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, int64_t dimension);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_broadcast_to_copy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_broadcast_to_copy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9c997e9deffc7a1f5643d56a6a0c4123f5dce004
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_broadcast_to_copy_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 & _sparse_broadcast_to_copy_out(const at::Tensor & self, at::IntArrayRef size, at::Tensor & out);
+TORCH_API at::Tensor _sparse_broadcast_to_copy(const at::Tensor & self, at::IntArrayRef size);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_and_tensors_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_and_tensors_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..60f504df027ff3f52b5032c68c5fbf3d4fe649af
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_and_tensors_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 & _sparse_coo_tensor_with_dims_and_tensors_out(at::Tensor & out, int64_t sparse_dim, int64_t dense_dim, at::IntArrayRef size, const at::Tensor & indices, const at::Tensor & values, ::std::optional<bool> is_coalesced=::std::nullopt);
+TORCH_API at::Tensor & _sparse_coo_tensor_with_dims_and_tensors_outf(int64_t sparse_dim, int64_t dense_dim, at::IntArrayRef size, const at::Tensor & indices, const at::Tensor & values, ::std::optional<bool> is_coalesced, at::Tensor & out);
+TORCH_API at::Tensor & _sparse_coo_tensor_with_dims_and_tensors_symint_out(at::Tensor & out, int64_t sparse_dim, int64_t dense_dim, c10::SymIntArrayRef size, const at::Tensor & indices, const at::Tensor & values, ::std::optional<bool> is_coalesced=::std::nullopt);
+TORCH_API at::Tensor & _sparse_coo_tensor_with_dims_and_tensors_symint_outf(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 compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..87a19490246e32a018e7feb708af386ba430ef7c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_coo_tensor_with_dims_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_coo_tensor_with_dims_out(at::Tensor & out, int64_t sparse_dim, int64_t dense_dim, at::IntArrayRef size);
+TORCH_API at::Tensor & _sparse_coo_tensor_with_dims_outf(int64_t sparse_dim, int64_t dense_dim, at::IntArrayRef size, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_csr_prod.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_csr_prod.h
new file mode 100644
index 0000000000000000000000000000000000000000..61b8a4e0fc820ba6bd2f04456992a8d9fc1fbd1d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_csr_prod.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_csr_prod_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_csr_prod.dim_dtype(Tensor self, int[1] dim, bool keepdim=False, *, ScalarType? dtype=None) -> Tensor
+inline at::Tensor _sparse_csr_prod(const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt) {
+    return at::_ops::_sparse_csr_prod_dim_dtype::call(self, dim, keepdim, dtype);
+}
+
+// aten::_sparse_csr_prod.dim_dtype_out(Tensor self, int[1] dim, bool keepdim=False, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_csr_prod_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt) {
+    return at::_ops::_sparse_csr_prod_dim_dtype_out::call(self, dim, keepdim, dtype, out);
+}
+// aten::_sparse_csr_prod.dim_dtype_out(Tensor self, int[1] dim, bool keepdim=False, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_csr_prod_outf(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out) {
+    return at::_ops::_sparse_csr_prod_dim_dtype_out::call(self, dim, keepdim, dtype, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_csr_prod_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_csr_prod_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..818a8b670f47758d4097887635ca1b7c4bb1c91d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_csr_prod_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_prod_dim_dtype {
+  using schema = at::Tensor (const at::Tensor &, at::IntArrayRef, bool, ::std::optional<at::ScalarType>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sparse_csr_prod";
+  static constexpr const char* overload_name = "dim_dtype";
+  static constexpr const char* schema_str = "_sparse_csr_prod.dim_dtype(Tensor self, int[1] dim, bool keepdim=False, *, ScalarType? dtype=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype);
+};
+
+struct TORCH_API _sparse_csr_prod_dim_dtype_out {
+  using schema = at::Tensor & (const at::Tensor &, at::IntArrayRef, 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::_sparse_csr_prod";
+  static constexpr const char* overload_name = "dim_dtype_out";
+  static constexpr const char* schema_str = "_sparse_csr_prod.dim_dtype_out(Tensor self, int[1] dim, bool keepdim=False, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef 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/_sparse_csr_sum_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_csr_sum_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7d75dbb755f7096bb6ef482bd5f87191aef7ca5d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_csr_sum_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_csr_sum_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & _sparse_csr_sum_outf(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..17628632e20e6ed0310e9c14d22308114255e07f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm_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 _sparse_mm(const at::Tensor & sparse, const at::Tensor & dense);
+TORCH_API at::Tensor _sparse_mm(const at::Tensor & sparse, const at::Tensor & dense, c10::string_view reduce);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm_reduce_impl_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm_reduce_impl_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..cbb1df963f611ebe6aa5a60d67ae08b3fc1f4eb9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm_reduce_impl_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/_sparse_mm_reduce_impl_backward_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_mm_reduce_impl_backward(Tensor self, Tensor grad_out, Tensor weight, str reduce, Tensor arg_out, bool[2] output_mask) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _sparse_mm_reduce_impl_backward(const at::Tensor & self, const at::Tensor & grad_out, const at::Tensor & weight, c10::string_view reduce, const at::Tensor & arg_out, ::std::array<bool,2> output_mask) {
+    return at::_ops::_sparse_mm_reduce_impl_backward::call(self, grad_out, weight, reduce, arg_out, output_mask);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_addmm.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_addmm.h
new file mode 100644
index 0000000000000000000000000000000000000000..e825a080a64d4127541dcf51e590b7218545e00c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_addmm.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/_sparse_semi_structured_addmm_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_semi_structured_addmm(Tensor input, Tensor mat1, Tensor mat1_meta, Tensor mat2, *, Scalar alpha=1, Scalar beta=1, ScalarType? out_dtype=None) -> Tensor
+inline at::Tensor _sparse_semi_structured_addmm(const at::Tensor & input, const at::Tensor & mat1, const at::Tensor & mat1_meta, const at::Tensor & mat2, const at::Scalar & alpha=1, const at::Scalar & beta=1, ::std::optional<at::ScalarType> out_dtype=::std::nullopt) {
+    return at::_ops::_sparse_semi_structured_addmm::call(input, mat1, mat1_meta, mat2, alpha, beta, out_dtype);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_addmm_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_addmm_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..6a8bb544a35aa0ca0e97dddb11c199b9890bf42a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_addmm_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_addmm {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Scalar &, const at::Scalar &, ::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_addmm";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_sparse_semi_structured_addmm(Tensor input, Tensor mat1, Tensor mat1_meta, Tensor mat2, *, Scalar alpha=1, Scalar beta=1, ScalarType? out_dtype=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & input, const at::Tensor & mat1, const at::Tensor & mat1_meta, const at::Tensor & mat2, const at::Scalar & alpha, const at::Scalar & beta, ::std::optional<at::ScalarType> out_dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const at::Tensor & mat1, const at::Tensor & mat1_meta, const at::Tensor & mat2, const at::Scalar & alpha, const at::Scalar & beta, ::std::optional<at::ScalarType> out_dtype);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_apply.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_apply.h
new file mode 100644
index 0000000000000000000000000000000000000000..17182096b671b68be92428b1a36bfaf892fb871e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_apply.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/_sparse_semi_structured_apply_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_semi_structured_apply(Tensor input, Tensor thread_masks) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _sparse_semi_structured_apply(const at::Tensor & input, const at::Tensor & thread_masks) {
+    return at::_ops::_sparse_semi_structured_apply::call(input, thread_masks);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_linear.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_linear.h
new file mode 100644
index 0000000000000000000000000000000000000000..27ce33cc90d5958de5783b66b3d9e2528ba4430f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_linear.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/_sparse_semi_structured_linear_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_semi_structured_linear(Tensor input, Tensor weight, Tensor meta, *, Tensor? bias=None, str? activation=None, ScalarType? out_dtype=None) -> Tensor
+inline at::Tensor _sparse_semi_structured_linear(const at::Tensor & input, const at::Tensor & weight, const at::Tensor & meta, const ::std::optional<at::Tensor> & bias={}, ::std::optional<c10::string_view> activation=::std::nullopt, ::std::optional<at::ScalarType> out_dtype=::std::nullopt) {
+    return at::_ops::_sparse_semi_structured_linear::call(input, weight, meta, bias, activation, out_dtype);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_softmax_backward_data.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_softmax_backward_data.h
new file mode 100644
index 0000000000000000000000000000000000000000..02d254fd98fe4f4543ca957f9151c7bd3082d297
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_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_softmax_backward_data_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_softmax_backward_data(Tensor grad_output, Tensor output, int dim, Tensor self) -> Tensor
+inline at::Tensor _sparse_softmax_backward_data(const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, const at::Tensor & self) {
+    return at::_ops::_sparse_softmax_backward_data::call(grad_output, output, dim, self);
+}
+
+// aten::_sparse_softmax_backward_data.out(Tensor grad_output, Tensor output, int dim, Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_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_softmax_backward_data_out::call(grad_output, output, dim, self, out);
+}
+// aten::_sparse_softmax_backward_data.out(Tensor grad_output, Tensor output, int dim, Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_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_softmax_backward_data_out::call(grad_output, output, dim, self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_softmax_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_softmax_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..fc7c5dcdbf3b690b67e94e9a48bf8262f0f35825
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_softmax_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 _sparse_softmax(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor _sparse_softmax(const at::Tensor & self, at::Dimname dim, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & _sparse_softmax_out(const at::Tensor & self, int64_t dim, bool half_to_float, at::Tensor & out);
+TORCH_API at::Tensor softmax_sparse_cpu(const at::Tensor & self, int64_t dim, bool half_to_float);
+TORCH_API at::Tensor softmax_sparse_cuda(const at::Tensor & self, int64_t dim, bool half_to_float);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_standard_gamma_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_standard_gamma_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..11e36708e9a5332a4973c788e13a07d1a0d25fb2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_standard_gamma_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 & _standard_gamma_out(const at::Tensor & self, ::std::optional<at::Generator> generator, at::Tensor & out);
+TORCH_API at::Tensor _s_gamma_cpu(const at::Tensor & self, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor _s_gamma_cuda(const at::Tensor & self, ::std::optional<at::Generator> generator=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch_view.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch_view.h
new file mode 100644
index 0000000000000000000000000000000000000000..eb5b1c7cc75ef24100d3fca5145e84aa9605a3ec
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch_view.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/_test_autograd_multiple_dispatch_view_ops.h>
+
+namespace at {
+
+
+// aten::_test_autograd_multiple_dispatch_view(Tensor(a) self) -> Tensor(a)
+inline at::Tensor _test_autograd_multiple_dispatch_view(const at::Tensor & self) {
+    return at::_ops::_test_autograd_multiple_dispatch_view::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch_view_copy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch_view_copy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6400cb646ca8eaf751f358070dc1c2c7c5f2f159
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch_view_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 _test_autograd_multiple_dispatch_view_copy(const at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_floatlist_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_floatlist_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..004f03fc12d771b7b3ea7e751bd8685333cb852c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_floatlist_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 _test_optional_floatlist {
+  using schema = at::Tensor (const at::Tensor &, ::std::optional<at::ArrayRef<double>>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_test_optional_floatlist";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_test_optional_floatlist(Tensor values, float[]? addends) -> Tensor";
+  static at::Tensor call(const at::Tensor & values, ::std::optional<at::ArrayRef<double>> addends);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & values, ::std::optional<at::ArrayRef<double>> addends);
+};
+
+struct TORCH_API _test_optional_floatlist_out {
+  using schema = at::Tensor & (const at::Tensor &, ::std::optional<at::ArrayRef<double>>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_test_optional_floatlist";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_test_optional_floatlist.out(Tensor values, float[]? addends, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & values, ::std::optional<at::ArrayRef<double>> addends, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & values, ::std::optional<at::ArrayRef<double>> addends, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_intlist_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_intlist_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..ea10a97ea04ad177b2ce22fc4d673b51ac1d7f40
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_intlist_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 & _test_optional_intlist_out(const at::Tensor & values, at::OptionalIntArrayRef addends, at::Tensor & out);
+TORCH_API at::Tensor _test_optional_intlist(const at::Tensor & values, at::OptionalIntArrayRef addends);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_parallel_materialize.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_parallel_materialize.h
new file mode 100644
index 0000000000000000000000000000000000000000..fbf894fd5561b8b7aa8a657ab9c77e049e97fdf0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_parallel_materialize.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/_test_parallel_materialize_ops.h>
+
+namespace at {
+
+
+// aten::_test_parallel_materialize(Tensor self, int num_parallel, bool skip_first=False) -> Tensor
+inline at::Tensor _test_parallel_materialize(const at::Tensor & self, int64_t num_parallel, bool skip_first=false) {
+    return at::_ops::_test_parallel_materialize::call(self, num_parallel, skip_first);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_serialization_subcmul.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_serialization_subcmul.h
new file mode 100644
index 0000000000000000000000000000000000000000..567b8541556a485b06f4fa8927fc22334d834376
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_serialization_subcmul.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/_test_serialization_subcmul_ops.h>
+
+namespace at {
+
+
+// aten::_test_serialization_subcmul(Tensor self, Tensor other, Scalar alpha=1) -> Tensor
+inline at::Tensor _test_serialization_subcmul(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1) {
+    return at::_ops::_test_serialization_subcmul::call(self, other, alpha);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1f0cd632407f47578c44d644769cfb5a683e4810
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_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,at::Tensor,at::Tensor,at::Tensor> _thnn_fused_lstm_cell_backward(const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, const at::Tensor & cx, const at::Tensor & cy, const at::Tensor & workspace, bool has_bias);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_to_dense_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_dense_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7fa8f7957604f3ae465863d3640e23904421edbf
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_dense_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 & _to_dense_out(at::Tensor & out, const at::Tensor & self, ::std::optional<at::ScalarType> dtype=::std::nullopt, ::std::optional<bool> masked_grad=::std::nullopt);
+TORCH_API at::Tensor & _to_dense_outf(const at::Tensor & self, ::std::optional<at::ScalarType> dtype, ::std::optional<bool> masked_grad, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_bsr_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_bsr_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..28100ec684c4ad99cc13990e27db42a53df8f08d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_bsr_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 & _to_sparse_bsr_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef blocksize, ::std::optional<int64_t> dense_dim=::std::nullopt);
+TORCH_API at::Tensor & _to_sparse_bsr_outf(const at::Tensor & self, at::IntArrayRef blocksize, ::std::optional<int64_t> dense_dim, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_csc_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_csc_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d18ae6e1d52fc20ca080206f10fc0d5426fa45a8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_csc_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 & _to_sparse_csc_out(const at::Tensor & self, ::std::optional<int64_t> dense_dim, at::Tensor & out);
+TORCH_API at::Tensor dense_to_sparse_csc(const at::Tensor & self, ::std::optional<int64_t> dense_dim=::std::nullopt);
+TORCH_API at::Tensor coo_to_sparse_csc(const at::Tensor & self, ::std::optional<int64_t> dense_dim=::std::nullopt);
+TORCH_API at::Tensor sparse_compressed_to_sparse_csc(const at::Tensor & self, ::std::optional<int64_t> dense_dim=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_triton_scaled_dot_attention_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_triton_scaled_dot_attention_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..70f5d19f469d67da0644ce977858958d000b80f8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_triton_scaled_dot_attention_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 & _triton_scaled_dot_attention_out(const at::Tensor & q, const at::Tensor & k, const at::Tensor & v, double dropout_p, at::Tensor & out);
+TORCH_API at::Tensor triton_scaled_dot_attention(const at::Tensor & q, const at::Tensor & k, const at::Tensor & v, double dropout_p=0.0);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_unique2.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_unique2.h
new file mode 100644
index 0000000000000000000000000000000000000000..a8f7145a566eba30a8bbaaec490d247c260e6c72
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_unique2.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/_unique2_ops.h>
+
+namespace at {
+
+
+// aten::_unique2(Tensor self, bool sorted=True, bool return_inverse=False, bool return_counts=False) -> (Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _unique2(const at::Tensor & self, bool sorted=true, bool return_inverse=false, bool return_counts=false) {
+    return at::_ops::_unique2::call(self, sorted, return_inverse, return_counts);
+}
+
+// aten::_unique2.out(Tensor self, bool sorted=True, 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 &> _unique2_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & self, bool sorted=true, bool return_inverse=false, bool return_counts=false) {
+    return at::_ops::_unique2_out::call(self, sorted, return_inverse, return_counts, out0, out1, out2);
+}
+// aten::_unique2.out(Tensor self, bool sorted=True, 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 &> _unique2_outf(const at::Tensor & self, bool sorted, bool return_inverse, bool return_counts, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2) {
+    return at::_ops::_unique2_out::call(self, sorted, return_inverse, return_counts, out0, out1, out2);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_unsafe_index_put_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_unsafe_index_put_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..ecac6314ad0c210b5507c2577d9e7231ff7c0ba2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_unsafe_index_put_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 _unsafe_index_put(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bicubic2d_aa_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bicubic2d_aa_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2fce473f8a8a75d2255da60095cb0035073e83f2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bicubic2d_aa_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_bicubic2d_aa(const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor _upsample_bicubic2d_aa_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_bicubic2d_aa_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_bicubic2d_aa_outf(const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+TORCH_API at::Tensor & _upsample_bicubic2d_aa_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_bicubic2d_aa_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::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/_upsample_bicubic2d_aa_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bicubic2d_aa_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..660435a04a499f03aee7b9e9e6f78fff4704eab8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bicubic2d_aa_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_bicubic2d_aa(const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor _upsample_bicubic2d_aa_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_bicubic2d_aa_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_bicubic2d_aa_outf(const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+TORCH_API at::Tensor & _upsample_bicubic2d_aa_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_bicubic2d_aa_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bilinear2d_aa_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bilinear2d_aa_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d460b50a48a2bce49c432dc68af1523294f288d3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bilinear2d_aa_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_bilinear2d_aa(const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor _upsample_bilinear2d_aa_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..edb473f41914dd1a5399a6e26d72f65abbd00ceb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_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__upsample_nearest_exact1d : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, at::ArrayRef<int64_t> output_size, ::std::optional<double> scales);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact2d_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact2d_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a89ab4e0ed08fe71b4f326b498a067d5f0ad4be7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact2d_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_exact2d(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor _upsample_nearest_exact2d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_nearest_exact2d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_nearest_exact2d_outf(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+TORCH_API at::Tensor & _upsample_nearest_exact2d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_nearest_exact2d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact3d_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact3d_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b72b4b21b0b02f8d7ef7ae1c2e6867befc24855e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact3d_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_nearest_exact3d(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_nearest_exact3d_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);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact3d_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact3d_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e559d5f7bd0f9077d0e94583623b01e13c994180
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact3d_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_nearest_exact3d(const at::Tensor & input, at::OptionalIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors);
+TORCH_API at::Tensor _upsample_nearest_exact3d_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/_use_cudnn_ctc_loss_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_use_cudnn_ctc_loss_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..e159622e3382c62b1db8750eb988e5b76b295a59
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_use_cudnn_ctc_loss_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 _use_cudnn_ctc_loss {
+  using schema = bool (const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_use_cudnn_ctc_loss";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_use_cudnn_ctc_loss(Tensor log_probs, Tensor targets, int[] input_lengths, int[] target_lengths, int blank) -> bool";
+  static bool call(const at::Tensor & log_probs, const at::Tensor & targets, at::IntArrayRef input_lengths, at::IntArrayRef target_lengths, int64_t blank);
+  static bool redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & log_probs, const at::Tensor & targets, at::IntArrayRef input_lengths, at::IntArrayRef target_lengths, int64_t blank);
+};
+
+struct TORCH_API _use_cudnn_ctc_loss_Tensor {
+  using schema = bool (const 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::_use_cudnn_ctc_loss";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "_use_cudnn_ctc_loss.Tensor(Tensor log_probs, Tensor targets, Tensor input_lengths, Tensor target_lengths, int blank) -> bool";
+  static bool call(const at::Tensor & log_probs, const at::Tensor & targets, const at::Tensor & input_lengths, const at::Tensor & target_lengths, int64_t blank);
+  static bool redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & log_probs, const at::Tensor & targets, const at::Tensor & input_lengths, const at::Tensor & target_lengths, int64_t blank);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_bsr_tensor_args_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_bsr_tensor_args_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..dfc04c0255489d59b146d6824a65b449d4a3339f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_bsr_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_bsr_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/_values_copy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_values_copy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..aa460e7de9ee8159633f4e10091b464b36db561d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_values_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 _values_copy(const at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_version.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_version.h
new file mode 100644
index 0000000000000000000000000000000000000000..f227514c58b3d5b65db0eacc8abc2f1adf57f2ea
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_version.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/_version_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..24570368f7eb789d6461b39e384258aabbf2bd99
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_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 ::std::tuple<at::Tensor &,at::Tensor &> _weight_norm_interface_out(const at::Tensor & v, const at::Tensor & g, int64_t dim, at::Tensor & out0, at::Tensor & out1);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> weight_norm_cpu(const at::Tensor & v, const at::Tensor & g, int64_t dim=0);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> weight_norm_cuda(const at::Tensor & v, const at::Tensor & g, int64_t dim=0);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..0e69f838c2e88dab92866a5af174ca1952d348fa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_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 _weight_norm {
+  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::_weight_norm";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_weight_norm(Tensor v, Tensor g, int dim=0) -> Tensor";
+  static at::Tensor call(const at::Tensor & v, const at::Tensor & g, int64_t dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & v, const at::Tensor & g, int64_t dim);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/acos_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/acos_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5f98c953afb1bb664c888f33ad8c4e948409345e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/acos_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 acos(const at::Tensor & self);
+TORCH_API at::Tensor & acos_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & acos_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & acos_(at::Tensor & self);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/acosh_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/acosh_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..eaa3193749230bc524e44279a354e625ed347b5a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/acosh_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 acosh {
+  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::acosh";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "acosh(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 acosh_ {
+  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::acosh_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "acosh_(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 acosh_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::acosh";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "acosh.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_pool3d_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool3d_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..5bcbef5de55b99865011aae3c19f91eb2171c193
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool3d_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 adaptive_avg_pool3d_backward_grad_input {
+  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::adaptive_avg_pool3d_backward";
+  static constexpr const char* overload_name = "grad_input";
+  static constexpr const char* schema_str = "adaptive_avg_pool3d_backward.grad_input(Tensor grad_output, Tensor self, *, Tensor(a!) grad_input) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & grad_output, const at::Tensor & self, at::Tensor & grad_input);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, at::Tensor & grad_input);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool3d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool3d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..f27f8ca31079ad187b7839e97a17371355d45f1e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool3d_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 adaptive_avg_pool3d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size);
+TORCH_API at::Tensor & adaptive_avg_pool3d_out_cpu(const at::Tensor & self, at::IntArrayRef output_size, at::Tensor & out);
+TORCH_API at::Tensor & adaptive_avg_pool3d_out_cuda(const at::Tensor & self, at::IntArrayRef output_size, at::Tensor & out);
+TORCH_API at::Tensor & adaptive_avg_pool3d_out_quantized_cpu(const at::Tensor & self, at::IntArrayRef output_size, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..e6870534f54c029a5069a0e603b1bca6a04e531d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_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/adaptive_max_pool2d_backward_ops.h>
+
+namespace at {
+
+
+// aten::adaptive_max_pool2d_backward.grad_input(Tensor grad_output, Tensor self, Tensor indices, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & adaptive_max_pool2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & indices) {
+    return at::_ops::adaptive_max_pool2d_backward_grad_input::call(grad_output, self, indices, grad_input);
+}
+// aten::adaptive_max_pool2d_backward.grad_input(Tensor grad_output, Tensor self, Tensor indices, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & adaptive_max_pool2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & indices, at::Tensor & grad_input) {
+    return at::_ops::adaptive_max_pool2d_backward_grad_input::call(grad_output, self, indices, grad_input);
+}
+
+// aten::adaptive_max_pool2d_backward(Tensor grad_output, Tensor self, Tensor indices) -> Tensor
+inline at::Tensor adaptive_max_pool2d_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & indices) {
+    return at::_ops::adaptive_max_pool2d_backward::call(grad_output, self, indices);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..448a08dbb36e722245c72b8c0c5c94badd600a7d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_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/adaptive_max_pool3d_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_adaptive_max_pool3d_out_cpu : public at::meta::structured_adaptive_max_pool3d {
+void impl(const at::Tensor & self, at::IntArrayRef output_size, const at::Tensor & out, const at::Tensor & indices);
+};
+struct TORCH_API structured_adaptive_max_pool3d_out_cuda : public at::meta::structured_adaptive_max_pool3d {
+void impl(const at::Tensor & self, at::IntArrayRef output_size, const at::Tensor & out, const at::Tensor & indices);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addbmm_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addbmm_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4491fdd247be48c7d52b96254591a22014c558da
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addbmm_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 & addbmm_(at::Tensor & self, const at::Tensor & batch1, const at::Tensor & batch2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addcdiv_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addcdiv_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..a91cbac3a1f82f3a27090f277991f0b611d88ed1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addcdiv_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_addcdiv : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1b4b38e8f3123842fef65e6eff9295c0ef0c9f2c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul_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 addcmul(const at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value=1);
+TORCH_API at::Tensor & addcmul_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value=1);
+TORCH_API at::Tensor & addcmul_outf(const at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value, at::Tensor & out);
+TORCH_API at::Tensor & addcmul_(at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value=1);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addmm_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addmm_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e945077c8928a64ce4bbfec00f4bdfff663dfe84
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addmm_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 addmm(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & addmm_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & addmm_outf(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta, const at::Scalar & alpha, at::Tensor & out);
+TORCH_API at::Tensor & addmm_(at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/affine_grid_generator_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/affine_grid_generator_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..0ea0d0394314dbb1ffc26ab3c8b7faa924f1cdf9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/affine_grid_generator_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/affine_grid_generator_backward_ops.h>
+
+namespace at {
+
+
+// aten::affine_grid_generator_backward(Tensor grad, SymInt[] size, bool align_corners) -> Tensor
+inline at::Tensor affine_grid_generator_backward(const at::Tensor & grad, at::IntArrayRef size, bool align_corners) {
+    return at::_ops::affine_grid_generator_backward::call(grad, c10::fromIntArrayRefSlow(size), align_corners);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor affine_grid_generator_backward(const at::Tensor & grad, at::IntArrayRef size, bool align_corners) {
+    return at::_ops::affine_grid_generator_backward::call(grad, c10::fromIntArrayRefSlow(size), align_corners);
+  }
+}
+
+// aten::affine_grid_generator_backward(Tensor grad, SymInt[] size, bool align_corners) -> Tensor
+inline at::Tensor affine_grid_generator_backward_symint(const at::Tensor & grad, c10::SymIntArrayRef size, bool align_corners) {
+    return at::_ops::affine_grid_generator_backward::call(grad, size, align_corners);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor affine_grid_generator_backward(const at::Tensor & grad, c10::SymIntArrayRef size, bool align_corners) {
+    return at::_ops::affine_grid_generator_backward::call(grad, size, align_corners);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/affine_grid_generator_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/affine_grid_generator_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..08e608404e4ddcf753d0a9e777561887b1a77462
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/affine_grid_generator_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 affine_grid_generator(const at::Tensor & theta, at::IntArrayRef size, bool align_corners);
+TORCH_API at::Tensor & affine_grid_generator_out_symint(const at::Tensor & theta, c10::SymIntArrayRef size, bool align_corners, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/alias_copy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/alias_copy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4b2b8e738af1639b54020aa79537b872931fab5e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/alias_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 alias_copy(const at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/all_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/all_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..f5a99d497857f2d6137755ee494499d0606e579a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/all_meta.h
@@ -0,0 +1,37 @@
+#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_all_dim : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, int64_t dim, bool keepdim);
+};
+struct TORCH_API structured_all_dims : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim);
+};
+struct TORCH_API structured_all : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/angle_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/angle_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c09727f87089fa64a305cf51c185393520942257
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/angle_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 angle(const at::Tensor & self);
+TORCH_API at::Tensor & angle_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & angle_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arange.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arange.h
new file mode 100644
index 0000000000000000000000000000000000000000..1bee7b7bb4a41d805be1d17734c9e7f9a31986cf
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arange.h
@@ -0,0 +1,70 @@
+#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/arange_ops.h>
+
+namespace at {
+
+
+// aten::arange(Scalar end, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor arange(const at::Scalar & end, at::TensorOptions options={}) {
+    return at::_ops::arange::call(end, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::arange(Scalar end, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor arange(const at::Scalar & end, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::arange::call(end, dtype, layout, device, pin_memory);
+}
+
+// aten::arange.start(Scalar start, Scalar end, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor arange(const at::Scalar & start, const at::Scalar & end, at::TensorOptions options={}) {
+    return at::_ops::arange_start::call(start, end, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::arange.start(Scalar start, Scalar end, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor arange(const at::Scalar & start, const at::Scalar & end, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::arange_start::call(start, end, dtype, layout, device, pin_memory);
+}
+
+// aten::arange.start_step(Scalar start, Scalar end, Scalar step=1, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor arange(const at::Scalar & start, const at::Scalar & end, const at::Scalar & step, at::TensorOptions options={}) {
+    return at::_ops::arange_start_step::call(start, end, step, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::arange.start_step(Scalar start, Scalar end, Scalar step=1, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor arange(const at::Scalar & start, const at::Scalar & end, const at::Scalar & step, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::arange_start_step::call(start, end, step, dtype, layout, device, pin_memory);
+}
+
+// aten::arange.out(Scalar end, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arange_out(at::Tensor & out, const at::Scalar & end) {
+    return at::_ops::arange_out::call(end, out);
+}
+// aten::arange.out(Scalar end, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arange_outf(const at::Scalar & end, at::Tensor & out) {
+    return at::_ops::arange_out::call(end, out);
+}
+
+// aten::arange.start_out(Scalar start, Scalar end, Scalar step=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arange_out(at::Tensor & out, const at::Scalar & start, const at::Scalar & end, const at::Scalar & step) {
+    return at::_ops::arange_start_out::call(start, end, step, out);
+}
+// aten::arange.start_out(Scalar start, Scalar end, Scalar step=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arange_outf(const at::Scalar & start, const at::Scalar & end, const at::Scalar & step, at::Tensor & out) {
+    return at::_ops::arange_start_out::call(start, end, step, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arccosh_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arccosh_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5646fa0d8641bd5abc6e9ff49b1ad333ca64228d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arccosh_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 arccosh(const at::Tensor & self);
+TORCH_API at::Tensor & arccosh_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & arccosh_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & arccosh_(at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arcsin_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arcsin_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..04ad1809fc686618f5d1808ee5ec6779e9aaf4e4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arcsin_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 arcsin(const at::Tensor & self);
+TORCH_API at::Tensor & arcsin_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & arcsin_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & arcsin_(at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arctan2_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arctan2_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..1fc96d27c73b6993102d89c7dd73cd2d2839d6b7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arctan2_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 arctan2(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & arctan2_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & arctan2_(at::Tensor & self, const at::Tensor & other);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/argsort_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/argsort_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..5ea66b6678065467d3932a247aa9a745a11097a7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/argsort_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 argsort {
+  using schema = at::Tensor (const at::Tensor &, int64_t, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::argsort";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "argsort(Tensor self, int dim=-1, bool descending=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t dim, bool descending);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, bool descending);
+};
+
+struct TORCH_API argsort_stable {
+  using schema = at::Tensor (const at::Tensor &, bool, int64_t, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::argsort";
+  static constexpr const char* overload_name = "stable";
+  static constexpr const char* schema_str = "argsort.stable(Tensor self, *, bool stable, int dim=-1, bool descending=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, bool stable, int64_t dim, bool descending);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, bool stable, int64_t dim, bool descending);
+};
+
+struct TORCH_API argsort_stable_out {
+  using schema = at::Tensor & (const at::Tensor &, bool, int64_t, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::argsort";
+  static constexpr const char* overload_name = "stable_out";
+  static constexpr const char* schema_str = "argsort.stable_out(Tensor self, *, bool stable, int dim=-1, bool descending=False, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, bool stable, int64_t dim, bool descending, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, bool stable, int64_t dim, bool descending, at::Tensor & out);
+};
+
+struct TORCH_API argsort_dimname {
+  using schema = at::Tensor (const at::Tensor &, at::Dimname, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::argsort";
+  static constexpr const char* overload_name = "dimname";
+  static constexpr const char* schema_str = "argsort.dimname(Tensor self, Dimname dim, bool descending=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::Dimname dim, bool descending);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname dim, bool descending);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/as_strided_copy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/as_strided_copy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4e5f95b885ec111119c01df1a5685870a1ba2d7e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/as_strided_copy_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 & as_strided_copy_out_symint(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<c10::SymInt> storage_offset, at::Tensor & out);
+TORCH_API at::Tensor as_strided_copy_symint(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<c10::SymInt> storage_offset=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/as_strided_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/as_strided_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c194e98c608f4f1eaa2cc8fbd217d13ff688d081
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/as_strided_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 as_strided_tensorimpl(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride, ::std::optional<int64_t> storage_offset=::std::nullopt);
+TORCH_API at::Tensor as_strided_tensorimpl_meta_symint(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<c10::SymInt> storage_offset=::std::nullopt);
+TORCH_API at::Tensor as_strided_qtensorimpl(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride, ::std::optional<int64_t> storage_offset=::std::nullopt);
+TORCH_API const at::Tensor & as_strided__symint(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<c10::SymInt> storage_offset=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/asinh_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/asinh_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a46699b1b7bb12db09dbba8d01679620eaed14db
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/asinh_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 asinh {
+  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::asinh";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "asinh(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 asinh_ {
+  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::asinh_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "asinh_(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 asinh_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::asinh";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "asinh.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/atan2_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/atan2_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8a2f5831e1a9510e71dc5b35072f722566219b2c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/atan2_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 atan2(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & atan2_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & atan2_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & atan2_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool3d_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool3d_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f9bd50a263aa89b3e6b84f3a9aea210ff213d7d7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_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 at::Tensor avg_pool3d(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, bool ceil_mode=false, bool count_include_pad=true, ::std::optional<int64_t> divisor_override=::std::nullopt);
+TORCH_API at::Tensor & avg_pool3d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, bool ceil_mode=false, bool count_include_pad=true, ::std::optional<int64_t> divisor_override=::std::nullopt);
+TORCH_API at::Tensor & avg_pool3d_outf(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 & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..14f581b2575a6c5aaf8db987811d39d6f18ccb8e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/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> batch_norm_backward(const at::Tensor & grad_out, const at::Tensor & input, 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, bool update, double eps, ::std::array<bool,3> output_mask, const at::Tensor & reserve);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_backward_elemt_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_backward_elemt_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c87507efece85d0baa1fb5d2362546f679cf676f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_backward_elemt_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 & batch_norm_backward_elemt_out(const at::Tensor & grad_out, const at::Tensor & input, const at::Tensor & mean, const at::Tensor & invstd, const ::std::optional<at::Tensor> & weight, const at::Tensor & sum_dy, const at::Tensor & sum_dy_xmu, const at::Tensor & count, at::Tensor & out);
+TORCH_API at::Tensor batch_norm_backward_elemt_cuda(const at::Tensor & grad_out, const at::Tensor & input, const at::Tensor & mean, const at::Tensor & invstd, const ::std::optional<at::Tensor> & weight, const at::Tensor & sum_dy, const at::Tensor & sum_dy_xmu, const at::Tensor & count);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b86c0b04f9829fd18c3261d00294e062d6d16007
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_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_right_shift(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_right_shift_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_xor_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_xor_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..889e22f940ee2963ee549750dd8ce15132532b07
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_xor_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 bitwise_xor(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_xor_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_xor_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & bitwise_xor_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bucketize.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bucketize.h
new file mode 100644
index 0000000000000000000000000000000000000000..14c2b82548395e0e1218c688ea8f45d0288d044b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bucketize.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/bucketize_ops.h>
+
+namespace at {
+
+
+// aten::bucketize.Tensor(Tensor self, Tensor boundaries, *, bool out_int32=False, bool right=False) -> Tensor
+inline at::Tensor bucketize(const at::Tensor & self, const at::Tensor & boundaries, bool out_int32=false, bool right=false) {
+    return at::_ops::bucketize_Tensor::call(self, boundaries, out_int32, right);
+}
+
+// aten::bucketize.Tensor_out(Tensor self, Tensor boundaries, *, bool out_int32=False, bool right=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & bucketize_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & boundaries, bool out_int32=false, bool right=false) {
+    return at::_ops::bucketize_Tensor_out::call(self, boundaries, out_int32, right, out);
+}
+// aten::bucketize.Tensor_out(Tensor self, Tensor boundaries, *, bool out_int32=False, bool right=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & bucketize_outf(const at::Tensor & self, const at::Tensor & boundaries, bool out_int32, bool right, at::Tensor & out) {
+    return at::_ops::bucketize_Tensor_out::call(self, boundaries, out_int32, right, out);
+}
+
+// aten::bucketize.Scalar(Scalar self, Tensor boundaries, *, bool out_int32=False, bool right=False) -> Tensor
+inline at::Tensor bucketize(const at::Scalar & self, const at::Tensor & boundaries, bool out_int32=false, bool right=false) {
+    return at::_ops::bucketize_Scalar::call(self, boundaries, out_int32, right);
+}
+
+// aten::bucketize.Scalar_out(Scalar self, Tensor boundaries, *, bool out_int32=False, bool right=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & bucketize_out(at::Tensor & out, const at::Scalar & self, const at::Tensor & boundaries, bool out_int32=false, bool right=false) {
+    return at::_ops::bucketize_Scalar_out::call(self, boundaries, out_int32, right, out);
+}
+// aten::bucketize.Scalar_out(Scalar self, Tensor boundaries, *, bool out_int32=False, bool right=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & bucketize_outf(const at::Scalar & self, const at::Tensor & boundaries, bool out_int32, bool right, at::Tensor & out) {
+    return at::_ops::bucketize_Scalar_out::call(self, boundaries, out_int32, right, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cat_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cat_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..36f6c7f8c89009bb4e201b045263d8ebde3776fb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cat_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/cat_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_cat_out_cpu : public at::meta::structured_cat {
+void impl(const at::ITensorListRef & tensors, int64_t dim, int64_t valid, bool all_contiguous, bool all_same_dtype, bool all_same_sizes_and_stride, at::MemoryFormat memory_format, const at::Tensor & out);
+};
+struct TORCH_API structured_cat_out_cuda : public at::meta::structured_cat {
+void impl(const at::ITensorListRef & tensors, int64_t dim, int64_t valid, bool all_contiguous, bool all_same_dtype, bool all_same_sizes_and_stride, at::MemoryFormat memory_format, const at::Tensor & out);
+};
+TORCH_API at::Tensor cat_nested(const at::ITensorListRef & tensors, int64_t dim=0);
+TORCH_API at::Tensor cat_sparse(const at::ITensorListRef & tensors, int64_t dim=0);
+TORCH_API at::Tensor cat_quantized_cpu(const at::ITensorListRef & tensors, int64_t dim=0);
+TORCH_API at::Tensor & cat_out_quantized_cpu(const at::ITensorListRef & tensors, int64_t dim, at::Tensor & out);
+TORCH_API at::Tensor cat(at::TensorList tensors, at::Dimname dim);
+TORCH_API at::Tensor & cat_out(at::TensorList tensors, at::Dimname dim, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cdist_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cdist_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7095c24ec524eedec8d6347a353c0a160c64bba8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cdist_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 cdist(const at::Tensor & x1, const at::Tensor & x2, double p=2, ::std::optional<int64_t> compute_mode=::std::nullopt);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/channel_shuffle_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/channel_shuffle_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4c73c3a91fc3792f03300f0ec6a97e7b22c2218a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/channel_shuffle_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 channel_shuffle(const at::Tensor & self, int64_t groups);
+TORCH_API at::Tensor channel_shuffle_symint(const at::Tensor & self, c10::SymInt groups);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cholesky_inverse.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cholesky_inverse.h
new file mode 100644
index 0000000000000000000000000000000000000000..7188fc83a8948eb4424a5eb71871bfd8ce88d129
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cholesky_inverse.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/cholesky_inverse_ops.h>
+
+namespace at {
+
+
+// aten::cholesky_inverse(Tensor self, bool upper=False) -> Tensor
+inline at::Tensor cholesky_inverse(const at::Tensor & self, bool upper=false) {
+    return at::_ops::cholesky_inverse::call(self, upper);
+}
+
+// aten::cholesky_inverse.out(Tensor self, bool upper=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cholesky_inverse_out(at::Tensor & out, const at::Tensor & self, bool upper=false) {
+    return at::_ops::cholesky_inverse_out::call(self, upper, out);
+}
+// aten::cholesky_inverse.out(Tensor self, bool upper=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cholesky_inverse_outf(const at::Tensor & self, bool upper, at::Tensor & out) {
+    return at::_ops::cholesky_inverse_out::call(self, upper, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cholesky_inverse_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cholesky_inverse_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..86aa245954b91ed1837388a38d28806cd0723a31
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cholesky_inverse_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_inverse(const at::Tensor & self, bool upper=false);
+TORCH_API at::Tensor & cholesky_inverse_out(at::Tensor & out, const at::Tensor & self, bool upper=false);
+TORCH_API at::Tensor & cholesky_inverse_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/clamp_max_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/clamp_max_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..9222a0677b2f10ad9545dbc1eb85272f8a522757
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/clamp_max_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_clamp_max : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Scalar & max);
+};
+struct TORCH_API structured_clamp_max_Tensor : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Tensor & max);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/clamp_min_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/clamp_min_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..74c113dd37f3927b5f0fabbbe007bed95b8f6bb6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/clamp_min_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/clamp_min_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_clamp_min_out : public at::meta::structured_clamp_min {
+void impl(const at::Tensor & self, const at::Scalar & min, const at::Tensor & out);
+};
+struct TORCH_API structured_clamp_min_Tensor_out : public at::meta::structured_clamp_min_Tensor {
+void impl(const at::Tensor & self, const at::Tensor & min, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/col2im.h b/lib/python3.12/site-packages/torch/include/ATen/ops/col2im.h
new file mode 100644
index 0000000000000000000000000000000000000000..4337bf50d4b18afa092b3425c95fcd25b85032be
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/col2im.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/col2im_ops.h>
+
+namespace at {
+
+
+// aten::col2im.out(Tensor self, SymInt[2] output_size, int[2] kernel_size, int[2] dilation, int[2] padding, int[2] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & col2im_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride) {
+    return at::_ops::col2im_out::call(self, c10::fromIntArrayRefSlow(output_size), kernel_size, dilation, padding, stride, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & col2im_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride) {
+    return at::_ops::col2im_out::call(self, c10::fromIntArrayRefSlow(output_size), kernel_size, dilation, padding, stride, out);
+  }
+}
+
+// aten::col2im.out(Tensor self, SymInt[2] output_size, int[2] kernel_size, int[2] dilation, int[2] padding, int[2] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & col2im_outf(const at::Tensor & self, at::IntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride, at::Tensor & out) {
+    return at::_ops::col2im_out::call(self, c10::fromIntArrayRefSlow(output_size), kernel_size, dilation, padding, stride, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & col2im_outf(const at::Tensor & self, at::IntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride, at::Tensor & out) {
+    return at::_ops::col2im_out::call(self, c10::fromIntArrayRefSlow(output_size), kernel_size, dilation, padding, stride, out);
+  }
+}
+
+// aten::col2im.out(Tensor self, SymInt[2] output_size, int[2] kernel_size, int[2] dilation, int[2] padding, int[2] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & col2im_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride) {
+    return at::_ops::col2im_out::call(self, output_size, kernel_size, dilation, padding, stride, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & col2im_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride) {
+    return at::_ops::col2im_out::call(self, output_size, kernel_size, dilation, padding, stride, out);
+  }
+}
+
+// aten::col2im.out(Tensor self, SymInt[2] output_size, int[2] kernel_size, int[2] dilation, int[2] padding, int[2] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & col2im_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride, at::Tensor & out) {
+    return at::_ops::col2im_out::call(self, output_size, kernel_size, dilation, padding, stride, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & col2im_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride, at::Tensor & out) {
+    return at::_ops::col2im_out::call(self, output_size, kernel_size, dilation, padding, stride, out);
+  }
+}
+
+// aten::col2im(Tensor self, SymInt[2] output_size, int[2] kernel_size, int[2] dilation, int[2] padding, int[2] stride) -> Tensor
+inline at::Tensor col2im(const at::Tensor & self, at::IntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride) {
+    return at::_ops::col2im::call(self, c10::fromIntArrayRefSlow(output_size), kernel_size, dilation, padding, stride);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor col2im(const at::Tensor & self, at::IntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride) {
+    return at::_ops::col2im::call(self, c10::fromIntArrayRefSlow(output_size), kernel_size, dilation, padding, stride);
+  }
+}
+
+// aten::col2im(Tensor self, SymInt[2] output_size, int[2] kernel_size, int[2] dilation, int[2] padding, int[2] stride) -> Tensor
+inline at::Tensor col2im_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride) {
+    return at::_ops::col2im::call(self, output_size, kernel_size, dilation, padding, stride);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor col2im(const at::Tensor & self, c10::SymIntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride) {
+    return at::_ops::col2im::call(self, output_size, kernel_size, dilation, padding, stride);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices.h b/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices.h
new file mode 100644
index 0000000000000000000000000000000000000000..d8e7efd91b9a3404647f51dab062dea977198514
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices.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/col_indices_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/combinations_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/combinations_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6f57c07e72733fcbdaacb98be9ffdf164b9f9276
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/combinations_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 combinations(const at::Tensor & self, int64_t r=2, bool with_replacement=false);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/complex_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/complex_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..ea08bde133fb04b300bfed32e09ab80338c9e07f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/complex_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 complex {
+  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::complex";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "complex(Tensor real, Tensor imag) -> Tensor";
+  static at::Tensor call(const at::Tensor & real, const at::Tensor & imag);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & real, const at::Tensor & imag);
+};
+
+struct TORCH_API complex_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::complex";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "complex.out(Tensor real, Tensor imag, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & real, const at::Tensor & imag, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & real, const at::Tensor & imag, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/conj.h b/lib/python3.12/site-packages/torch/include/ATen/ops/conj.h
new file mode 100644
index 0000000000000000000000000000000000000000..6c5569b0293e24ae5ba71965ef5cbf42f6abfee7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/conj.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/conj_ops.h>
+
+namespace at {
+
+
+// aten::conj(Tensor(a) self) -> Tensor(a)
+inline at::Tensor __dispatch_conj(const at::Tensor & self) {
+    return at::_ops::conj::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/conv_transpose1d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/conv_transpose1d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9d2935f6d2e2bf81be8bf7b47eb7f700f5ceac63
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/conv_transpose1d_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 conv_transpose1d {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymInt, c10::SymIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::conv_transpose1d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "conv_transpose1d(Tensor input, Tensor weight, Tensor? bias=None, SymInt[1] stride=1, SymInt[1] padding=0, SymInt[1] output_padding=0, SymInt groups=1, SymInt[1] dilation=1) -> Tensor";
+  static at::Tensor call(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymInt groups, c10::SymIntArrayRef dilation);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymInt groups, c10::SymIntArrayRef dilation);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..2c5ae1cd19163d28dedcfd380191b557af7eae3b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_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 ::std::tuple<at::Tensor,at::Tensor,at::Tensor> convolution_backward(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & weight, at::OptionalIntArrayRef bias_sizes, 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_out_symint(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & weight, at::OptionalSymIntArrayRef bias_sizes, 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 native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_overrideable.h b/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_overrideable.h
new file mode 100644
index 0000000000000000000000000000000000000000..1327359b0fd7a54498bc1d4c9b549d7a6a7414ed
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_overrideable.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/convolution_overrideable_ops.h>
+
+namespace at {
+
+
+// aten::convolution_overrideable(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, SymInt[] output_padding, SymInt groups) -> Tensor
+inline 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) {
+    return at::_ops::convolution_overrideable::call(input, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), transposed, c10::fromIntArrayRefSlow(output_padding), groups);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  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) {
+    return at::_ops::convolution_overrideable::call(input, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), transposed, c10::fromIntArrayRefSlow(output_padding), groups);
+  }
+}
+
+// aten::convolution_overrideable(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, SymInt[] output_padding, SymInt groups) -> Tensor
+inline 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) {
+    return at::_ops::convolution_overrideable::call(input, weight, bias, stride, padding, dilation, transposed, output_padding, groups);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor convolution_overrideable(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) {
+    return at::_ops::convolution_overrideable::call(input, weight, bias, stride, padding, dilation, transposed, output_padding, groups);
+  }
+}
+
+// aten::convolution_overrideable.out(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, SymInt[] output_padding, SymInt groups, *, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::convolution_overrideable_out::call(input, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), transposed, c10::fromIntArrayRefSlow(output_padding), groups, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  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) {
+    return at::_ops::convolution_overrideable_out::call(input, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), transposed, c10::fromIntArrayRefSlow(output_padding), groups, out);
+  }
+}
+
+// aten::convolution_overrideable.out(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, SymInt[] output_padding, SymInt groups, *, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::convolution_overrideable_out::call(input, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), transposed, c10::fromIntArrayRefSlow(output_padding), groups, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  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) {
+    return at::_ops::convolution_overrideable_out::call(input, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), transposed, c10::fromIntArrayRefSlow(output_padding), groups, out);
+  }
+}
+
+// aten::convolution_overrideable.out(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, SymInt[] output_padding, SymInt groups, *, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::convolution_overrideable_out::call(input, weight, bias, stride, padding, dilation, transposed, output_padding, groups, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & convolution_overrideable_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) {
+    return at::_ops::convolution_overrideable_out::call(input, weight, bias, stride, padding, dilation, transposed, output_padding, groups, out);
+  }
+}
+
+// aten::convolution_overrideable.out(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, SymInt[] output_padding, SymInt groups, *, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::convolution_overrideable_out::call(input, weight, bias, stride, padding, dilation, transposed, output_padding, groups, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & convolution_overrideable_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) {
+    return at::_ops::convolution_overrideable_out::call(input, weight, bias, stride, padding, dilation, transposed, output_padding, groups, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cosh.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cosh.h
new file mode 100644
index 0000000000000000000000000000000000000000..227a63c68f86d71681fe1dcc0e3da6fcdce94cb5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cosh.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/cosh_ops.h>
+
+namespace at {
+
+
+// aten::cosh(Tensor self) -> Tensor
+inline at::Tensor cosh(const at::Tensor & self) {
+    return at::_ops::cosh::call(self);
+}
+
+// aten::cosh_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & cosh_(at::Tensor & self) {
+    return at::_ops::cosh_::call(self);
+}
+
+// aten::cosh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cosh_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::cosh_out::call(self, out);
+}
+// aten::cosh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cosh_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::cosh_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cosh_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cosh_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8711d99791445f2daa6db706570daf4e6844b012
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cosh_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 cosh(const at::Tensor & self);
+TORCH_API at::Tensor & cosh_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & cosh_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & cosh_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cosh_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cosh_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..56631d7dbabd5bfed2864749ce1fbc6da2e7ae3a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cosh_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 cosh {
+  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::cosh";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "cosh(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 cosh_ {
+  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::cosh_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "cosh_(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 cosh_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::cosh";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "cosh.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/cosine_similarity.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cosine_similarity.h
new file mode 100644
index 0000000000000000000000000000000000000000..e30450bde1fed046908ae6d6469ee99a7c6dc362
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cosine_similarity.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_similarity_ops.h>
+
+namespace at {
+
+
+// aten::cosine_similarity(Tensor x1, Tensor x2, int dim=1, float eps=1e-08) -> Tensor
+inline at::Tensor cosine_similarity(const at::Tensor & x1, const at::Tensor & x2, int64_t dim=1, double eps=1e-08) {
+    return at::_ops::cosine_similarity::call(x1, x2, dim, eps);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/count_nonzero_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/count_nonzero_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..89f24ac04b8da1b1752d8631d98c6ba045ad2fd2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/count_nonzero_compositeexplicitautograd_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 compositeexplicitautograd {
+
+TORCH_API at::Tensor & count_nonzero_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim);
+TORCH_API at::Tensor & count_nonzero_outf(const at::Tensor & self, at::IntArrayRef dim, at::Tensor & out);
+TORCH_API at::Tensor count_nonzero(const at::Tensor & self, ::std::optional<int64_t> dim=::std::nullopt);
+TORCH_API at::Tensor & count_nonzero_out(at::Tensor & out, const at::Tensor & self, ::std::optional<int64_t> dim=::std::nullopt);
+TORCH_API at::Tensor & count_nonzero_outf(const at::Tensor & self, ::std::optional<int64_t> dim, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cross_entropy_loss.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cross_entropy_loss.h
new file mode 100644
index 0000000000000000000000000000000000000000..578ccc605af8d1fdeb8ffd36474afe0f20fbf2a2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cross_entropy_loss.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/cross_entropy_loss_ops.h>
+
+namespace at {
+
+
+// aten::cross_entropy_loss(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100, float label_smoothing=0.0) -> Tensor
+inline at::Tensor cross_entropy_loss(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean, int64_t ignore_index=-100, double label_smoothing=0.0) {
+    return at::_ops::cross_entropy_loss::call(self, target, weight, reduction, ignore_index, label_smoothing);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor cross_entropy_loss(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean, int64_t ignore_index=-100, double label_smoothing=0.0) {
+    return at::_ops::cross_entropy_loss::call(self, target, weight, reduction, ignore_index, label_smoothing);
+  }
+}
+
+// aten::cross_entropy_loss(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100, float label_smoothing=0.0) -> Tensor
+inline at::Tensor cross_entropy_loss_symint(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean, c10::SymInt ignore_index=-100, double label_smoothing=0.0) {
+    return at::_ops::cross_entropy_loss::call(self, target, weight, reduction, ignore_index, label_smoothing);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor cross_entropy_loss(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean, c10::SymInt ignore_index=-100, double label_smoothing=0.0) {
+    return at::_ops::cross_entropy_loss::call(self, target, weight, reduction, ignore_index, label_smoothing);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/crow_indices_copy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/crow_indices_copy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e7eef9e550737d2ec979f2ab0c661187cc47e94d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/crow_indices_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 crow_indices_copy(const at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/ctc_loss_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/ctc_loss_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..10cabd7f7d93bc31c59c07e18a1c86b7d0a564a4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/ctc_loss_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 ctc_loss_IntList {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, int64_t, int64_t, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::ctc_loss";
+  static constexpr const char* overload_name = "IntList";
+  static constexpr const char* schema_str = "ctc_loss.IntList(Tensor log_probs, Tensor targets, int[] input_lengths, int[] target_lengths, int blank=0, int reduction=Mean, bool zero_infinity=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & log_probs, const at::Tensor & targets, at::IntArrayRef input_lengths, at::IntArrayRef target_lengths, int64_t blank, int64_t reduction, bool zero_infinity);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & log_probs, const at::Tensor & targets, at::IntArrayRef input_lengths, at::IntArrayRef target_lengths, int64_t blank, int64_t reduction, bool zero_infinity);
+};
+
+struct TORCH_API ctc_loss_Tensor {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, int64_t, int64_t, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::ctc_loss";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "ctc_loss.Tensor(Tensor log_probs, Tensor targets, Tensor input_lengths, Tensor target_lengths, int blank=0, int reduction=Mean, bool zero_infinity=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & log_probs, const at::Tensor & targets, const at::Tensor & input_lengths, const at::Tensor & target_lengths, int64_t blank, int64_t reduction, bool zero_infinity);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & log_probs, const at::Tensor & targets, const at::Tensor & input_lengths, const at::Tensor & target_lengths, int64_t blank, int64_t reduction, bool zero_infinity);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_is_acceptable_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_is_acceptable_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..19fc5443bf54c731166a5742bdff08a33249ff5b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_is_acceptable_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 cudnn_is_acceptable(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cummax_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cummax_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1aa5f0ca9344e05a1c5afa77b29d1941fa3bee6e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cummax_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 ::std::tuple<at::Tensor,at::Tensor> cummax(const at::Tensor & self, at::Dimname dim);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> cummax_out(at::Tensor & values, at::Tensor & indices, const at::Tensor & self, at::Dimname dim);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> cummax_outf(const at::Tensor & self, at::Dimname dim, at::Tensor & values, at::Tensor & indices);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cumprod_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cumprod_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8b5a4fe97d5188395ecafbd51738af7a8b09782d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cumprod_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 cumprod(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & cumprod_out(at::Tensor & out, const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & cumprod_outf(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+TORCH_API at::Tensor & cumprod_(at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum.h
new file mode 100644
index 0000000000000000000000000000000000000000..cbc2f10a3865b50095c7666127af6a61fa69d1cd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum.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/cumsum_ops.h>
+
+namespace at {
+
+
+// aten::cumsum(Tensor self, int dim, *, ScalarType? dtype=None) -> Tensor
+inline at::Tensor cumsum(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype=::std::nullopt) {
+    return at::_ops::cumsum::call(self, dim, dtype);
+}
+
+// aten::cumsum.out(Tensor self, int dim, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cumsum_out(at::Tensor & out, const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype=::std::nullopt) {
+    return at::_ops::cumsum_out::call(self, dim, dtype, out);
+}
+// aten::cumsum.out(Tensor self, int dim, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cumsum_outf(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype, at::Tensor & out) {
+    return at::_ops::cumsum_out::call(self, dim, dtype, out);
+}
+
+// aten::cumsum.dimname(Tensor self, Dimname dim, *, ScalarType? dtype=None) -> Tensor
+inline at::Tensor cumsum(const at::Tensor & self, at::Dimname dim, ::std::optional<at::ScalarType> dtype=::std::nullopt) {
+    return at::_ops::cumsum_dimname::call(self, dim, dtype);
+}
+
+// aten::cumsum.dimname_out(Tensor self, Dimname dim, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cumsum_out(at::Tensor & out, const at::Tensor & self, at::Dimname dim, ::std::optional<at::ScalarType> dtype=::std::nullopt) {
+    return at::_ops::cumsum_dimname_out::call(self, dim, dtype, out);
+}
+// aten::cumsum.dimname_out(Tensor self, Dimname dim, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cumsum_outf(const at::Tensor & self, at::Dimname dim, ::std::optional<at::ScalarType> dtype, at::Tensor & out) {
+    return at::_ops::cumsum_dimname_out::call(self, dim, dtype, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/data_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/data_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8dee6aef370181014cf43e5e303114cece2930d0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/data_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 data(const at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/dequantize_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/dequantize_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ebd880412244368efd83a11b39d9cbdc716ca652
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/dequantize_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 & dequantize_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & dequantize_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API void dequantize_out(at::TensorList out, at::TensorList tensors);
+TORCH_API void dequantize_outf(at::TensorList tensors, at::TensorList out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/dequantize_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/dequantize_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..4a30b5e5dc2295707ffd2c588a61a6b467789132
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/dequantize_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 dequantize_self {
+  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::dequantize";
+  static constexpr const char* overload_name = "self";
+  static constexpr const char* schema_str = "dequantize.self(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 dequantize_tensors {
+  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::dequantize";
+  static constexpr const char* overload_name = "tensors";
+  static constexpr const char* schema_str = "dequantize.tensors(Tensor[] tensors) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(at::TensorList tensors);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList tensors);
+};
+
+struct TORCH_API dequantize_self_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::dequantize";
+  static constexpr const char* overload_name = "self_out";
+  static constexpr const char* schema_str = "dequantize.self_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);
+};
+
+struct TORCH_API dequantize_tensors_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::dequantize";
+  static constexpr const char* overload_name = "tensors_out";
+  static constexpr const char* schema_str = "dequantize.tensors_out(Tensor[] tensors, *, Tensor(a!)[] out) -> ()";
+  static void call(at::TensorList tensors, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList tensors, at::TensorList out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/dsplit.h b/lib/python3.12/site-packages/torch/include/ATen/ops/dsplit.h
new file mode 100644
index 0000000000000000000000000000000000000000..b026baded694fa923552f859f99263d32bcf7721
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/dsplit.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/dsplit_ops.h>
+
+namespace at {
+
+
+// aten::dsplit.int(Tensor(a -> *) self, int sections) -> Tensor(a)[]
+inline ::std::vector<at::Tensor> dsplit(const at::Tensor & self, int64_t sections) {
+    return at::_ops::dsplit_int::call(self, sections);
+}
+
+// aten::dsplit.array(Tensor(a -> *) self, int[] indices) -> Tensor(a)[]
+inline ::std::vector<at::Tensor> dsplit(const at::Tensor & self, at::IntArrayRef indices) {
+    return at::_ops::dsplit_array::call(self, indices);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/einsum_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/einsum_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..ade5e8c4e559cf87585e0b3a56261bafa7b66411
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/einsum_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 einsum(c10::string_view equation, at::TensorList tensors, at::OptionalIntArrayRef path=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/embedding.h b/lib/python3.12/site-packages/torch/include/ATen/ops/embedding.h
new file mode 100644
index 0000000000000000000000000000000000000000..e984644cd1d3313ed53e8ce99f6e8dcb60943156
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/embedding.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/embedding_ops.h>
+
+namespace at {
+
+
+// aten::embedding(Tensor weight, Tensor indices, SymInt padding_idx=-1, bool scale_grad_by_freq=False, bool sparse=False) -> Tensor
+inline at::Tensor embedding(const at::Tensor & weight, const at::Tensor & indices, int64_t padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false) {
+    return at::_ops::embedding::call(weight, indices, padding_idx, scale_grad_by_freq, sparse);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor embedding(const at::Tensor & weight, const at::Tensor & indices, int64_t padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false) {
+    return at::_ops::embedding::call(weight, indices, padding_idx, scale_grad_by_freq, sparse);
+  }
+}
+
+// aten::embedding(Tensor weight, Tensor indices, SymInt padding_idx=-1, bool scale_grad_by_freq=False, bool sparse=False) -> Tensor
+inline at::Tensor embedding_symint(const at::Tensor & weight, const at::Tensor & indices, c10::SymInt padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false) {
+    return at::_ops::embedding::call(weight, indices, padding_idx, scale_grad_by_freq, sparse);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor embedding(const at::Tensor & weight, const at::Tensor & indices, c10::SymInt padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false) {
+    return at::_ops::embedding::call(weight, indices, padding_idx, scale_grad_by_freq, sparse);
+  }
+}
+
+// aten::embedding.out(Tensor weight, Tensor indices, SymInt padding_idx=-1, bool scale_grad_by_freq=False, bool sparse=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & embedding_out(at::Tensor & out, const at::Tensor & weight, const at::Tensor & indices, int64_t padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false) {
+    return at::_ops::embedding_out::call(weight, indices, padding_idx, scale_grad_by_freq, sparse, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & embedding_out(at::Tensor & out, const at::Tensor & weight, const at::Tensor & indices, int64_t padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false) {
+    return at::_ops::embedding_out::call(weight, indices, padding_idx, scale_grad_by_freq, sparse, out);
+  }
+}
+
+// aten::embedding.out(Tensor weight, Tensor indices, SymInt padding_idx=-1, bool scale_grad_by_freq=False, bool sparse=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & embedding_outf(const at::Tensor & weight, const at::Tensor & indices, int64_t padding_idx, bool scale_grad_by_freq, bool sparse, at::Tensor & out) {
+    return at::_ops::embedding_out::call(weight, indices, padding_idx, scale_grad_by_freq, sparse, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & embedding_outf(const at::Tensor & weight, const at::Tensor & indices, int64_t padding_idx, bool scale_grad_by_freq, bool sparse, at::Tensor & out) {
+    return at::_ops::embedding_out::call(weight, indices, padding_idx, scale_grad_by_freq, sparse, out);
+  }
+}
+
+// aten::embedding.out(Tensor weight, Tensor indices, SymInt padding_idx=-1, bool scale_grad_by_freq=False, bool sparse=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & embedding_symint_out(at::Tensor & out, const at::Tensor & weight, const at::Tensor & indices, c10::SymInt padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false) {
+    return at::_ops::embedding_out::call(weight, indices, padding_idx, scale_grad_by_freq, sparse, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & embedding_out(at::Tensor & out, const at::Tensor & weight, const at::Tensor & indices, c10::SymInt padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false) {
+    return at::_ops::embedding_out::call(weight, indices, padding_idx, scale_grad_by_freq, sparse, out);
+  }
+}
+
+// aten::embedding.out(Tensor weight, Tensor indices, SymInt padding_idx=-1, bool scale_grad_by_freq=False, bool sparse=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & embedding_symint_outf(const at::Tensor & weight, const at::Tensor & indices, c10::SymInt padding_idx, bool scale_grad_by_freq, bool sparse, at::Tensor & out) {
+    return at::_ops::embedding_out::call(weight, indices, padding_idx, scale_grad_by_freq, sparse, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & embedding_outf(const at::Tensor & weight, const at::Tensor & indices, c10::SymInt padding_idx, bool scale_grad_by_freq, bool sparse, at::Tensor & out) {
+    return at::_ops::embedding_out::call(weight, indices, padding_idx, scale_grad_by_freq, sparse, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_bag.h b/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_bag.h
new file mode 100644
index 0000000000000000000000000000000000000000..d1b51e4c918f136d04fb075d66dacbaba5d11e23
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_bag.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/embedding_bag_ops.h>
+
+namespace at {
+
+
+// aten::embedding_bag(Tensor weight, Tensor indices, Tensor offsets, bool scale_grad_by_freq=False, int mode=0, bool sparse=False, Tensor? per_sample_weights=None, bool include_last_offset=False) -> (Tensor, Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> embedding_bag(const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq=false, int64_t mode=0, bool sparse=false, const ::std::optional<at::Tensor> & per_sample_weights={}, bool include_last_offset=false) {
+    return at::_ops::embedding_bag::call(weight, indices, offsets, scale_grad_by_freq, mode, sparse, per_sample_weights, include_last_offset);
+}
+
+// aten::embedding_bag.padding_idx(Tensor weight, Tensor indices, Tensor offsets, bool scale_grad_by_freq, int mode, bool sparse, Tensor? per_sample_weights, bool include_last_offset, int? padding_idx) -> (Tensor, Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> embedding_bag(const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, bool include_last_offset, ::std::optional<int64_t> padding_idx) {
+    return at::_ops::embedding_bag_padding_idx::call(weight, indices, offsets, scale_grad_by_freq, mode, sparse, per_sample_weights, include_last_offset, padding_idx);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/empty_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..acbefa80cfbfa8b5ce89b2903b58fb8ee82ac0f1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_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 empty(at::IntArrayRef size, ::std::optional<at::DimnameList> names, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API 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);
+TORCH_API at::Tensor & empty_out(at::Tensor & out, at::IntArrayRef size, ::std::optional<at::DimnameList> names, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor & empty_outf(at::IntArrayRef size, ::std::optional<at::DimnameList> names, ::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/empty_strided_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_strided_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..07046c9beede8b3f725d673a56a6131ed04641c7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_strided_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 empty_strided {
+  using schema = at::Tensor (c10::SymIntArrayRef, 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::empty_strided";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "empty_strided(SymInt[] size, SymInt[] stride, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(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);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+struct TORCH_API empty_strided_out {
+  using schema = at::Tensor & (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::empty_strided";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "empty_strided.out(SymInt[] size, SymInt[] stride, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymIntArrayRef size, c10::SymIntArrayRef stride, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/eq_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/eq_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a15f08555822b803f3b3400e9b0da39ea3247f43
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/eq_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 eq(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & eq_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & eq_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & eq_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor eq(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & eq_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & eq_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & eq_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/erf_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/erf_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2b0b85d47a375a6cbee206e61b1b45768b2ed9a1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/erf_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 erf(const at::Tensor & self);
+TORCH_API at::Tensor & erf_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & erf_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & erf_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/exp.h b/lib/python3.12/site-packages/torch/include/ATen/ops/exp.h
new file mode 100644
index 0000000000000000000000000000000000000000..afe799a2748094f63c0f33dc8b2258a2d0a71854
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/exp.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/exp_ops.h>
+
+namespace at {
+
+
+// aten::exp(Tensor self) -> Tensor
+inline at::Tensor exp(const at::Tensor & self) {
+    return at::_ops::exp::call(self);
+}
+
+// aten::exp_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & exp_(at::Tensor & self) {
+    return at::_ops::exp_::call(self);
+}
+
+// aten::exp.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & exp_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::exp_out::call(self, out);
+}
+// aten::exp.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & exp_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::exp_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/exp2_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/exp2_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5de55711b372a748f094271b769672b46a4500f2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/exp2_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 exp2(const at::Tensor & self);
+TORCH_API at::Tensor & exp2_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & exp2_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & exp2_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_pack_quantized_matrix_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_pack_quantized_matrix_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..59c0b33958f224a4254e38e4b94db8e831990539
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_pack_quantized_matrix_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 fbgemm_pack_quantized_matrix(const at::Tensor & input);
+TORCH_API at::Tensor fbgemm_pack_quantized_matrix(const at::Tensor & input, int64_t K, int64_t N);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/feature_alpha_dropout_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/feature_alpha_dropout_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..29cdd6392e00467bc08f0fac60061e91d980664b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/feature_alpha_dropout_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 feature_alpha_dropout(const at::Tensor & input, double p, bool train);
+TORCH_API at::Tensor & feature_alpha_dropout_(at::Tensor & self, double p, bool train);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_hfftn_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_hfftn_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1794b2acef3265270d3595563a4c30570974ef86
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_hfftn_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_hfftn {
+  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_hfftn";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "fft_hfftn(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_hfftn_out {
+  using schema = const at::Tensor & (const at::Tensor &, at::OptionalSymIntArrayRef, at::OptionalIntArrayRef, ::std::optional<c10::string_view>, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fft_hfftn";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "fft_hfftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static const at::Tensor & call(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, const at::Tensor & out);
+  static const at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, const at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft.h
new file mode 100644
index 0000000000000000000000000000000000000000..df441c5904204162eef3e9e6b3f9429eff7a45aa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft.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_ifft_ops.h>
+
+namespace at {
+
+
+// aten::fft_ifft(Tensor self, SymInt? n=None, int dim=-1, str? norm=None) -> Tensor
+inline 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) {
+    return at::_ops::fft_ifft::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_ifft(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_ifft::call(self, n.has_value() ? ::std::make_optional(c10::SymInt(*n)) : ::std::nullopt, dim, norm);
+  }
+}
+
+// aten::fft_ifft(Tensor self, SymInt? n=None, int dim=-1, str? norm=None) -> Tensor
+inline 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) {
+    return at::_ops::fft_ifft::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_ifft(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_ifft::call(self, n, dim, norm);
+  }
+}
+
+// aten::fft_ifft.out(Tensor self, SymInt? n=None, int dim=-1, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::fft_ifft_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_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) {
+    return at::_ops::fft_ifft_out::call(self, n.has_value() ? ::std::make_optional(c10::SymInt(*n)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_ifft.out(Tensor self, SymInt? n=None, int dim=-1, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::fft_ifft_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_ifft_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_ifft_out::call(self, n.has_value() ? ::std::make_optional(c10::SymInt(*n)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_ifft.out(Tensor self, SymInt? n=None, int dim=-1, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::fft_ifft_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_ifft_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_ifft_out::call(self, n, dim, norm, out);
+  }
+}
+
+// aten::fft_ifft.out(Tensor self, SymInt? n=None, int dim=-1, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::fft_ifft_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_ifft_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_ifft_out::call(self, n, dim, norm, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ihfft2.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ihfft2.h
new file mode 100644
index 0000000000000000000000000000000000000000..a6439af1e25edb2193b8186c7a89ae34e63526fb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ihfft2.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_ihfft2_ops.h>
+
+namespace at {
+
+
+// aten::fft_ihfft2(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None) -> Tensor
+inline at::Tensor fft_ihfft2(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ihfft2::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_ihfft2(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ihfft2::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm);
+  }
+}
+
+// aten::fft_ihfft2(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None) -> Tensor
+inline at::Tensor fft_ihfft2_symint(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ihfft2::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_ihfft2(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ihfft2::call(self, s, dim, norm);
+  }
+}
+
+// aten::fft_ihfft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline const at::Tensor & fft_ihfft2_out(const 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) {
+    return at::_ops::fft_ihfft2_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>>>
+  const at::Tensor & fft_ihfft2_out(const 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) {
+    return at::_ops::fft_ihfft2_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_ihfft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline const at::Tensor & fft_ihfft2_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, const at::Tensor & out) {
+    return at::_ops::fft_ihfft2_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>>>
+  const at::Tensor & fft_ihfft2_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, const at::Tensor & out) {
+    return at::_ops::fft_ihfft2_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_ihfft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline const at::Tensor & fft_ihfft2_symint_out(const 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) {
+    return at::_ops::fft_ihfft2_out::call(self, s, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  const at::Tensor & fft_ihfft2_out(const 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) {
+    return at::_ops::fft_ihfft2_out::call(self, s, dim, norm, out);
+  }
+}
+
+// aten::fft_ihfft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline const at::Tensor & fft_ihfft2_symint_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, const at::Tensor & out) {
+    return at::_ops::fft_ihfft2_out::call(self, s, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  const at::Tensor & fft_ihfft2_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, const at::Tensor & out) {
+    return at::_ops::fft_ihfft2_out::call(self, s, dim, norm, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fill_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fill_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6b8dd4e8a5906ce0be42b0b49045176c038b60ff
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fill_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 & fill_(at::Tensor & self, const at::Scalar & value);
+TORCH_API at::Tensor & fill_(at::Tensor & self, const at::Tensor & value);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/flatten_dense_tensors_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/flatten_dense_tensors_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3d805f35c5dce9d3b7ecea9fcef08ec5d3e75711
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/flatten_dense_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 at::Tensor flatten_dense_tensors(at::TensorList tensors);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/floor_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/floor_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..28541d74214abc405ba8655c6a0c6f16ac0b3df4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/floor_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 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 cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fmin_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fmin_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b7c1a54fe39fcfd062131299d3fc28a37372cd6a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fmin_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 fmin(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & fmin_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & fmin_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fmod_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fmod_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..43aeea62323226f94fed66a1271aaaf81ebd6137
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fmod_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/fmod_meta.h>
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor fmod(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & fmod_out(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & fmod_(at::Tensor & self, const at::Scalar & other);
+struct TORCH_API structured_fmod_out : public at::meta::structured_fmod_Tensor {
+void impl(const at::Tensor & self, const at::Tensor & other, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/frac.h b/lib/python3.12/site-packages/torch/include/ATen/ops/frac.h
new file mode 100644
index 0000000000000000000000000000000000000000..7cbec262d076a6770f63aac05c15a99c5f71a757
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/frac.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/frac_ops.h>
+
+namespace at {
+
+
+// aten::frac(Tensor self) -> Tensor
+inline at::Tensor frac(const at::Tensor & self) {
+    return at::_ops::frac::call(self);
+}
+
+// aten::frac_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & frac_(at::Tensor & self) {
+    return at::_ops::frac_::call(self);
+}
+
+// aten::frac.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & frac_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::frac_out::call(self, out);
+}
+// aten::frac.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & frac_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::frac_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/frexp_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/frexp_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7f23ed030ecac100ae8bc0be877ab6b85a91a3e6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/frexp_compositeexplicitautograd_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 compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> frexp(const at::Tensor & self);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/full_like_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/full_like_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..7fd33b7465a0e8d3c81210adbde21b5e43970208
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/full_like_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 full_like {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &, ::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::full_like";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "full_like(Tensor self, Scalar fill_value, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Scalar & fill_value, ::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, const at::Tensor & self, const at::Scalar & fill_value, ::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 full_like_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, ::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::full_like";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "full_like.out(Tensor self, Scalar fill_value, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Scalar & fill_value, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & fill_value, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/full_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/full_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..2e714be1dea653028f78c0bf47652c23b6d420e5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/full_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 full(at::IntArrayRef size, const at::Scalar & fill_value, ::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={});
+TORCH_API at::Tensor & full_names_out(at::IntArrayRef size, const at::Scalar & fill_value, ::std::optional<at::DimnameList> names, at::Tensor & out);
+TORCH_API at::Tensor full(at::IntArrayRef size, const at::Scalar & fill_value, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={});
+TORCH_API at::Tensor & full_out(at::IntArrayRef size, const at::Scalar & fill_value, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gather_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gather_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8b15c129ebe178ce5ed6e1a17b61e2b2aa75bada
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gather_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 gather(const at::Tensor & self, int64_t dim, const at::Tensor & index, bool sparse_grad=false);
+TORCH_API at::Tensor & gather_out(at::Tensor & out, const at::Tensor & self, int64_t dim, const at::Tensor & index, bool sparse_grad=false);
+TORCH_API at::Tensor & gather_outf(const at::Tensor & self, int64_t dim, const at::Tensor & index, bool sparse_grad, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/ge_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/ge_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8a943808f081eab3f6287e62a7de913874172403
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/ge_native.h
@@ -0,0 +1,31 @@
+#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/ge_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_ge_Scalar_out : public at::meta::structured_ge_Scalar {
+void impl(const at::Tensor & self, const at::Scalar & other, const at::Tensor & out);
+};
+TORCH_API at::Tensor ge_scalar_nested(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor ge_quantized_cpu(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & ge_out_quantized_cpu(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+struct TORCH_API structured_ge_Tensor_out : public at::meta::structured_ge_Tensor {
+void impl(const at::Tensor & self, const at::Tensor & other, const at::Tensor & out);
+};
+TORCH_API at::Tensor ge_quantized_cpu(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & ge_out_quantized_cpu(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/gelu_backward_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gelu_backward_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..82b9c8e32c4956c128bd8d36504c57e5fe4bae1b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gelu_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_gelu_backward : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & grad_output, const at::Tensor & self, c10::string_view approximate);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/glu.h b/lib/python3.12/site-packages/torch/include/ATen/ops/glu.h
new file mode 100644
index 0000000000000000000000000000000000000000..4889e7196b589ceb76471267e1de5d749ded0934
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/glu.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/glu_ops.h>
+
+namespace at {
+
+
+// aten::glu.out(Tensor self, int dim=-1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & glu_out(at::Tensor & out, const at::Tensor & self, int64_t dim=-1) {
+    return at::_ops::glu_out::call(self, dim, out);
+}
+// aten::glu.out(Tensor self, int dim=-1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & glu_outf(const at::Tensor & self, int64_t dim, at::Tensor & out) {
+    return at::_ops::glu_out::call(self, dim, out);
+}
+
+// aten::glu(Tensor self, int dim=-1) -> Tensor
+inline at::Tensor glu(const at::Tensor & self, int64_t dim=-1) {
+    return at::_ops::glu::call(self, dim);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/glu_jvp_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/glu_jvp_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ca9787cfdd0220309bfbe2d4ea679a6803440a94
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/glu_jvp_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 glu_jvp(const at::Tensor & glu, const at::Tensor & x, const at::Tensor & dx, int64_t dim);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_2d_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_2d_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d0a004abcf0d0559eea81053433c3dd147578455
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_2d_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 grid_sampler_2d(const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_2d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_2d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..94f4da9f0341faeb59e080eaf6cc4a2ae8169548
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_2d_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 grid_sampler_2d {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, int64_t, int64_t, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::grid_sampler_2d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "grid_sampler_2d(Tensor input, Tensor grid, int interpolation_mode, int padding_mode, bool align_corners) -> Tensor";
+  static at::Tensor call(const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners);
+};
+
+struct TORCH_API grid_sampler_2d_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, int64_t, int64_t, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::grid_sampler_2d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "grid_sampler_2d.out(Tensor input, Tensor grid, int interpolation_mode, int padding_mode, bool align_corners, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gt.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gt.h
new file mode 100644
index 0000000000000000000000000000000000000000..42e004c8d0409f1245f341c9c871b4ebae0c478a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gt.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/gt_ops.h>
+
+namespace at {
+
+
+// aten::gt.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & gt_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::gt_Scalar_out::call(self, other, out);
+}
+// aten::gt.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & gt_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out) {
+    return at::_ops::gt_Scalar_out::call(self, other, out);
+}
+
+// aten::gt.Scalar(Tensor self, Scalar other) -> Tensor
+inline at::Tensor gt(const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::gt_Scalar::call(self, other);
+}
+
+// aten::gt.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & gt_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::gt_Tensor_out::call(self, other, out);
+}
+// aten::gt.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & gt_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::gt_Tensor_out::call(self, other, out);
+}
+
+// aten::gt.Tensor(Tensor self, Tensor other) -> Tensor
+inline at::Tensor gt(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::gt_Tensor::call(self, other);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..dd66907d167f44ea1372dbbc8f5c395b38aa86db
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_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/hardshrink_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_hardshrink_out : public at::meta::structured_hardshrink {
+void impl(const at::Tensor & self, const at::Scalar & lambd, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_backward_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_backward_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4de83514317b1f429fce7c809a5db850a1f58859
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_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 hardsigmoid_backward(const at::Tensor & grad_output, const at::Tensor & self);
+TORCH_API at::Tensor & hardsigmoid_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self);
+TORCH_API at::Tensor & hardsigmoid_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::Tensor & grad_input);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..387902a87bbb0fc83e2e0eee66d98c9c3d7b11da
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_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/hardsigmoid_backward_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_hardsigmoid_backward_out : public at::meta::structured_hardsigmoid_backward {
+void impl(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & grad_input);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardswish_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardswish_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0f29b0293645726c227b2264e4eaf866a1e179a4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardswish_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 hardswish(const at::Tensor & self);
+TORCH_API at::Tensor & hardswish_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & hardswish_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & hardswish_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardswish_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardswish_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..32efe4cd7cf26bb2768bf5fa978bd650c24302f9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardswish_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 hardswish(const at::Tensor & self);
+TORCH_API at::Tensor & hardswish_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & hardswish_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/heaviside_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/heaviside_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a94a49152aeb72f8283411169d401bad5db46b5e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/heaviside_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 heaviside_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::heaviside";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "heaviside.out(Tensor self, Tensor values, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & values, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & values, at::Tensor & out);
+};
+
+struct TORCH_API heaviside {
+  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::heaviside";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "heaviside(Tensor self, Tensor values) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & values);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & values);
+};
+
+struct TORCH_API heaviside_ {
+  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::heaviside_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "heaviside_(Tensor(a!) self, Tensor values) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & values);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & values);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/histogram.h b/lib/python3.12/site-packages/torch/include/ATen/ops/histogram.h
new file mode 100644
index 0000000000000000000000000000000000000000..ca6bafcb2ab6d08a6e38fc8e666094ddd0d3d776
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/histogram.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/histogram_ops.h>
+
+namespace at {
+
+
+// aten::histogram.bins_tensor_out(Tensor self, Tensor bins, *, Tensor? weight=None, bool density=False, Tensor(a!) hist, Tensor(b!) bin_edges) -> (Tensor(a!) hist, Tensor(b!) bin_edges)
+inline ::std::tuple<at::Tensor &,at::Tensor &> histogram_out(at::Tensor & hist, at::Tensor & bin_edges, const at::Tensor & self, const at::Tensor & bins, const ::std::optional<at::Tensor> & weight={}, bool density=false) {
+    return at::_ops::histogram_bins_tensor_out::call(self, bins, weight, density, hist, bin_edges);
+}
+// aten::histogram.bins_tensor_out(Tensor self, Tensor bins, *, Tensor? weight=None, bool density=False, Tensor(a!) hist, Tensor(b!) bin_edges) -> (Tensor(a!) hist, Tensor(b!) bin_edges)
+inline ::std::tuple<at::Tensor &,at::Tensor &> histogram_outf(const at::Tensor & self, const at::Tensor & bins, const ::std::optional<at::Tensor> & weight, bool density, at::Tensor & hist, at::Tensor & bin_edges) {
+    return at::_ops::histogram_bins_tensor_out::call(self, bins, weight, density, hist, bin_edges);
+}
+
+// aten::histogram.bins_tensor(Tensor self, Tensor bins, *, Tensor? weight=None, bool density=False) -> (Tensor hist, Tensor bin_edges)
+inline ::std::tuple<at::Tensor,at::Tensor> histogram(const at::Tensor & self, const at::Tensor & bins, const ::std::optional<at::Tensor> & weight={}, bool density=false) {
+    return at::_ops::histogram_bins_tensor::call(self, bins, weight, density);
+}
+
+// aten::histogram.bin_ct_out(Tensor self, int bins=100, *, float[]? range=None, Tensor? weight=None, bool density=False, Tensor(a!) hist, Tensor(b!) bin_edges) -> (Tensor(a!) hist, Tensor(b!) bin_edges)
+inline ::std::tuple<at::Tensor &,at::Tensor &> histogram_out(at::Tensor & hist, at::Tensor & bin_edges, const at::Tensor & self, int64_t bins=100, ::std::optional<at::ArrayRef<double>> range=::std::nullopt, const ::std::optional<at::Tensor> & weight={}, bool density=false) {
+    return at::_ops::histogram_bin_ct_out::call(self, bins, range, weight, density, hist, bin_edges);
+}
+// aten::histogram.bin_ct_out(Tensor self, int bins=100, *, float[]? range=None, Tensor? weight=None, bool density=False, Tensor(a!) hist, Tensor(b!) bin_edges) -> (Tensor(a!) hist, Tensor(b!) bin_edges)
+inline ::std::tuple<at::Tensor &,at::Tensor &> histogram_outf(const at::Tensor & self, int64_t bins, ::std::optional<at::ArrayRef<double>> range, const ::std::optional<at::Tensor> & weight, bool density, at::Tensor & hist, at::Tensor & bin_edges) {
+    return at::_ops::histogram_bin_ct_out::call(self, bins, range, weight, density, hist, bin_edges);
+}
+
+// aten::histogram.bin_ct(Tensor self, int bins=100, *, float[]? range=None, Tensor? weight=None, bool density=False) -> (Tensor hist, Tensor bin_edges)
+inline ::std::tuple<at::Tensor,at::Tensor> histogram(const at::Tensor & self, int64_t bins=100, ::std::optional<at::ArrayRef<double>> range=::std::nullopt, const ::std::optional<at::Tensor> & weight={}, bool density=false) {
+    return at::_ops::histogram_bin_ct::call(self, bins, range, weight, density);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/i0_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/i0_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..071a436948fc7d9220df45c3d8cb98f91349061f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/i0_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 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 cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/im2col_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/im2col_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..166d718e2b0bc5f8c2b08c5761f198a3a329f403
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/im2col_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 im2col_cpu(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride);
+TORCH_API at::Tensor & im2col_out_cpu(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride, at::Tensor & out);
+TORCH_API at::Tensor im2col_cuda(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride);
+TORCH_API at::Tensor & im2col_out_cuda(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/imag_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/imag_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3ed78d4b3036eda7306082c445c405e7b846461a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/imag_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 imag(const at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/imag_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/imag_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..7b78198656d5c7cc5bd5f0dad6212bea705e4849
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/imag_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 imag {
+  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::imag";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "imag(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/index_add_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_add_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d6f58f70ebba01bb01617ae3b57b5cbfbaa6b340
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_add_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/index_add_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_index_add_cpu_out : public at::meta::structured_index_add {
+void impl(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source, const at::Scalar & alpha, const at::Tensor & out);
+};
+struct TORCH_API structured_index_add_cuda_out : public at::meta::structured_index_add {
+void impl(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source, const at::Scalar & alpha, const at::Tensor & out);
+};
+TORCH_API at::Tensor index_add(const at::Tensor & self, at::Dimname dim, const at::Tensor & index, const at::Tensor & source, const at::Scalar & alpha=1);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_copy_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_copy_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6092003dd6d727ee6338ccd7f02646d9316a10a5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_copy_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 index_copy(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source);
+TORCH_API at::Tensor & index_copy_out(at::Tensor & out, const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source);
+TORCH_API at::Tensor & index_copy_outf(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source, at::Tensor & out);
+TORCH_API at::Tensor & index_copy_(at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_copy_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_copy_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f3c852ed8eb41ef9a3d24690b0d41a599430fa45
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_copy_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_copy(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source);
+TORCH_API at::Tensor & index_copy_out(at::Tensor & out, const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source);
+TORCH_API at::Tensor & index_copy_outf(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source, at::Tensor & out);
+TORCH_API at::Tensor & index_copy_(at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a2ae14581517ba37e337fa89f41500694d221020
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_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 index(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices);
+TORCH_API at::Tensor & index_out(at::Tensor & out, const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices);
+TORCH_API at::Tensor & index_outf(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..fb2f49de750bab6ded1f9d8b49f05a7fd34695c5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_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 index_Tensor {
+  using schema = at::Tensor (const at::Tensor &, const c10::List<::std::optional<at::Tensor>> &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::index";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "index.Tensor(Tensor self, Tensor?[] indices) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices);
+};
+
+struct TORCH_API index_Tensor_out {
+  using schema = at::Tensor & (const at::Tensor &, const c10::List<::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::index";
+  static constexpr const char* overload_name = "Tensor_out";
+  static constexpr const char* schema_str = "index.Tensor_out(Tensor self, Tensor?[] indices, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_put_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_put_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..eef12242d8e1deced44037fee9b977f59c37dff9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_put_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 index_put(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false);
+TORCH_API at::Tensor & index_put_out(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate, at::Tensor & out);
+TORCH_API at::Tensor & index_put_(at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/indices_copy_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/indices_copy_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..48edad1cfd1a26c5daf93d027ef4254b48a66f7b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/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 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::indices_copy";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "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 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::indices_copy";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "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/is_conj_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_conj_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5e568d603352ba758941614431cff4493f24fdae
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_conj_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 is_conj(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_distributed_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_distributed_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..611cd70d6eba2f39bb0ffba54c106bc249be00d1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_distributed_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_distributed {
+  using schema = bool (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::is_distributed";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "is_distributed(Tensor self) -> bool";
+  static bool call(const at::Tensor & self);
+  static bool redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_floating_point.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_floating_point.h
new file mode 100644
index 0000000000000000000000000000000000000000..4b65188bde642c37a0af6344a03db949cae62ee9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_floating_point.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_floating_point_ops.h>
+
+namespace at {
+
+
+// aten::is_floating_point(Tensor self) -> bool
+inline bool __dispatch_is_floating_point(const at::Tensor & self) {
+    return at::_ops::is_floating_point::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_set_to_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_set_to_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..2661c8794e0f42d384090a10816cac3a5cdc8351
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_set_to_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_set_to {
+  using schema = bool (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::is_set_to";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "is_set_to(Tensor self, Tensor tensor) -> bool";
+  static bool call(const at::Tensor & self, const at::Tensor & tensor);
+  static bool redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & tensor);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/isin.h b/lib/python3.12/site-packages/torch/include/ATen/ops/isin.h
new file mode 100644
index 0000000000000000000000000000000000000000..26ddb87828759176020ae4ac55c174abc844b85f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/isin.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/isin_ops.h>
+
+namespace at {
+
+
+// aten::isin.Tensor_Tensor_out(Tensor elements, Tensor test_elements, *, bool assume_unique=False, bool invert=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & isin_out(at::Tensor & out, const at::Tensor & elements, const at::Tensor & test_elements, bool assume_unique=false, bool invert=false) {
+    return at::_ops::isin_Tensor_Tensor_out::call(elements, test_elements, assume_unique, invert, out);
+}
+// aten::isin.Tensor_Tensor_out(Tensor elements, Tensor test_elements, *, bool assume_unique=False, bool invert=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & isin_outf(const at::Tensor & elements, const at::Tensor & test_elements, bool assume_unique, bool invert, at::Tensor & out) {
+    return at::_ops::isin_Tensor_Tensor_out::call(elements, test_elements, assume_unique, invert, out);
+}
+
+// aten::isin.Tensor_Tensor(Tensor elements, Tensor test_elements, *, bool assume_unique=False, bool invert=False) -> Tensor
+inline at::Tensor isin(const at::Tensor & elements, const at::Tensor & test_elements, bool assume_unique=false, bool invert=false) {
+    return at::_ops::isin_Tensor_Tensor::call(elements, test_elements, assume_unique, invert);
+}
+
+// aten::isin.Tensor_Scalar_out(Tensor elements, Scalar test_element, *, bool assume_unique=False, bool invert=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & isin_out(at::Tensor & out, const at::Tensor & elements, const at::Scalar & test_element, bool assume_unique=false, bool invert=false) {
+    return at::_ops::isin_Tensor_Scalar_out::call(elements, test_element, assume_unique, invert, out);
+}
+// aten::isin.Tensor_Scalar_out(Tensor elements, Scalar test_element, *, bool assume_unique=False, bool invert=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & isin_outf(const at::Tensor & elements, const at::Scalar & test_element, bool assume_unique, bool invert, at::Tensor & out) {
+    return at::_ops::isin_Tensor_Scalar_out::call(elements, test_element, assume_unique, invert, out);
+}
+
+// aten::isin.Tensor_Scalar(Tensor elements, Scalar test_element, *, bool assume_unique=False, bool invert=False) -> Tensor
+inline at::Tensor isin(const at::Tensor & elements, const at::Scalar & test_element, bool assume_unique=false, bool invert=false) {
+    return at::_ops::isin_Tensor_Scalar::call(elements, test_element, assume_unique, invert);
+}
+
+// aten::isin.Scalar_Tensor_out(Scalar element, Tensor test_elements, *, bool assume_unique=False, bool invert=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & isin_out(at::Tensor & out, const at::Scalar & element, const at::Tensor & test_elements, bool assume_unique=false, bool invert=false) {
+    return at::_ops::isin_Scalar_Tensor_out::call(element, test_elements, assume_unique, invert, out);
+}
+// aten::isin.Scalar_Tensor_out(Scalar element, Tensor test_elements, *, bool assume_unique=False, bool invert=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & isin_outf(const at::Scalar & element, const at::Tensor & test_elements, bool assume_unique, bool invert, at::Tensor & out) {
+    return at::_ops::isin_Scalar_Tensor_out::call(element, test_elements, assume_unique, invert, out);
+}
+
+// aten::isin.Scalar_Tensor(Scalar element, Tensor test_elements, *, bool assume_unique=False, bool invert=False) -> Tensor
+inline at::Tensor isin(const at::Scalar & element, const at::Tensor & test_elements, bool assume_unique=false, bool invert=false) {
+    return at::_ops::isin_Scalar_Tensor::call(element, test_elements, assume_unique, invert);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/isneginf_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/isneginf_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..62cb78b646300d90cd75928887187b1f0842a192
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/isneginf_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/isneginf_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_isneginf_out : public at::meta::structured_isneginf {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+TORCH_API at::Tensor NestedTensor_isneginf(const at::Tensor & self);
+TORCH_API at::Tensor isneginf_sparse(const at::Tensor & self);
+TORCH_API at::Tensor & isneginf_sparse_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor isneginf_sparse_csr(const at::Tensor & self);
+TORCH_API at::Tensor & isneginf_sparse_csr_out(const at::Tensor & self, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/isposinf.h b/lib/python3.12/site-packages/torch/include/ATen/ops/isposinf.h
new file mode 100644
index 0000000000000000000000000000000000000000..7d3cb943b2d02ed5fa81bed791f784e04fedf1c6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/isposinf.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/isposinf_ops.h>
+
+namespace at {
+
+
+// aten::isposinf(Tensor self) -> Tensor
+inline at::Tensor isposinf(const at::Tensor & self) {
+    return at::_ops::isposinf::call(self);
+}
+
+// aten::isposinf.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & isposinf_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::isposinf_out::call(self, out);
+}
+// aten::isposinf.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & isposinf_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::isposinf_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_backward_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_backward_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8bd257b124e91c289c6103716780a26926147928
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_backward_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 leaky_relu_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & negative_slope, bool self_is_result);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lift_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lift_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..d9f6e53ed57b6f995de96e871b10bfb85279ff8b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lift_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 lift {
+  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::lift";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "lift(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 lift_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::lift";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "lift.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/linalg_cholesky.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cholesky.h
new file mode 100644
index 0000000000000000000000000000000000000000..980a1214670c303be32d9b30bd7f52477f15e7e0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cholesky.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_cholesky_ops.h>
+
+namespace at {
+
+
+// aten::linalg_cholesky(Tensor self, *, bool upper=False) -> Tensor
+inline at::Tensor linalg_cholesky(const at::Tensor & self, bool upper=false) {
+    return at::_ops::linalg_cholesky::call(self, upper);
+}
+
+// aten::linalg_cholesky.out(Tensor self, *, bool upper=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_cholesky_out(at::Tensor & out, const at::Tensor & self, bool upper=false) {
+    return at::_ops::linalg_cholesky_out::call(self, upper, out);
+}
+// aten::linalg_cholesky.out(Tensor self, *, bool upper=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_cholesky_outf(const at::Tensor & self, bool upper, at::Tensor & out) {
+    return at::_ops::linalg_cholesky_out::call(self, upper, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_diagonal_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_diagonal_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..76359b305432ecaa7e027b8dad0ad15b0aa61194
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_diagonal_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 linalg_diagonal(const at::Tensor & A, int64_t offset=0, int64_t dim1=-2, int64_t dim2=-1);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_diagonal_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_diagonal_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..55bbfca2c44568f04ba169de740bb0342c72c76f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_diagonal_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 linalg_diagonal {
+  using schema = at::Tensor (const at::Tensor &, 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::linalg_diagonal";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_diagonal(Tensor(a) A, *, int offset=0, int dim1=-2, int dim2=-1) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & A, int64_t offset, int64_t dim1, int64_t dim2);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, int64_t offset, int64_t dim1, int64_t dim2);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_ex_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_ex_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9b74f9aec1da4fc6a5d5060dee1704b5bea050de
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_ex_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/linalg_ldl_factor_ex_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_linalg_ldl_factor_ex_out : public at::meta::structured_linalg_ldl_factor_ex {
+void impl(const at::Tensor & self, bool hermitian, bool check_errors, const at::Tensor & LD, const at::Tensor & pivots, const at::Tensor & info);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a8b36fd84df929f5cf94e6763695fea39c60ca88
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor_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_lu_factor {
+  using schema = ::std::tuple<at::Tensor,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::linalg_lu_factor";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_lu_factor(Tensor A, *, bool pivot=True) -> (Tensor LU, Tensor pivots)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & A, bool pivot);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, bool pivot);
+};
+
+struct TORCH_API linalg_lu_factor_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const at::Tensor &, bool, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_lu_factor";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "linalg_lu_factor.out(Tensor A, *, bool pivot=True, Tensor(a!) LU, Tensor(b!) pivots) -> (Tensor(a!) LU, Tensor(b!) pivots)";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & A, bool pivot, at::Tensor & LU, at::Tensor & pivots);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, bool pivot, at::Tensor & LU, at::Tensor & pivots);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..516271a86276590fbfe15c72a66593448eed1e07
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_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_linalg_lu : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & A, bool pivot);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..df3146b6818216ccb077ab9aa20c1d6b8851b1a5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_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_lu {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,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::linalg_lu";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_lu(Tensor A, *, bool pivot=True) -> (Tensor P, Tensor L, Tensor U)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & A, bool pivot);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, bool pivot);
+};
+
+struct TORCH_API linalg_lu_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> (const at::Tensor &, bool, 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_lu";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "linalg_lu.out(Tensor A, *, bool pivot=True, Tensor(a!) P, Tensor(b!) L, Tensor(c!) U) -> (Tensor(a!) P, Tensor(b!) L, Tensor(c!) U)";
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> call(const at::Tensor & A, bool pivot, at::Tensor & P, at::Tensor & L, at::Tensor & U);
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, bool pivot, at::Tensor & P, at::Tensor & L, at::Tensor & U);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_norm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_norm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5aff85b0b3ae90a9921569d298a4a182ce9b8cda
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_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 linalg_matrix_norm(const at::Tensor & self, const at::Scalar & ord, at::IntArrayRef dim={-2,-1}, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & linalg_matrix_norm_out(const at::Tensor & self, const at::Scalar & ord, at::IntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+TORCH_API at::Tensor linalg_matrix_norm(const at::Tensor & self, c10::string_view ord="fro", at::IntArrayRef dim={-2,-1}, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & linalg_matrix_norm_out(const at::Tensor & self, c10::string_view ord, at::IntArrayRef 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/linalg_pinv_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_pinv_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..21d98b09b6588eeb8fa824aa0680b889514efb3f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_pinv_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>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & linalg_pinv_out(const at::Tensor & self, const ::std::optional<at::Tensor> & atol, const ::std::optional<at::Tensor> & rtol, bool hermitian, at::Tensor & out);
+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);
+TORCH_API at::Tensor linalg_pinv(const at::Tensor & self, ::std::optional<double> atol=::std::nullopt, ::std::optional<double> rtol=::std::nullopt, bool hermitian=false);
+TORCH_API at::Tensor & linalg_pinv_out(const at::Tensor & self, ::std::optional<double> atol, ::std::optional<double> rtol, bool hermitian, at::Tensor & out);
+TORCH_API at::Tensor linalg_pinv(const at::Tensor & self, double rcond, bool hermitian=false);
+TORCH_API at::Tensor & linalg_pinv_out(const at::Tensor & self, double rcond, bool hermitian, at::Tensor & out);
+TORCH_API at::Tensor linalg_pinv(const at::Tensor & self, const at::Tensor & rcond, bool hermitian=false);
+TORCH_API at::Tensor & linalg_pinv_out(const at::Tensor & self, const at::Tensor & rcond, bool hermitian, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linear_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linear_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..62ca66fecf640a860ad77d55e620e3e0c784fb1b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linear_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 linear_backward {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, ::std::array<bool,3>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linear_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linear_backward(Tensor self, Tensor grad_output, Tensor weight, bool[3] output_mask) -> (Tensor, Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, ::std::array<bool,3> output_mask);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & grad_output, const at::Tensor & weight, ::std::array<bool,3> output_mask);
+};
+
+struct TORCH_API linear_backward_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> (const at::Tensor &, const at::Tensor &, const at::Tensor &, ::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::linear_backward";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "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!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> call(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);
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log1p.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log1p.h
new file mode 100644
index 0000000000000000000000000000000000000000..44313aba12e8bb2b184161f4b9be4e6da72ad678
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log1p.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/log1p_ops.h>
+
+namespace at {
+
+
+// aten::log1p(Tensor self) -> Tensor
+inline at::Tensor log1p(const at::Tensor & self) {
+    return at::_ops::log1p::call(self);
+}
+
+// aten::log1p_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & log1p_(at::Tensor & self) {
+    return at::_ops::log1p_::call(self);
+}
+
+// aten::log1p.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & log1p_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::log1p_out::call(self, out);
+}
+// aten::log1p.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & log1p_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::log1p_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log2_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log2_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f885a233fbed2c46b8d73c615b32dc327be5cae5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log2_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 log2(const at::Tensor & self);
+TORCH_API at::Tensor & log2_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & log2_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & log2_(at::Tensor & self);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_forward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_forward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a0b4bf022178e9e8b752e4f59a3b7991f6eedcd0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_forward_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_sigmoid_forward_output {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const 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::log_sigmoid_forward";
+  static constexpr const char* overload_name = "output";
+  static constexpr const char* schema_str = "log_sigmoid_forward.output(Tensor self, *, Tensor(a!) output, Tensor(b!) buffer) -> (Tensor(a!), Tensor(b!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & self, at::Tensor & output, at::Tensor & buffer);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & output, at::Tensor & buffer);
+};
+
+struct TORCH_API log_sigmoid_forward {
+  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::log_sigmoid_forward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "log_sigmoid_forward(Tensor self) -> (Tensor output, Tensor buffer)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self);
+  static ::std::tuple<at::Tensor,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/log_sigmoid_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..810a9ddd6a83c84d7790a05b1ef8732e030aafcd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_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 log_sigmoid(const at::Tensor & self);
+TORCH_API at::Tensor & log_sigmoid_out(const at::Tensor & self, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logical_and_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_and_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..a899a2f167d74ffb533cef846c0ec4e4ef1ba688
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_and_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 logical_and(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & logical_and_(at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & logical_and_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/logical_not_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_not_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e930571b7dc3a059f8834acbaa37da40558408f9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_not_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 & logical_not_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & logical_not_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logical_xor.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_xor.h
new file mode 100644
index 0000000000000000000000000000000000000000..2e53164fb8198d0947f0b664d489733c95d3da04
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_xor.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/logical_xor_ops.h>
+
+namespace at {
+
+
+// aten::logical_xor(Tensor self, Tensor other) -> Tensor
+inline at::Tensor logical_xor(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::logical_xor::call(self, other);
+}
+
+// aten::logical_xor.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & logical_xor_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::logical_xor_out::call(self, other, out);
+}
+// aten::logical_xor.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & logical_xor_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::logical_xor_out::call(self, other, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logical_xor_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_xor_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2d587f1a4a38ad074d5e84659e00e39d9db55b83
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_xor_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 logical_xor(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & logical_xor_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logit_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logit_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b50dd218f4db1041b7b54a59d9d66da6f9e6bf10
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logit_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 & logit_(at::Tensor & self, ::std::optional<double> eps=::std::nullopt);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logsumexp_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logsumexp_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..92f5012700bd51ea42f688b41154d118b7ee0fb7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logsumexp_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 logsumexp {
+  using schema = at::Tensor (const at::Tensor &, at::IntArrayRef, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logsumexp";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "logsumexp(Tensor self, int[1] dim, bool keepdim=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::IntArrayRef dim, bool keepdim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef dim, bool keepdim);
+};
+
+struct TORCH_API logsumexp_out {
+  using schema = at::Tensor & (const at::Tensor &, at::IntArrayRef, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logsumexp";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "logsumexp.out(Tensor self, int[1] dim, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef dim, bool keepdim, at::Tensor & out);
+};
+
+struct TORCH_API logsumexp_names {
+  using schema = at::Tensor (const at::Tensor &, at::DimnameList, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logsumexp";
+  static constexpr const char* overload_name = "names";
+  static constexpr const char* schema_str = "logsumexp.names(Tensor self, Dimname[1] dim, bool keepdim=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::DimnameList dim, bool keepdim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::DimnameList dim, bool keepdim);
+};
+
+struct TORCH_API logsumexp_names_out {
+  using schema = at::Tensor & (const at::Tensor &, at::DimnameList, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logsumexp";
+  static constexpr const char* overload_name = "names_out";
+  static constexpr const char* schema_str = "logsumexp.names_out(Tensor self, Dimname[1] dim, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::DimnameList dim, bool keepdim, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::DimnameList dim, bool keepdim, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/masked_scatter_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_scatter_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f25e6c8b83895439ed78c1a3ca3d978473c3f3fa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_scatter_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 masked_scatter_ {
+  using schema = at::Tensor & (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::masked_scatter_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "masked_scatter_(Tensor(a!) self, Tensor mask, Tensor source) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & mask, const at::Tensor & source);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & mask, const at::Tensor & source);
+};
+
+struct TORCH_API masked_scatter {
+  using schema = at::Tensor (const 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::masked_scatter";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "masked_scatter(Tensor self, Tensor mask, Tensor source) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & mask, const at::Tensor & source);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & mask, const at::Tensor & source);
+};
+
+struct TORCH_API masked_scatter_out {
+  using schema = at::Tensor & (const 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::masked_scatter";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "masked_scatter.out(Tensor self, Tensor mask, Tensor source, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & mask, const at::Tensor & source, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & mask, const at::Tensor & source, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select_backward_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select_backward_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6e09228250fa2131e4e1c6a3bbbb9ce072848553
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select_backward_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 masked_select_backward(const at::Tensor & grad, const at::Tensor & input, const at::Tensor & mask);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/matrix_exp.h b/lib/python3.12/site-packages/torch/include/ATen/ops/matrix_exp.h
new file mode 100644
index 0000000000000000000000000000000000000000..104996cde379bc140469a78576a53978be9488a1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/matrix_exp.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_ops.h>
+
+namespace at {
+
+
+// aten::matrix_exp(Tensor self) -> Tensor
+inline at::Tensor matrix_exp(const at::Tensor & self) {
+    return at::_ops::matrix_exp::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max.h
new file mode 100644
index 0000000000000000000000000000000000000000..9c8c01f738d89a5c0139f8688ff2a0bfa6bea6ac
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max.h
@@ -0,0 +1,81 @@
+#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/max_ops.h>
+
+namespace at {
+
+
+// aten::max.dim(Tensor self, int dim, bool keepdim=False) -> (Tensor values, Tensor indices)
+inline ::std::tuple<at::Tensor,at::Tensor> max(const at::Tensor & self, int64_t dim, bool keepdim=false) {
+    return at::_ops::max_dim::call(self, dim, keepdim);
+}
+
+// aten::max.dim_max(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) max, Tensor(b!) max_values) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::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) {
+    return at::_ops::max_dim_max::call(self, dim, keepdim, max, max_values);
+}
+// aten::max.dim_max(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) max, Tensor(b!) max_values) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::std::tuple<at::Tensor &,at::Tensor &> max_outf(const at::Tensor & self, int64_t dim, bool keepdim, at::Tensor & max, at::Tensor & max_values) {
+    return at::_ops::max_dim_max::call(self, dim, keepdim, max, max_values);
+}
+
+// aten::max.names_dim(Tensor self, Dimname dim, bool keepdim=False) -> (Tensor values, Tensor indices)
+inline ::std::tuple<at::Tensor,at::Tensor> max(const at::Tensor & self, at::Dimname dim, bool keepdim=false) {
+    return at::_ops::max_names_dim::call(self, dim, keepdim);
+}
+
+// aten::max.names_dim_max(Tensor self, Dimname dim, bool keepdim=False, *, Tensor(a!) max, Tensor(b!) max_values) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::std::tuple<at::Tensor &,at::Tensor &> max_out(at::Tensor & max, at::Tensor & max_values, const at::Tensor & self, at::Dimname dim, bool keepdim=false) {
+    return at::_ops::max_names_dim_max::call(self, dim, keepdim, max, max_values);
+}
+// aten::max.names_dim_max(Tensor self, Dimname dim, bool keepdim=False, *, Tensor(a!) max, Tensor(b!) max_values) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::std::tuple<at::Tensor &,at::Tensor &> max_outf(const at::Tensor & self, at::Dimname dim, bool keepdim, at::Tensor & max, at::Tensor & max_values) {
+    return at::_ops::max_names_dim_max::call(self, dim, keepdim, max, max_values);
+}
+
+// aten::max(Tensor self) -> Tensor
+inline at::Tensor max(const at::Tensor & self) {
+    return at::_ops::max::call(self);
+}
+
+// aten::max.other(Tensor self, Tensor other) -> Tensor
+inline at::Tensor max(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::max_other::call(self, other);
+}
+
+// aten::max.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & max_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::max_out::call(self, other, out);
+}
+// aten::max.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & max_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::max_out::call(self, other, out);
+}
+
+// aten::max.unary_out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & max_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::max_unary_out::call(self, out);
+}
+// aten::max.unary_out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & max_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::max_unary_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..8aacf14534ca83d0460e5fdb1c4bff929ed866fb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_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_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::max_pool3d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "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);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mean_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mean_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..229eca9154922bc64cc07676c35782f648ecc92e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mean_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 mean(const at::Tensor & self, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & mean_out(at::Tensor & out, const at::Tensor & self, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & mean_outf(const at::Tensor & self, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/minimum_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/minimum_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b87be6278a0c7e38199cd1e1f5f176a32c07517d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/minimum_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 minimum(const at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..1bb184b9331050b4d6b48b76aada66c58cfcdb59
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_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/miopen_rnn_backward_ops.h>
+
+namespace at {
+
+
+// aten::miopen_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, int hidden_size, int num_layers, bool batch_first, float dropout, bool train, bool bidirectional, int[] batch_sizes, Tensor? dropout_state, Tensor reserve, bool[4] output_mask) -> (Tensor, Tensor, Tensor, Tensor[])
+inline ::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) {
+    return at::_ops::miopen_rnn_backward::call(input, weight, weight_stride0, weight_buf, hx, cx, output, grad_output, grad_hy, grad_cy, mode, hidden_size, num_layers, batch_first, dropout, train, bidirectional, batch_sizes, dropout_state, reserve, output_mask);
+}
+
+// aten::miopen_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, int hidden_size, int num_layers, bool batch_first, float dropout, bool train, bool bidirectional, int[] batch_sizes, Tensor? dropout_state, Tensor reserve, bool[4] output_mask, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2, Tensor(d!)[] out3) -> ()
+inline void miopen_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 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) {
+    return at::_ops::miopen_rnn_backward_out::call(input, weight, weight_stride0, weight_buf, hx, cx, output, grad_output, grad_hy, grad_cy, mode, hidden_size, num_layers, batch_first, dropout, train, bidirectional, batch_sizes, dropout_state, reserve, output_mask, out0, out1, out2, out3);
+}
+// aten::miopen_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, int hidden_size, int num_layers, bool batch_first, float dropout, bool train, bool bidirectional, int[] batch_sizes, Tensor? dropout_state, Tensor reserve, bool[4] output_mask, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2, Tensor(d!)[] out3) -> ()
+inline void miopen_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 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) {
+    return at::_ops::miopen_rnn_backward_out::call(input, weight, weight_stride0, weight_buf, hx, cx, output, grad_output, grad_hy, grad_cy, mode, hidden_size, num_layers, batch_first, dropout, train, bidirectional, batch_sizes, dropout_state, reserve, output_mask, out0, out1, out2, out3);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_backward_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_backward_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..22fa857fb9329d902a932eb287d2046821dbf083
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_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 void miopen_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 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 miopen_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 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);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d.h
new file mode 100644
index 0000000000000000000000000000000000000000..18463073d31e9fc767e6a7378f73106ea80bbc72
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d.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/mkldnn_max_pool2d_ops.h>
+
+namespace at {
+
+
+// aten::mkldnn_max_pool2d(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False) -> Tensor
+inline at::Tensor mkldnn_max_pool2d(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, bool ceil_mode=false) {
+    return at::_ops::mkldnn_max_pool2d::call(self, kernel_size, stride, padding, dilation, ceil_mode);
+}
+
+// aten::mkldnn_max_pool2d.out(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & mkldnn_max_pool2d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, bool ceil_mode=false) {
+    return at::_ops::mkldnn_max_pool2d_out::call(self, kernel_size, stride, padding, dilation, ceil_mode, out);
+}
+// aten::mkldnn_max_pool2d.out(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & mkldnn_max_pool2d_outf(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out) {
+    return at::_ops::mkldnn_max_pool2d_out::call(self, kernel_size, stride, padding, dilation, ceil_mode, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool3d_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool3d_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b38b58ff23c0742acf3145df7ec4d90493c38f06
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool3d_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_max_pool3d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, bool ceil_mode=false);
+TORCH_API at::Tensor & mkldnn_max_pool3d_outf(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_reorder_conv2d_weight_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_reorder_conv2d_weight_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b193bae9116fe40a466b548d8343a369ecee74af
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_reorder_conv2d_weight_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 & mkldnn_reorder_conv2d_weight_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef padding=0, at::IntArrayRef stride=1, at::IntArrayRef dilation=1, int64_t groups=1, at::OptionalIntArrayRef input_size=::std::nullopt);
+TORCH_API at::Tensor & mkldnn_reorder_conv2d_weight_outf(const at::Tensor & self, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, at::OptionalIntArrayRef input_size, at::Tensor & out);
+TORCH_API at::Tensor & mkldnn_reorder_conv2d_weight_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef padding=c10::SymInt(0), c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef dilation=c10::SymInt(1), c10::SymInt groups=1, at::OptionalSymIntArrayRef input_size=::std::nullopt);
+TORCH_API at::Tensor & mkldnn_reorder_conv2d_weight_symint_outf(const at::Tensor & self, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, at::OptionalSymIntArrayRef input_size, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_reorder_conv2d_weight_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_reorder_conv2d_weight_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..cf7e916e70b20bb39d1de1faf91987c3d17f0464
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_reorder_conv2d_weight_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 & mkldnn_reorder_conv2d_weight_out_symint(const at::Tensor & self, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, at::OptionalSymIntArrayRef input_size, at::Tensor & out);
+TORCH_API at::Tensor mkldnn_reorder_conv2d_weight(const at::Tensor & self, at::IntArrayRef padding=0, at::IntArrayRef stride=1, at::IntArrayRef dilation=1, int64_t groups=1, at::OptionalIntArrayRef input_size=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mse_loss_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mse_loss_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1a32b18426139275f9e8c2779fb19954fda5b02a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mse_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 mse_loss(const at::Tensor & self, const at::Tensor & target, int64_t reduction=at::Reduction::Mean);
+TORCH_API at::Tensor & mse_loss_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & target, int64_t reduction=at::Reduction::Mean);
+TORCH_API at::Tensor & mse_loss_outf(const at::Tensor & self, const at::Tensor & target, int64_t reduction, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mul.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mul.h
new file mode 100644
index 0000000000000000000000000000000000000000..965baad9b7920c9af3c0ca04a14adc62658a6b7a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mul.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/mul_ops.h>
+
+namespace at {
+
+
+// aten::mul.Tensor(Tensor self, Tensor other) -> Tensor
+inline at::Tensor mul(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::mul_Tensor::call(self, other);
+}
+
+// aten::mul.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & mul_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::mul_out::call(self, other, out);
+}
+// aten::mul.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & mul_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::mul_out::call(self, other, out);
+}
+
+// aten::mul.Scalar(Tensor self, Scalar other) -> Tensor
+inline at::Tensor mul(const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::mul_Scalar::call(self, other);
+}
+
+// aten::mul.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & mul_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::mul_Scalar_out::call(self, other, out);
+}
+// aten::mul.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & mul_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out) {
+    return at::_ops::mul_Scalar_out::call(self, other, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mul_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mul_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d782ab7eb92a28508672316dfe4485a83af2be80
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mul_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 mul(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & mul_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & mul_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & mul_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_loss_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_loss_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f58d0304fef288a72b864ff879ddf02d3ea2fa8b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_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 multilabel_margin_loss_backward_grad_input {
+  using schema = at::Tensor & (const at::Tensor &, const 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::multilabel_margin_loss_backward";
+  static constexpr const char* overload_name = "grad_input";
+  static constexpr const char* schema_str = "multilabel_margin_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, int reduction, Tensor is_target, *, 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, const at::Tensor & is_target, 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, const at::Tensor & is_target, at::Tensor & grad_input);
+};
+
+struct TORCH_API multilabel_margin_loss_backward {
+  using schema = at::Tensor (const at::Tensor &, const 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::multilabel_margin_loss_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "multilabel_margin_loss_backward(Tensor grad_output, Tensor self, Tensor target, int reduction, Tensor is_target) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction, const at::Tensor & is_target);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction, const at::Tensor & is_target);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_loss_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_loss_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..02af87e0adeb1242c07e0451f8fef79b486719f6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_loss_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 multilabel_margin_loss_out {
+  using schema = 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::multilabel_margin_loss";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "multilabel_margin_loss.out(Tensor self, Tensor target, int reduction=Mean, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & target, int64_t reduction, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & target, int64_t reduction, at::Tensor & out);
+};
+
+struct TORCH_API multilabel_margin_loss {
+  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::multilabel_margin_loss";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "multilabel_margin_loss(Tensor self, Tensor target, int reduction=Mean) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & target, int64_t reduction);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, 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/multiply_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/multiply_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..069e973e28ac62d3860d7b88f33af93e82c4c69f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/multiply_ops.h
@@ -0,0 +1,72 @@
+#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 multiply_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::multiply";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "multiply.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 multiply__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::multiply_";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "multiply_.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 multiply_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::multiply";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "multiply.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 multiply_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::multiply";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "multiply.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 multiply__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::multiply_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "multiply_.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);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mvlgamma.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mvlgamma.h
new file mode 100644
index 0000000000000000000000000000000000000000..8ad579c2c2f70c2dcbd72133e8bdb1bb7ae51b5e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mvlgamma.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/mvlgamma_ops.h>
+
+namespace at {
+
+
+// aten::mvlgamma.out(Tensor self, int p, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & mvlgamma_out(at::Tensor & out, const at::Tensor & self, int64_t p) {
+    return at::_ops::mvlgamma_out::call(self, p, out);
+}
+// aten::mvlgamma.out(Tensor self, int p, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & mvlgamma_outf(const at::Tensor & self, int64_t p, at::Tensor & out) {
+    return at::_ops::mvlgamma_out::call(self, p, out);
+}
+
+// aten::mvlgamma(Tensor self, int p) -> Tensor
+inline at::Tensor mvlgamma(const at::Tensor & self, int64_t p) {
+    return at::_ops::mvlgamma::call(self, p);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mvlgamma_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mvlgamma_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..73663ade8bdfe91372ae7254398c5725c6017d59
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mvlgamma_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 mvlgamma(const at::Tensor & self, int64_t p);
+TORCH_API at::Tensor & mvlgamma_(at::Tensor & self, int64_t p);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_channel_shuffle_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_channel_shuffle_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..3aea3cb988d1d11d9eca512b000b9db5d140b9e7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_channel_shuffle_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 native_channel_shuffle {
+  using schema = at::Tensor (const at::Tensor &, c10::SymInt);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::native_channel_shuffle";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "native_channel_shuffle(Tensor self, SymInt groups) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, c10::SymInt groups);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymInt groups);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_dropout_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_dropout_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c075861ca7dc656d1692b92a44355e8b5ef8f1e1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_dropout_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_dropout(const at::Tensor & input, double p, ::std::optional<bool> train);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_group_norm_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_group_norm_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e602e49c5c9f5a8ab5fa7c8638503b3e79153b07
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_group_norm_backward_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 ::std::tuple<at::Tensor,at::Tensor,at::Tensor> native_group_norm_backward(const at::Tensor & grad_out, const at::Tensor & input, const at::Tensor & mean, const at::Tensor & rstd, const ::std::optional<at::Tensor> & weight, int64_t N, int64_t C, int64_t HxW, int64_t group, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> native_group_norm_backward_symint(const at::Tensor & grad_out, const at::Tensor & input, const at::Tensor & mean, const at::Tensor & rstd, const ::std::optional<at::Tensor> & weight, c10::SymInt N, c10::SymInt C, c10::SymInt HxW, int64_t group, ::std::array<bool,3> output_mask);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/ne_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/ne_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..81257f73d74f27bbe76f7367de7fc05fdc848742
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/ne_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 ne(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & ne_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & ne_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & ne_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor ne(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & ne_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & ne_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & ne_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/neg.h b/lib/python3.12/site-packages/torch/include/ATen/ops/neg.h
new file mode 100644
index 0000000000000000000000000000000000000000..0fa59dd0f6250499c1f3c4842ee8ee3cfd929948
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/neg.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/neg_ops.h>
+
+namespace at {
+
+
+// aten::neg(Tensor self) -> Tensor
+inline at::Tensor neg(const at::Tensor & self) {
+    return at::_ops::neg::call(self);
+}
+
+// aten::neg_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & neg_(at::Tensor & self) {
+    return at::_ops::neg_::call(self);
+}
+
+// aten::neg.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & neg_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::neg_out::call(self, out);
+}
+// aten::neg.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & neg_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::neg_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/neg_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/neg_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d23d9963d4dc909ba091a20a941581cc56b2c70e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/neg_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 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 cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/new_zeros_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/new_zeros_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8a61c380bd1646c1f8656dcd503398be68d0aba2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/new_zeros_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 new_zeros(const at::Tensor & self, 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 & new_zeros_out_symint(const at::Tensor & self, c10::SymIntArrayRef size, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss2d_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss2d_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..87156b90bee56f01ea1f9bb9ee6847e44f2d4b32
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss2d_backward_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 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);
+TORCH_API 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);
+TORCH_API 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);
+TORCH_API 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);
+TORCH_API 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);
+TORCH_API 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);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_nd_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_nd_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..94305d86947d442f7f0842db5596bff1730e3ed0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_nd_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 nll_loss_nd {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const ::std::optional<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::nll_loss_nd";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "nll_loss_nd(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_numpy_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_numpy_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0b1a2ff600d2752711597d3cfbea4fdb0c3b0493
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nonzero_numpy_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> nonzero_numpy(const at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nuclear_norm_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nuclear_norm_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f97a30b1b4f0e09f07f50c4f1991ede9ad24592b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nuclear_norm_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 nuclear_norm(const at::Tensor & self, bool keepdim=false);
+TORCH_API at::Tensor & nuclear_norm_out(at::Tensor & out, const at::Tensor & self, bool keepdim=false);
+TORCH_API at::Tensor & nuclear_norm_outf(const at::Tensor & self, bool keepdim, at::Tensor & out);
+TORCH_API at::Tensor nuclear_norm(const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false);
+TORCH_API at::Tensor & nuclear_norm_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false);
+TORCH_API at::Tensor & nuclear_norm_outf(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/one_hot_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/one_hot_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5be948361d64c577972d38d4146266963bb55b0a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/one_hot_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 one_hot(const at::Tensor & self, int64_t num_classes=-1);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/or.h b/lib/python3.12/site-packages/torch/include/ATen/ops/or.h
new file mode 100644
index 0000000000000000000000000000000000000000..877ce691462a165000ebc9732a237bc35379f8db
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/or.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/or_ops.h>
+
+namespace at {
+
+
+// aten::__or__.Scalar(Tensor self, Scalar other) -> Tensor
+inline at::Tensor __or__(const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::__or___Scalar::call(self, other);
+}
+
+// aten::__or__.Tensor(Tensor self, Tensor other) -> Tensor
+inline at::Tensor __or__(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::__or___Tensor::call(self, other);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/pad_sequence.h b/lib/python3.12/site-packages/torch/include/ATen/ops/pad_sequence.h
new file mode 100644
index 0000000000000000000000000000000000000000..019056794e53849c638cd4ce5fe8ed97dd3c4757
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/pad_sequence.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/pad_sequence_ops.h>
+
+namespace at {
+
+
+// aten::pad_sequence(Tensor[] sequences, bool batch_first=False, float padding_value=0.0, str padding_side="right") -> Tensor
+inline at::Tensor pad_sequence(at::TensorList sequences, bool batch_first=false, double padding_value=0.0, c10::string_view padding_side="right") {
+    return at::_ops::pad_sequence::call(sequences, batch_first, padding_value, padding_side);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/pdist_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/pdist_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..1beecea797276f1141f8295629802482477b9b8b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/pdist_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 pdist(const at::Tensor & self, double p=2);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/permute.h b/lib/python3.12/site-packages/torch/include/ATen/ops/permute.h
new file mode 100644
index 0000000000000000000000000000000000000000..7524a584c32eefa730e7a48f7b56d9c7ab00651a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/permute.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/permute_ops.h>
+
+namespace at {
+
+
+// aten::permute(Tensor(a) self, int[] dims) -> Tensor(a)
+inline at::Tensor permute(const at::Tensor & self, at::IntArrayRef dims) {
+    return at::_ops::permute::call(self, dims);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/pixel_shuffle_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/pixel_shuffle_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..c81615895e8204051ae934b250676f24535cc775
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/pixel_shuffle_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 pixel_shuffle {
+  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::pixel_shuffle";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "pixel_shuffle(Tensor self, int upscale_factor) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t upscale_factor);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t upscale_factor);
+};
+
+struct TORCH_API pixel_shuffle_out {
+  using schema = 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::pixel_shuffle";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "pixel_shuffle.out(Tensor self, int upscale_factor, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, int64_t upscale_factor, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t upscale_factor, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/pow_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/pow_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..208d8a14dbe5569521972e18e07de2d0d53a83f2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/pow_meta.h
@@ -0,0 +1,37 @@
+#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_pow_Tensor_Tensor : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Tensor & exponent);
+};
+struct TORCH_API structured_pow_Scalar : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Scalar & self, const at::Tensor & exponent);
+};
+struct TORCH_API structured_pow_Tensor_Scalar : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Scalar & exponent);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/pow_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/pow_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1818ca1baf04bbb7cd808d6d56530860899af869
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/pow_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 pow_Tensor_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::pow";
+  static constexpr const char* overload_name = "Tensor_Tensor_out";
+  static constexpr const char* schema_str = "pow.Tensor_Tensor_out(Tensor self, Tensor exponent, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & exponent, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & exponent, at::Tensor & out);
+};
+
+struct TORCH_API pow_Tensor_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::pow";
+  static constexpr const char* overload_name = "Tensor_Tensor";
+  static constexpr const char* schema_str = "pow.Tensor_Tensor(Tensor self, Tensor exponent) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & exponent);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & exponent);
+};
+
+struct TORCH_API pow_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::pow";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "pow.Scalar_out(Scalar self, Tensor exponent, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Scalar & self, const at::Tensor & exponent, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & self, const at::Tensor & exponent, at::Tensor & out);
+};
+
+struct TORCH_API pow_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::pow";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "pow.Scalar(Scalar self, Tensor exponent) -> Tensor";
+  static at::Tensor call(const at::Scalar & self, const at::Tensor & exponent);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & self, const at::Tensor & exponent);
+};
+
+struct TORCH_API pow_Tensor_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::pow";
+  static constexpr const char* overload_name = "Tensor_Scalar_out";
+  static constexpr const char* schema_str = "pow.Tensor_Scalar_out(Tensor self, Scalar exponent, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Scalar & exponent, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & exponent, at::Tensor & out);
+};
+
+struct TORCH_API pow_Tensor_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::pow";
+  static constexpr const char* overload_name = "Tensor_Scalar";
+  static constexpr const char* schema_str = "pow.Tensor_Scalar(Tensor self, Scalar exponent) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Scalar & exponent);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & exponent);
+};
+
+struct TORCH_API pow__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::pow_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "pow_.Scalar(Tensor(a!) self, Scalar exponent) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Scalar & exponent);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Scalar & exponent);
+};
+
+struct TORCH_API pow__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::pow_";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "pow_.Tensor(Tensor(a!) self, Tensor exponent) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & exponent);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & exponent);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/q_per_channel_scales_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/q_per_channel_scales_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..724bc5ae0937ae39a805b9d72f93c4aafb1fe5a8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/q_per_channel_scales_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 & q_per_channel_scales_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor q_per_channel_scales(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/q_per_channel_scales_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/q_per_channel_scales_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..5bca47c9d2873273d03065c1d47a55451d7f2cdb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/q_per_channel_scales_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 q_per_channel_scales {
+  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::q_per_channel_scales";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "q_per_channel_scales(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 q_per_channel_scales_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::q_per_channel_scales";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "q_per_channel_scales.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/q_per_channel_zero_points.h b/lib/python3.12/site-packages/torch/include/ATen/ops/q_per_channel_zero_points.h
new file mode 100644
index 0000000000000000000000000000000000000000..64889d65f5ca4c057632d9efa38a55862a3b6489
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/q_per_channel_zero_points.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/q_per_channel_zero_points_ops.h>
+
+namespace at {
+
+
+// aten::q_per_channel_zero_points(Tensor self) -> Tensor
+inline at::Tensor q_per_channel_zero_points(const at::Tensor & self) {
+    return at::_ops::q_per_channel_zero_points::call(self);
+}
+
+// aten::q_per_channel_zero_points.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & q_per_channel_zero_points_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::q_per_channel_zero_points_out::call(self, out);
+}
+// aten::q_per_channel_zero_points.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & q_per_channel_zero_points_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::q_per_channel_zero_points_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/q_scale_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/q_scale_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..f494443543046695b27af70d48722a679ffa6f51
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/q_scale_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 double q_scale_quant(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/qscheme_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/qscheme_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d77aaaf3c87785ed415f3921d6d9b09a62fb525d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/qscheme_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::QScheme qscheme_quant(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantile_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantile_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a87c195f8b41f420d4fe2a7f40694f17f8a585b4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantile_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 quantile {
+  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::quantile";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "quantile(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 quantile_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::quantile";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "quantile.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 quantile_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::quantile";
+  static constexpr const char* overload_name = "scalar";
+  static constexpr const char* schema_str = "quantile.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 quantile_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::quantile";
+  static constexpr const char* overload_name = "scalar_out";
+  static constexpr const char* schema_str = "quantile.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/quantize_per_channel.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_channel.h
new file mode 100644
index 0000000000000000000000000000000000000000..37a90691aeabf57f9da2fccbaa2beeb98c27575c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_channel.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/quantize_per_channel_ops.h>
+
+namespace at {
+
+
+// aten::quantize_per_channel(Tensor self, Tensor scales, Tensor zero_points, int axis, ScalarType dtype) -> Tensor
+inline at::Tensor quantize_per_channel(const at::Tensor & self, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, at::ScalarType dtype) {
+    return at::_ops::quantize_per_channel::call(self, scales, zero_points, axis, dtype);
+}
+
+// aten::quantize_per_channel.out(Tensor self, Tensor scales, Tensor zero_points, int axis, ScalarType dtype, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & quantize_per_channel_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, at::ScalarType dtype) {
+    return at::_ops::quantize_per_channel_out::call(self, scales, zero_points, axis, dtype, out);
+}
+// aten::quantize_per_channel.out(Tensor self, Tensor scales, Tensor zero_points, int axis, ScalarType dtype, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & quantize_per_channel_outf(const at::Tensor & self, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, at::ScalarType dtype, at::Tensor & out) {
+    return at::_ops::quantize_per_channel_out::call(self, scales, zero_points, axis, dtype, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_channel_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_channel_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a8c3b8a49af8b416c2d0f9aa74cee9fa3c725946
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_channel_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 quantize_per_channel {
+  using schema = at::Tensor (const 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::quantize_per_channel";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "quantize_per_channel(Tensor self, Tensor scales, Tensor zero_points, int axis, ScalarType dtype) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, at::ScalarType dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, at::ScalarType dtype);
+};
+
+struct TORCH_API quantize_per_channel_out {
+  using schema = at::Tensor & (const 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::quantize_per_channel";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "quantize_per_channel.out(Tensor self, Tensor scales, Tensor zero_points, int axis, ScalarType dtype, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, at::ScalarType dtype, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, at::ScalarType dtype, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor.h
new file mode 100644
index 0000000000000000000000000000000000000000..059ea6224926a94e74fa2e037025b005cc416c29
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor.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/quantize_per_tensor_ops.h>
+
+namespace at {
+
+
+// aten::quantize_per_tensor(Tensor self, float scale, int zero_point, ScalarType dtype) -> Tensor
+inline at::Tensor quantize_per_tensor(const at::Tensor & self, double scale, int64_t zero_point, at::ScalarType dtype) {
+    return at::_ops::quantize_per_tensor::call(self, scale, zero_point, dtype);
+}
+
+// aten::quantize_per_tensor.tensor_qparams(Tensor self, Tensor scale, Tensor zero_point, ScalarType dtype) -> Tensor
+inline at::Tensor quantize_per_tensor(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, at::ScalarType dtype) {
+    return at::_ops::quantize_per_tensor_tensor_qparams::call(self, scale, zero_point, dtype);
+}
+
+// aten::quantize_per_tensor.tensors(Tensor[] tensors, Tensor scales, Tensor zero_points, ScalarType dtype) -> Tensor[]
+inline ::std::vector<at::Tensor> quantize_per_tensor(at::TensorList tensors, const at::Tensor & scales, const at::Tensor & zero_points, at::ScalarType dtype) {
+    return at::_ops::quantize_per_tensor_tensors::call(tensors, scales, zero_points, dtype);
+}
+
+// aten::quantize_per_tensor.out(Tensor self, float scale, int zero_point, ScalarType dtype, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & quantize_per_tensor_out(at::Tensor & out, const at::Tensor & self, double scale, int64_t zero_point, at::ScalarType dtype) {
+    return at::_ops::quantize_per_tensor_out::call(self, scale, zero_point, dtype, out);
+}
+// aten::quantize_per_tensor.out(Tensor self, float scale, int zero_point, ScalarType dtype, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & quantize_per_tensor_outf(const at::Tensor & self, double scale, int64_t zero_point, at::ScalarType dtype, at::Tensor & out) {
+    return at::_ops::quantize_per_tensor_out::call(self, scale, zero_point, dtype, out);
+}
+
+// aten::quantize_per_tensor.tensor_qparams_out(Tensor self, Tensor scale, Tensor zero_point, ScalarType dtype, *, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::quantize_per_tensor_tensor_qparams_out::call(self, scale, zero_point, dtype, out);
+}
+// aten::quantize_per_tensor.tensor_qparams_out(Tensor self, Tensor scale, Tensor zero_point, ScalarType dtype, *, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::quantize_per_tensor_tensor_qparams_out::call(self, scale, zero_point, dtype, out);
+}
+
+// aten::quantize_per_tensor.tensors_out(Tensor[] tensors, Tensor scales, Tensor zero_points, ScalarType dtype, *, Tensor(a!)[] out) -> ()
+inline void quantize_per_tensor_out(at::TensorList out, at::TensorList tensors, const at::Tensor & scales, const at::Tensor & zero_points, at::ScalarType dtype) {
+    return at::_ops::quantize_per_tensor_tensors_out::call(tensors, scales, zero_points, dtype, out);
+}
+// aten::quantize_per_tensor.tensors_out(Tensor[] tensors, Tensor scales, Tensor zero_points, ScalarType dtype, *, Tensor(a!)[] out) -> ()
+inline void quantize_per_tensor_outf(at::TensorList tensors, const at::Tensor & scales, const at::Tensor & zero_points, at::ScalarType dtype, at::TensorList out) {
+    return at::_ops::quantize_per_tensor_tensors_out::call(tensors, scales, zero_points, dtype, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a49cea555f322ad11dbe5ac72db7a173d446f07e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor_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 quantize_per_tensor(const at::Tensor & self, double scale, int64_t zero_point, at::ScalarType dtype);
+TORCH_API at::Tensor quantize_per_tensor(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, at::ScalarType dtype);
+TORCH_API ::std::vector<at::Tensor> quantize_per_tensor(at::TensorList tensors, const at::Tensor & scales, const at::Tensor & zero_points, at::ScalarType dtype);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rand_like.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rand_like.h
new file mode 100644
index 0000000000000000000000000000000000000000..becd71d2640647d55a2f27ba6325df883c5383b5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rand_like.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/rand_like_ops.h>
+
+namespace at {
+
+
+// aten::rand_like(Tensor self, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor
+inline at::Tensor rand_like(const at::Tensor & self, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::rand_like::call(self, 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::rand_like(Tensor self, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor
+inline at::Tensor rand_like(const at::Tensor & self, ::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::rand_like::call(self, dtype, layout, device, pin_memory, memory_format);
+}
+
+// aten::rand_like.out(Tensor self, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & rand_like_out(at::Tensor & out, const at::Tensor & self, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::rand_like_out::call(self, memory_format, out);
+}
+// aten::rand_like.out(Tensor self, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & rand_like_outf(const at::Tensor & self, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out) {
+    return at::_ops::rand_like_out::call(self, memory_format, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rand_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rand_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a9c4b5eb360929bd884b037a521f8b61372795ed
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rand_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 rand_names {
+  using schema = at::Tensor (c10::SymIntArrayRef, ::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::rand";
+  static constexpr const char* overload_name = "names";
+  static constexpr const char* schema_str = "rand.names(SymInt[] size, *, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(c10::SymIntArrayRef 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, c10::SymIntArrayRef 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 rand_generator_with_names {
+  using schema = at::Tensor (c10::SymIntArrayRef, ::std::optional<at::Generator>, ::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::rand";
+  static constexpr const char* overload_name = "generator_with_names";
+  static constexpr const char* schema_str = "rand.generator_with_names(SymInt[] size, *, Generator? generator, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(c10::SymIntArrayRef size, ::std::optional<at::Generator> generator, ::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, c10::SymIntArrayRef size, ::std::optional<at::Generator> generator, ::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 rand {
+  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::rand";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "rand(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 rand_generator {
+  using schema = at::Tensor (c10::SymIntArrayRef, ::std::optional<at::Generator>, ::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::rand";
+  static constexpr const char* overload_name = "generator";
+  static constexpr const char* schema_str = "rand.generator(SymInt[] size, *, Generator? generator, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(c10::SymIntArrayRef size, ::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);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, ::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);
+};
+
+struct TORCH_API rand_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::rand";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "rand.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 rand_generator_out {
+  using schema = at::Tensor & (c10::SymIntArrayRef, ::std::optional<at::Generator>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::rand";
+  static constexpr const char* overload_name = "generator_out";
+  static constexpr const char* schema_str = "rand.generator_out(SymInt[] size, *, Generator? generator, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymIntArrayRef size, ::std::optional<at::Generator> generator, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, ::std::optional<at::Generator> generator, at::Tensor & out);
+};
+
+struct TORCH_API rand_names_out {
+  using schema = at::Tensor & (c10::SymIntArrayRef, ::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::rand";
+  static constexpr const char* overload_name = "names_out";
+  static constexpr const char* schema_str = "rand.names_out(SymInt[] size, *, Dimname[]? names, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymIntArrayRef size, ::std::optional<at::DimnameList> names, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, ::std::optional<at::DimnameList> names, at::Tensor & out);
+};
+
+struct TORCH_API rand_generator_with_names_out {
+  using schema = at::Tensor & (c10::SymIntArrayRef, ::std::optional<at::Generator>, ::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::rand";
+  static constexpr const char* overload_name = "generator_with_names_out";
+  static constexpr const char* schema_str = "rand.generator_with_names_out(SymInt[] size, *, Generator? generator, Dimname[]? names, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymIntArrayRef size, ::std::optional<at::Generator> generator, ::std::optional<at::DimnameList> names, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, ::std::optional<at::Generator> generator, ::std::optional<at::DimnameList> names, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/randperm_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/randperm_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..607e4571992f3dc4dac6ab40c52cce69d3e54a50
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/randperm_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 & randperm_out(at::Tensor & out, int64_t n, ::std::optional<at::Generator> generator);
+TORCH_API at::Tensor & randperm_outf(int64_t n, ::std::optional<at::Generator> generator, at::Tensor & out);
+TORCH_API at::Tensor & randperm_symint_out(at::Tensor & out, c10::SymInt n, ::std::optional<at::Generator> generator);
+TORCH_API at::Tensor & randperm_symint_outf(c10::SymInt n, ::std::optional<at::Generator> generator, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad1d_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad1d_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b79f15be13172a8f59ddd505115fa1cafdf49339
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad1d_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 reflection_pad1d(const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor reflection_pad1d_symint(const at::Tensor & self, c10::SymIntArrayRef padding);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/relu_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/relu_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6aea9684b15d4b02ed4b7883a0f0ad9a4e02124a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/relu_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 & relu_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & relu_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/renorm_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/renorm_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d1477d6cae160ee3f18740a8c08147fa055e729d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/renorm_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 renorm(const at::Tensor & self, const at::Scalar & p, int64_t dim, const at::Scalar & maxnorm);
+TORCH_API at::Tensor & renorm_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & p, int64_t dim, const at::Scalar & maxnorm);
+TORCH_API at::Tensor & renorm_outf(const at::Tensor & self, const at::Scalar & p, int64_t dim, const at::Scalar & maxnorm, at::Tensor & out);
+TORCH_API at::Tensor & renorm_(at::Tensor & self, const at::Scalar & p, int64_t dim, const at::Scalar & maxnorm);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_interleave_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_interleave_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3593c03901d2fefa7e732c0f74e9d49aa85005b9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_interleave_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 repeat_interleave(const at::Tensor & self, const at::Tensor & repeats, ::std::optional<int64_t> dim=::std::nullopt, ::std::optional<int64_t> output_size=::std::nullopt);
+TORCH_API at::Tensor repeat_interleave_symint(const at::Tensor & self, const at::Tensor & repeats, ::std::optional<int64_t> dim=::std::nullopt, ::std::optional<c10::SymInt> output_size=::std::nullopt);
+TORCH_API at::Tensor repeat_interleave(const at::Tensor & self, int64_t repeats, ::std::optional<int64_t> dim=::std::nullopt, ::std::optional<int64_t> output_size=::std::nullopt);
+TORCH_API at::Tensor repeat_interleave_symint(const at::Tensor & self, c10::SymInt repeats, ::std::optional<int64_t> dim=::std::nullopt, ::std::optional<c10::SymInt> output_size=::std::nullopt);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_interleave_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_interleave_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..851c4b87c6080476f1385eb04e63fe7575eaea43
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_interleave_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 repeat_interleave(const at::Tensor & repeats, ::std::optional<int64_t> output_size=::std::nullopt);
+TORCH_API at::Tensor repeat_interleave_symint(const at::Tensor & repeats, ::std::optional<c10::SymInt> output_size=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..3fa706538c859232d15d95d34ef35e33a4507e37
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_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 repeat(const at::Tensor & self, at::IntArrayRef repeats);
+TORCH_API at::Tensor & repeat_out_symint(const at::Tensor & self, c10::SymIntArrayRef repeats, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad1d_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad1d_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0fce1cb2483ef81ad59e4beb8e1a1575c3e640e5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad1d_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 replication_pad1d(const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor replication_pad1d_symint(const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & replication_pad1d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor & replication_pad1d_outf(const at::Tensor & self, at::IntArrayRef padding, at::Tensor & out);
+TORCH_API at::Tensor & replication_pad1d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & replication_pad1d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad2d_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad2d_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..1098a7f8c766db60ce5609bc5d6edd29ee28d264
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad2d_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/replication_pad2d_backward_ops.h>
+
+namespace at {
+
+
+// aten::replication_pad2d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & replication_pad2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, c10::fromIntArrayRefSlow(padding), grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & replication_pad2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, c10::fromIntArrayRefSlow(padding), grad_input);
+  }
+}
+
+// aten::replication_pad2d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & replication_pad2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding, at::Tensor & grad_input) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, c10::fromIntArrayRefSlow(padding), grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & replication_pad2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding, at::Tensor & grad_input) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, c10::fromIntArrayRefSlow(padding), grad_input);
+  }
+}
+
+// aten::replication_pad2d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & replication_pad2d_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, padding, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & replication_pad2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, padding, grad_input);
+  }
+}
+
+// aten::replication_pad2d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & replication_pad2d_backward_symint_outf(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & grad_input) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, padding, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & replication_pad2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & grad_input) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, padding, grad_input);
+  }
+}
+
+// aten::replication_pad2d_backward(Tensor grad_output, Tensor self, SymInt[4] padding) -> Tensor
+inline at::Tensor replication_pad2d_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward::call(grad_output, self, c10::fromIntArrayRefSlow(padding));
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor replication_pad2d_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward::call(grad_output, self, c10::fromIntArrayRefSlow(padding));
+  }
+}
+
+// aten::replication_pad2d_backward(Tensor grad_output, Tensor self, SymInt[4] padding) -> Tensor
+inline at::Tensor replication_pad2d_backward_symint(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward::call(grad_output, self, padding);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor replication_pad2d_backward(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward::call(grad_output, self, padding);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/reshape_as_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/reshape_as_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..67d22befeea9385580314a4ea11690db7e6ffe5d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/reshape_as_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 reshape_as(const at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/retain_grad_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/retain_grad_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..5cce81c83586a9ad3edb96ab23ce30179fb3cea7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/retain_grad_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 retain_grad {
+  using schema = void (at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::retain_grad";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "retain_grad(Tensor(a!) self) -> ()";
+  static void call(at::Tensor & self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rnn_tanh_cell_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rnn_tanh_cell_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9d79039778cb2d8a455fee50e999f0d3c9662d8a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rnn_tanh_cell_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 rnn_tanh_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={});
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/roll_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/roll_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0abe4414cfcb3550cbced2ed7b4ec232f8ff7219
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/roll_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 & roll_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef shifts, at::IntArrayRef dims={});
+TORCH_API at::Tensor & roll_outf(const at::Tensor & self, at::IntArrayRef shifts, at::IntArrayRef dims, at::Tensor & out);
+TORCH_API at::Tensor & roll_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef shifts, at::IntArrayRef dims={});
+TORCH_API at::Tensor & roll_symint_outf(const at::Tensor & self, c10::SymIntArrayRef shifts, at::IntArrayRef dims, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rot90.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rot90.h
new file mode 100644
index 0000000000000000000000000000000000000000..e7de710bbae91af71b3eec14481fcd172853cd66
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rot90.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/rot90_ops.h>
+
+namespace at {
+
+
+// aten::rot90(Tensor self, int k=1, int[] dims=[0,1]) -> Tensor
+inline at::Tensor rot90(const at::Tensor & self, int64_t k=1, at::IntArrayRef dims={0,1}) {
+    return at::_ops::rot90::call(self, k, dims);
+}
+
+// aten::rot90.out(Tensor self, int k=1, int[] dims=[0,1], *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & rot90_out(at::Tensor & out, const at::Tensor & self, int64_t k=1, at::IntArrayRef dims={0,1}) {
+    return at::_ops::rot90_out::call(self, k, dims, out);
+}
+// aten::rot90.out(Tensor self, int k=1, int[] dims=[0,1], *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & rot90_outf(const at::Tensor & self, int64_t k, at::IntArrayRef dims, at::Tensor & out) {
+    return at::_ops::rot90_out::call(self, k, dims, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/row_indices_copy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/row_indices_copy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..09f97ce215ac582e2c914ae9bf85f5752190d7dd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/row_indices_copy_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 & row_indices_copy_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor row_indices_copy(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5e79ef907833fe5860ba9b255170d7e0548ffaea
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_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 rrelu(const at::Tensor & self, const at::Scalar & lower=0.125, const at::Scalar & upper=0.3333333333333333, bool training=false, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor & rrelu_(at::Tensor & self, const at::Scalar & lower=0.125, const at::Scalar & upper=0.3333333333333333, bool training=false, ::std::optional<at::Generator> generator=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_with_noise_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_with_noise_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..67cbe34d3fe8afcfe9543c325a130a095433ebcb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_with_noise_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 & rrelu_with_noise_out(at::Tensor & out, const at::Tensor & self, at::Tensor & noise, const at::Scalar & lower=0.125, const at::Scalar & upper=0.3333333333333333, bool training=false, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor & rrelu_with_noise_outf(const at::Tensor & self, at::Tensor & noise, const at::Scalar & lower, const at::Scalar & upper, bool training, ::std::optional<at::Generator> generator, at::Tensor & out);
+TORCH_API at::Tensor & rrelu_with_noise_(at::Tensor & self, at::Tensor & noise, const at::Scalar & lower=0.125, const at::Scalar & upper=0.3333333333333333, bool training=false, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor rrelu_with_noise(const at::Tensor & self, at::Tensor & noise, const at::Scalar & lower=0.125, const at::Scalar & upper=0.3333333333333333, bool training=false, ::std::optional<at::Generator> generator=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rshift_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rshift_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..25d80bcc7baaf7d16f4854e219df4da41ec5a482
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rshift_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 & __rshift___out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & __rshift___outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & __rshift___out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & __rshift___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/rsub_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rsub_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1015fc80e978a306fe9e83c12ff4c2241ff46715
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rsub_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 rsub(const 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/scaled_dot_product_attention_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/scaled_dot_product_attention_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..488ea9e486d5fd00b5d8c01f3bb5b275ab1bd05b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/scaled_dot_product_attention_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 scaled_dot_product_attention(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_mask={}, double dropout_p=0.0, bool is_causal=false, ::std::optional<double> scale=::std::nullopt, bool enable_gqa=false);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/scatter.h b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter.h
new file mode 100644
index 0000000000000000000000000000000000000000..f41549f05382e4f44f0662aaf874e550cd9efb59
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter.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/scatter_ops.h>
+
+namespace at {
+
+
+// aten::scatter.src(Tensor self, int dim, Tensor index, Tensor src) -> Tensor
+inline at::Tensor scatter(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src) {
+    return at::_ops::scatter_src::call(self, dim, index, src);
+}
+
+// aten::scatter.src_out(Tensor self, int dim, Tensor index, Tensor src, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & scatter_out(at::Tensor & out, const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src) {
+    return at::_ops::scatter_src_out::call(self, dim, index, src, out);
+}
+// aten::scatter.src_out(Tensor self, int dim, Tensor index, Tensor src, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & scatter_outf(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, at::Tensor & out) {
+    return at::_ops::scatter_src_out::call(self, dim, index, src, out);
+}
+
+// aten::scatter.value(Tensor self, int dim, Tensor index, Scalar value) -> Tensor
+inline at::Tensor scatter(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Scalar & value) {
+    return at::_ops::scatter_value::call(self, dim, index, value);
+}
+
+// aten::scatter.value_out(Tensor self, int dim, Tensor index, Scalar value, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & scatter_out(at::Tensor & out, const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Scalar & value) {
+    return at::_ops::scatter_value_out::call(self, dim, index, value, out);
+}
+// aten::scatter.value_out(Tensor self, int dim, Tensor index, Scalar value, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & scatter_outf(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Scalar & value, at::Tensor & out) {
+    return at::_ops::scatter_value_out::call(self, dim, index, value, out);
+}
+
+// aten::scatter.reduce(Tensor self, int dim, Tensor index, Tensor src, *, str reduce) -> Tensor
+inline at::Tensor scatter(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, c10::string_view reduce) {
+    return at::_ops::scatter_reduce::call(self, dim, index, src, reduce);
+}
+
+// aten::scatter.reduce_out(Tensor self, int dim, Tensor index, Tensor src, *, str reduce, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & scatter_out(at::Tensor & out, const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, c10::string_view reduce) {
+    return at::_ops::scatter_reduce_out::call(self, dim, index, src, reduce, out);
+}
+// aten::scatter.reduce_out(Tensor self, int dim, Tensor index, Tensor src, *, str reduce, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & scatter_outf(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, c10::string_view reduce, at::Tensor & out) {
+    return at::_ops::scatter_reduce_out::call(self, dim, index, src, reduce, out);
+}
+
+// aten::scatter.value_reduce(Tensor self, int dim, Tensor index, Scalar value, *, str reduce) -> Tensor
+inline at::Tensor scatter(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Scalar & value, c10::string_view reduce) {
+    return at::_ops::scatter_value_reduce::call(self, dim, index, value, reduce);
+}
+
+// aten::scatter.value_reduce_out(Tensor self, int dim, Tensor index, Scalar value, *, str reduce, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & scatter_out(at::Tensor & out, const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Scalar & value, c10::string_view reduce) {
+    return at::_ops::scatter_value_reduce_out::call(self, dim, index, value, reduce, out);
+}
+// aten::scatter.value_reduce_out(Tensor self, int dim, Tensor index, Scalar value, *, str reduce, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & scatter_outf(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Scalar & value, c10::string_view reduce, at::Tensor & out) {
+    return at::_ops::scatter_value_reduce_out::call(self, dim, index, value, reduce, out);
+}
+
+// aten::scatter.dimname_src(Tensor self, Dimname dim, Tensor index, Tensor src) -> Tensor
+inline at::Tensor scatter(const at::Tensor & self, at::Dimname dim, const at::Tensor & index, const at::Tensor & src) {
+    return at::_ops::scatter_dimname_src::call(self, dim, index, src);
+}
+
+// aten::scatter.dimname_value(Tensor self, Dimname dim, Tensor index, Scalar value) -> Tensor
+inline at::Tensor scatter(const at::Tensor & self, at::Dimname dim, const at::Tensor & index, const at::Scalar & value) {
+    return at::_ops::scatter_dimname_value::call(self, dim, index, value);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_add.h b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_add.h
new file mode 100644
index 0000000000000000000000000000000000000000..6e6d620908d599c65d12a1f29b8d9bdc6058cb93
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_add.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/scatter_add_ops.h>
+
+namespace at {
+
+
+// aten::scatter_add(Tensor self, int dim, Tensor index, Tensor src) -> Tensor
+inline at::Tensor scatter_add(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src) {
+    return at::_ops::scatter_add::call(self, dim, index, src);
+}
+
+// aten::scatter_add.out(Tensor self, int dim, Tensor index, Tensor src, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & scatter_add_out(at::Tensor & out, const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src) {
+    return at::_ops::scatter_add_out::call(self, dim, index, src, out);
+}
+// aten::scatter_add.out(Tensor self, int dim, Tensor index, Tensor src, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & scatter_add_outf(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, at::Tensor & out) {
+    return at::_ops::scatter_add_out::call(self, dim, index, src, out);
+}
+
+// aten::scatter_add.dimname(Tensor self, Dimname dim, Tensor index, Tensor src) -> Tensor
+inline at::Tensor scatter_add(const at::Tensor & self, at::Dimname dim, const at::Tensor & index, const at::Tensor & src) {
+    return at::_ops::scatter_add_dimname::call(self, dim, index, src);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/searchsorted_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/searchsorted_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..84eba1a61b94610f2645f7f44256f04326d05762
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/searchsorted_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>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor searchsorted_cpu(const at::Tensor & sorted_sequence, const at::Tensor & self, bool out_int32=false, bool right=false, ::std::optional<c10::string_view> side=::std::nullopt, const ::std::optional<at::Tensor> & sorter={});
+TORCH_API at::Tensor & searchsorted_out_cpu(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);
+TORCH_API at::Tensor searchsorted_cuda(const at::Tensor & sorted_sequence, const at::Tensor & self, bool out_int32=false, bool right=false, ::std::optional<c10::string_view> side=::std::nullopt, const ::std::optional<at::Tensor> & sorter={});
+TORCH_API at::Tensor & searchsorted_out_cuda(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);
+TORCH_API at::Tensor searchsorted_cpu(const at::Tensor & sorted_sequence, const at::Scalar & self, bool out_int32=false, bool right=false, ::std::optional<c10::string_view> side=::std::nullopt, const ::std::optional<at::Tensor> & sorter={});
+TORCH_API at::Tensor & searchsorted_out_cpu(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);
+TORCH_API at::Tensor searchsorted_cuda(const at::Tensor & sorted_sequence, const at::Scalar & self, bool out_int32=false, bool right=false, ::std::optional<c10::string_view> side=::std::nullopt, const ::std::optional<at::Tensor> & sorter={});
+TORCH_API at::Tensor & searchsorted_out_cuda(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 native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/select_scatter.h b/lib/python3.12/site-packages/torch/include/ATen/ops/select_scatter.h
new file mode 100644
index 0000000000000000000000000000000000000000..a7fe447291507b941d056f993a40414abfe45bb1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/select_scatter.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/select_scatter_ops.h>
+
+namespace at {
+
+
+// aten::select_scatter(Tensor self, Tensor src, int dim, SymInt index) -> Tensor
+inline at::Tensor select_scatter(const at::Tensor & self, const at::Tensor & src, int64_t dim, int64_t index) {
+    return at::_ops::select_scatter::call(self, src, dim, index);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor select_scatter(const at::Tensor & self, const at::Tensor & src, int64_t dim, int64_t index) {
+    return at::_ops::select_scatter::call(self, src, dim, index);
+  }
+}
+
+// aten::select_scatter(Tensor self, Tensor src, int dim, SymInt index) -> Tensor
+inline at::Tensor select_scatter_symint(const at::Tensor & self, const at::Tensor & src, int64_t dim, c10::SymInt index) {
+    return at::_ops::select_scatter::call(self, src, dim, index);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor select_scatter(const at::Tensor & self, const at::Tensor & src, int64_t dim, c10::SymInt index) {
+    return at::_ops::select_scatter::call(self, src, dim, index);
+  }
+}
+
+// aten::select_scatter.out(Tensor self, Tensor src, int dim, SymInt index, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & select_scatter_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & src, int64_t dim, int64_t index) {
+    return at::_ops::select_scatter_out::call(self, src, dim, index, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & select_scatter_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & src, int64_t dim, int64_t index) {
+    return at::_ops::select_scatter_out::call(self, src, dim, index, out);
+  }
+}
+
+// aten::select_scatter.out(Tensor self, Tensor src, int dim, SymInt index, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & select_scatter_outf(const at::Tensor & self, const at::Tensor & src, int64_t dim, int64_t index, at::Tensor & out) {
+    return at::_ops::select_scatter_out::call(self, src, dim, index, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & select_scatter_outf(const at::Tensor & self, const at::Tensor & src, int64_t dim, int64_t index, at::Tensor & out) {
+    return at::_ops::select_scatter_out::call(self, src, dim, index, out);
+  }
+}
+
+// aten::select_scatter.out(Tensor self, Tensor src, int dim, SymInt index, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & select_scatter_symint_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & src, int64_t dim, c10::SymInt index) {
+    return at::_ops::select_scatter_out::call(self, src, dim, index, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & select_scatter_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & src, int64_t dim, c10::SymInt index) {
+    return at::_ops::select_scatter_out::call(self, src, dim, index, out);
+  }
+}
+
+// aten::select_scatter.out(Tensor self, Tensor src, int dim, SymInt index, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & select_scatter_symint_outf(const at::Tensor & self, const at::Tensor & src, int64_t dim, c10::SymInt index, at::Tensor & out) {
+    return at::_ops::select_scatter_out::call(self, src, dim, index, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & select_scatter_outf(const at::Tensor & self, const at::Tensor & src, int64_t dim, c10::SymInt index, at::Tensor & out) {
+    return at::_ops::select_scatter_out::call(self, src, dim, index, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/selu_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/selu_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..851fe594384dfe47782fe36d2a5fcf0fea9a0a4e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/selu_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 selu(const at::Tensor & self);
+TORCH_API at::Tensor & selu_(at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sgn_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sgn_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2e021d72923e3f2d6510dfd2c38f6bb3b66ab42f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sgn_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 sgn(const at::Tensor & self);
+TORCH_API at::Tensor & sgn_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & sgn_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & sgn_(at::Tensor & self);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid.h
new file mode 100644
index 0000000000000000000000000000000000000000..e61a77efe51d0252120a8656e3edd7f7ceaf821c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid.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/sigmoid_ops.h>
+
+namespace at {
+
+
+// aten::sigmoid(Tensor self) -> Tensor
+inline at::Tensor sigmoid(const at::Tensor & self) {
+    return at::_ops::sigmoid::call(self);
+}
+
+// aten::sigmoid_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & sigmoid_(at::Tensor & self) {
+    return at::_ops::sigmoid_::call(self);
+}
+
+// aten::sigmoid.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & sigmoid_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::sigmoid_out::call(self, out);
+}
+// aten::sigmoid.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & sigmoid_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::sigmoid_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2e614a6eb1dde781783a148f64d11fb159543b71
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_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 sigmoid(const at::Tensor & self);
+TORCH_API at::Tensor & sigmoid_(at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2e38e35a09c06cb5161e3bc04f22902d1eaa024c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_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 sigmoid(const at::Tensor & self);
+TORCH_API at::Tensor & sigmoid_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & sigmoid_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & sigmoid_(at::Tensor & self);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sin_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sin_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b38bdb546022fd0aa72dd89cd460c2d9fdf6fab5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sin_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 sin(const at::Tensor & self);
+TORCH_API at::Tensor & sin_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & sin_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & sin_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_transpose2d_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_transpose2d_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..411e0505aee45a1e3c306f70700c58e82a55e0c2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_transpose2d_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 slow_conv_transpose2d(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 output_padding=0, at::IntArrayRef dilation=1);
+TORCH_API at::Tensor slow_conv_transpose2d_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 output_padding=c10::SymInt(0), c10::SymIntArrayRef dilation=c10::SymInt(1));
+TORCH_API at::Tensor & slow_conv_transpose2d_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, at::IntArrayRef output_padding=0, at::IntArrayRef dilation=1);
+TORCH_API at::Tensor & slow_conv_transpose2d_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::IntArrayRef output_padding, at::IntArrayRef dilation, at::Tensor & out);
+TORCH_API at::Tensor & slow_conv_transpose2d_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), c10::SymIntArrayRef output_padding=c10::SymInt(0), c10::SymIntArrayRef dilation=c10::SymInt(1));
+TORCH_API at::Tensor & slow_conv_transpose2d_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, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef dilation, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/smm.h b/lib/python3.12/site-packages/torch/include/ATen/ops/smm.h
new file mode 100644
index 0000000000000000000000000000000000000000..da25c1b6aa154634c800a18d9d947d04155dd48f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/smm.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/smm_ops.h>
+
+namespace at {
+
+
+// aten::smm(Tensor self, Tensor mat2) -> Tensor
+inline at::Tensor smm(const at::Tensor & self, const at::Tensor & mat2) {
+    return at::_ops::smm::call(self, mat2);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/soft_margin_loss_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/soft_margin_loss_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..36d403e26e7c550965d0d7654984b7169a270e30
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/soft_margin_loss_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 soft_margin_loss_out {
+  using schema = 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::soft_margin_loss";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "soft_margin_loss.out(Tensor self, Tensor target, int reduction=Mean, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & target, int64_t reduction, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & target, int64_t reduction, at::Tensor & out);
+};
+
+struct TORCH_API soft_margin_loss {
+  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::soft_margin_loss";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "soft_margin_loss(Tensor self, Tensor target, int reduction=Mean) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & target, int64_t reduction);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, 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/softshrink_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/softshrink_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..c013e012e292294043b89b1104e34a76182a76c9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/softshrink_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 softshrink_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::softshrink";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "softshrink.out(Tensor self, Scalar lambd=0.5, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Scalar & lambd, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & lambd, at::Tensor & out);
+};
+
+struct TORCH_API softshrink {
+  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::softshrink";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "softshrink(Tensor self, Scalar lambd=0.5) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Scalar & lambd);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & lambd);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sort_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sort_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..21303b7e1abc95a649337eccf95814459ad30f84
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sort_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>
+#include <ATen/ops/sort_meta.h>
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> sort(const at::Tensor & self, int64_t dim=-1, bool descending=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> sort_out(const at::Tensor & self, int64_t dim, bool descending, at::Tensor & values, at::Tensor & indices);
+struct TORCH_API structured_sort_stable_out : public at::meta::structured_sort_stable {
+void impl(const at::Tensor & self, ::std::optional<bool> stable, int64_t dim, bool descending, const at::Tensor & values, const at::Tensor & indices);
+};
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> sort_quantized_cpu_stable(const at::Tensor & self, ::std::optional<bool> stable, int64_t dim=-1, bool descending=false);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> sort(const at::Tensor & self, at::Dimname dim, bool descending=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> sort_out(const at::Tensor & self, at::Dimname dim, bool descending, at::Tensor & values, at::Tensor & indices);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> sort(const at::Tensor & self, ::std::optional<bool> stable, at::Dimname dim, bool descending=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> sort_out(const at::Tensor & self, ::std::optional<bool> stable, at::Dimname dim, bool descending, at::Tensor & values, at::Tensor & indices);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_bsc_tensor.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_bsc_tensor.h
new file mode 100644
index 0000000000000000000000000000000000000000..3220ad0e2203cfabc9300a2341bf2fe97f954760
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_bsc_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_bsc_tensor_ops.h>
+
+namespace at {
+
+
+// aten::sparse_bsc_tensor.ccol_row_value_size(Tensor ccol_indices, Tensor row_indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_bsc_tensor(const at::Tensor & ccol_indices, const at::Tensor & row_indices, const at::Tensor & values, at::IntArrayRef size, at::TensorOptions options) {
+    return at::_ops::sparse_bsc_tensor_ccol_row_value_size::call(ccol_indices, row_indices, values, size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::sparse_bsc_tensor.ccol_row_value_size(Tensor ccol_indices, Tensor row_indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_bsc_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) {
+    return at::_ops::sparse_bsc_tensor_ccol_row_value_size::call(ccol_indices, row_indices, values, size, dtype, layout, device, pin_memory);
+}
+
+// aten::sparse_bsc_tensor.ccol_row_value(Tensor ccol_indices, Tensor row_indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_bsc_tensor(const at::Tensor & ccol_indices, const at::Tensor & row_indices, const at::Tensor & values, at::TensorOptions options) {
+    return at::_ops::sparse_bsc_tensor_ccol_row_value::call(ccol_indices, row_indices, values, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::sparse_bsc_tensor.ccol_row_value(Tensor ccol_indices, Tensor row_indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_bsc_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) {
+    return at::_ops::sparse_bsc_tensor_ccol_row_value::call(ccol_indices, row_indices, values, dtype, layout, device, pin_memory);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_compressed_tensor_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_compressed_tensor_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..048b22bcf2209b78e492713a62f4c23cb77fe92b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_compressed_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 sparse_compressed_tensor(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, at::IntArrayRef size, at::TensorOptions options);
+TORCH_API at::Tensor sparse_compressed_tensor(const at::Tensor & compressed_indices, const at::Tensor & plain_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_compressed_tensor_symint(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, c10::SymIntArrayRef size, at::TensorOptions options);
+TORCH_API at::Tensor sparse_compressed_tensor_symint(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, c10::SymIntArrayRef 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_compressed_tensor(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, at::TensorOptions options);
+TORCH_API at::Tensor sparse_compressed_tensor(const at::Tensor & compressed_indices, const at::Tensor & plain_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 compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csc_tensor.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csc_tensor.h
new file mode 100644
index 0000000000000000000000000000000000000000..1f0fd883ecf1203a4c396c2e5a83a6d7ac6fe138
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csc_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_csc_tensor_ops.h>
+
+namespace at {
+
+
+// aten::sparse_csc_tensor.ccol_row_value_size(Tensor ccol_indices, Tensor row_indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline 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) {
+    return at::_ops::sparse_csc_tensor_ccol_row_value_size::call(ccol_indices, row_indices, values, size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::sparse_csc_tensor.ccol_row_value_size(Tensor ccol_indices, Tensor row_indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline 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) {
+    return at::_ops::sparse_csc_tensor_ccol_row_value_size::call(ccol_indices, row_indices, values, size, dtype, layout, device, pin_memory);
+}
+
+// aten::sparse_csc_tensor.ccol_row_value(Tensor ccol_indices, Tensor row_indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_csc_tensor(const at::Tensor & ccol_indices, const at::Tensor & row_indices, const at::Tensor & values, at::TensorOptions options) {
+    return at::_ops::sparse_csc_tensor_ccol_row_value::call(ccol_indices, row_indices, values, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::sparse_csc_tensor.ccol_row_value(Tensor ccol_indices, Tensor row_indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline 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) {
+    return at::_ops::sparse_csc_tensor_ccol_row_value::call(ccol_indices, row_indices, values, dtype, layout, device, pin_memory);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_bessel_y0_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_bessel_y0_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..70f7758ce1882bc3cf8a46f328c0b796293dd086
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_bessel_y0_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_bessel_y0(const at::Tensor & self);
+TORCH_API at::Tensor & special_bessel_y0_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & special_bessel_y0_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_chebyshev_polynomial_t_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_t_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c380004c13e6ab26e01caea07ffd2214528af014
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_t_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_chebyshev_polynomial_t(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_chebyshev_polynomial_w_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_w_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..9b9a5201a03c908bf816a422ca5217b019b99bd9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_w_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_chebyshev_polynomial_w : 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_entr.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_entr.h
new file mode 100644
index 0000000000000000000000000000000000000000..0dd751608636ef4d745afd1b8e74ea86a5521b3b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_entr.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_entr_ops.h>
+
+namespace at {
+
+
+// aten::special_entr(Tensor self) -> Tensor
+inline at::Tensor special_entr(const at::Tensor & self) {
+    return at::_ops::special_entr::call(self);
+}
+
+// aten::special_entr.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_entr_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::special_entr_out::call(self, out);
+}
+// aten::special_entr.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_entr_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::special_entr_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_erfcx.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_erfcx.h
new file mode 100644
index 0000000000000000000000000000000000000000..c0c8636f4a314ede57d058f49852066c71667df6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_erfcx.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_erfcx_ops.h>
+
+namespace at {
+
+
+// aten::special_erfcx(Tensor self) -> Tensor
+inline at::Tensor special_erfcx(const at::Tensor & self) {
+    return at::_ops::special_erfcx::call(self);
+}
+
+// aten::special_erfcx.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_erfcx_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::special_erfcx_out::call(self, out);
+}
+// aten::special_erfcx.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_erfcx_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::special_erfcx_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_expm1_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_expm1_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..0434f01717e4b99e3d5ce9c6b45d6a9699e13f8c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_expm1_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 special_expm1(const at::Tensor & self);
+TORCH_API at::Tensor & special_expm1_out(const at::Tensor & self, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_hermite_polynomial_h_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_hermite_polynomial_h_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e5a3447069dea6ab0215b4c99a01267fcdd2ee03
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_hermite_polynomial_h_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 special_hermite_polynomial_h(const at::Tensor & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_hermite_polynomial_h_out(at::Tensor & out, const at::Tensor & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_hermite_polynomial_h_outf(const at::Tensor & x, const at::Tensor & n, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_hermite_polynomial_he_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_hermite_polynomial_he_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..638c1ffe8def6fc67131701c0af966fb5cdb13c5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_hermite_polynomial_he_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_hermite_polynomial_he_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_hermite_polynomial_he_out : public at::meta::structured_special_hermite_polynomial_he {
+void impl(const at::Tensor & x, const at::Tensor & n, const at::Tensor & out);
+};
+TORCH_API at::Tensor special_hermite_polynomial_he(const at::Scalar & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_hermite_polynomial_he_out(const at::Scalar & x, const at::Tensor & n, at::Tensor & out);
+TORCH_API at::Tensor special_hermite_polynomial_he(const at::Tensor & x, const at::Scalar & n);
+TORCH_API at::Tensor & special_hermite_polynomial_he_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_i0_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i0_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..01b87211ca503beba83e026f1764ba2685d14376
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i0_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_i0 {
+  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_i0";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_i0(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_i0_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_i0";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_i0.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_legendre_polynomial_p_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_legendre_polynomial_p_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..d79154b33b4e014f6c1466794af9a2715ee99bbc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_legendre_polynomial_p_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_legendre_polynomial_p {
+  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_legendre_polynomial_p";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_legendre_polynomial_p(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_legendre_polynomial_p_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_legendre_polynomial_p";
+  static constexpr const char* overload_name = "x_scalar";
+  static constexpr const char* schema_str = "special_legendre_polynomial_p.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_legendre_polynomial_p_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_legendre_polynomial_p";
+  static constexpr const char* overload_name = "n_scalar";
+  static constexpr const char* schema_str = "special_legendre_polynomial_p.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_legendre_polynomial_p_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_legendre_polynomial_p";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_legendre_polynomial_p.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_legendre_polynomial_p_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_legendre_polynomial_p";
+  static constexpr const char* overload_name = "x_scalar_out";
+  static constexpr const char* schema_str = "special_legendre_polynomial_p.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_legendre_polynomial_p_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_legendre_polynomial_p";
+  static constexpr const char* overload_name = "n_scalar_out";
+  static constexpr const char* schema_str = "special_legendre_polynomial_p.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_logit_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_logit_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..6e15171122decc0e5f2d792efe9972a159528d1c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_logit_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_logit {
+  using schema = at::Tensor (const at::Tensor &, ::std::optional<double>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_logit";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_logit(Tensor self, float? eps=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, ::std::optional<double> eps);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, ::std::optional<double> eps);
+};
+
+struct TORCH_API special_logit_out {
+  using schema = at::Tensor & (const at::Tensor &, ::std::optional<double>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_logit";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_logit.out(Tensor self, float? eps=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, ::std::optional<double> eps, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, ::std::optional<double> eps, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_i1_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_i1_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..27d72fb2608941f1fb059c368d4f8c043ea835cf
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_i1_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_modified_bessel_i1 : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_polygamma_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_polygamma_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..65c4be2fe71045779022f3a2de156c613af1612d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_polygamma_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_polygamma {
+  using schema = 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::special_polygamma";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_polygamma(int n, Tensor self) -> Tensor";
+  static at::Tensor call(int64_t n, const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t n, const at::Tensor & self);
+};
+
+struct TORCH_API special_polygamma_out {
+  using schema = 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::special_polygamma";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_polygamma.out(int n, Tensor self, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t n, const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t n, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_t_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_t_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..38112daf79c4d30cac50e799c1abd2f710175e5f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_t_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_shifted_chebyshev_polynomial_t(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_shifted_chebyshev_polynomial_t_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_t_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..ab56c3ef0a4b42cc6d17d81cf874a63a7dc9e341
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_t_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_t_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_shifted_chebyshev_polynomial_t_out : public at::meta::structured_special_shifted_chebyshev_polynomial_t {
+void impl(const at::Tensor & x, const at::Tensor & n, const at::Tensor & out);
+};
+TORCH_API at::Tensor special_shifted_chebyshev_polynomial_t(const at::Scalar & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_shifted_chebyshev_polynomial_t_out(const at::Scalar & x, const at::Tensor & n, at::Tensor & out);
+TORCH_API at::Tensor special_shifted_chebyshev_polynomial_t(const at::Tensor & x, const at::Scalar & n);
+TORCH_API at::Tensor & special_shifted_chebyshev_polynomial_t_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_shifted_chebyshev_polynomial_v_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_v_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8d0481f27706d95c46c4c90f3a310b1e065c9216
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_v_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_v_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_shifted_chebyshev_polynomial_v_out : public at::meta::structured_special_shifted_chebyshev_polynomial_v {
+void impl(const at::Tensor & x, const at::Tensor & n, const at::Tensor & out);
+};
+TORCH_API at::Tensor special_shifted_chebyshev_polynomial_v(const at::Scalar & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_shifted_chebyshev_polynomial_v_out(const at::Scalar & x, const at::Tensor & n, at::Tensor & out);
+TORCH_API at::Tensor special_shifted_chebyshev_polynomial_v(const at::Tensor & x, const at::Scalar & n);
+TORCH_API at::Tensor & special_shifted_chebyshev_polynomial_v_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_sinc_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_sinc_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..df1ab9b2bc6a1a55e1bc308134937c3ad2b6af88
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_sinc_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_sinc {
+  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_sinc";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_sinc(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_sinc_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_sinc";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_sinc.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_xlog1py.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlog1py.h
new file mode 100644
index 0000000000000000000000000000000000000000..96c3dd17edcd423fe4fdd6a77ad8db11489afd2d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlog1py.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/special_xlog1py_ops.h>
+
+namespace at {
+
+
+// aten::special_xlog1py(Tensor self, Tensor other) -> Tensor
+inline at::Tensor special_xlog1py(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::special_xlog1py::call(self, other);
+}
+
+// aten::special_xlog1py.self_scalar(Scalar self, Tensor other) -> Tensor
+inline at::Tensor special_xlog1py(const at::Scalar & self, const at::Tensor & other) {
+    return at::_ops::special_xlog1py_self_scalar::call(self, other);
+}
+
+// aten::special_xlog1py.other_scalar(Tensor self, Scalar other) -> Tensor
+inline at::Tensor special_xlog1py(const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::special_xlog1py_other_scalar::call(self, other);
+}
+
+// aten::special_xlog1py.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_xlog1py_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::special_xlog1py_out::call(self, other, out);
+}
+// aten::special_xlog1py.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_xlog1py_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::special_xlog1py_out::call(self, other, out);
+}
+
+// aten::special_xlog1py.self_scalar_out(Scalar self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_xlog1py_out(at::Tensor & out, const at::Scalar & self, const at::Tensor & other) {
+    return at::_ops::special_xlog1py_self_scalar_out::call(self, other, out);
+}
+// aten::special_xlog1py.self_scalar_out(Scalar self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_xlog1py_outf(const at::Scalar & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::special_xlog1py_self_scalar_out::call(self, other, out);
+}
+
+// aten::special_xlog1py.other_scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_xlog1py_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::special_xlog1py_other_scalar_out::call(self, other, out);
+}
+// aten::special_xlog1py.other_scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_xlog1py_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out) {
+    return at::_ops::special_xlog1py_other_scalar_out::call(self, other, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlogy_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlogy_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d81fe60e3167d8c37e9fd3d8e12945b7881fe3d8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlogy_compositeimplicitautograd_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 compositeimplicitautograd {
+
+TORCH_API at::Tensor special_xlogy(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & special_xlogy_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & special_xlogy_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor special_xlogy(const at::Scalar & self, const at::Tensor & other);
+TORCH_API at::Tensor & special_xlogy_out(at::Tensor & out, const at::Scalar & self, const at::Tensor & other);
+TORCH_API at::Tensor & special_xlogy_outf(const at::Scalar & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor special_xlogy(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & special_xlogy_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & special_xlogy_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlogy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlogy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..2e42fd18640399d9513ce1a6a958f4b9bffc21dc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlogy_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 special_xlogy(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & special_xlogy_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor special_xlogy(const at::Scalar & self, const at::Tensor & other);
+TORCH_API at::Tensor & special_xlogy_out(const at::Scalar & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor special_xlogy(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & special_xlogy_out(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/split_copy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/split_copy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3c5b8b5fa81b9aeddcc134e21adbe11fa11d2f8e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/split_copy_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 ::std::vector<at::Tensor> split_copy(const at::Tensor & self, int64_t split_size, int64_t dim=0);
+TORCH_API ::std::vector<at::Tensor> split_copy_symint(const at::Tensor & self, c10::SymInt split_size, int64_t dim=0);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze_copy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze_copy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8b7eb181051fc30d3e5e8745be9e1b718999e292
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze_copy_compositeexplicitautogradnonfunctional_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 compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor squeeze_copy(const at::Tensor & self);
+TORCH_API at::Tensor squeeze_copy(const at::Tensor & self, int64_t dim);
+TORCH_API at::Tensor squeeze_copy(const at::Tensor & self, at::IntArrayRef dim);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sub_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sub_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ff484b19f1bf1e3cea32192a2e4cad37162776ed
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sub_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 sub(const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & sub_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & sub_outf(const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha, at::Tensor & out);
+TORCH_API at::Tensor & sub_(at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha=1);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sum_to_size_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sum_to_size_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..02ba24eaef2fcb975b197c7e26ad4ca446addb62
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sum_to_size_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 sum_to_size {
+  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::sum_to_size";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "sum_to_size(Tensor self, SymInt[] size) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef size);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef size);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b9ca029d3f2ea1e530a33a5b5b14c2d77ce8c539
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_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 tanh_backward(const at::Tensor & grad_output, const at::Tensor & output);
+TORCH_API at::Tensor & tanh_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & output);
+TORCH_API at::Tensor & tanh_backward_outf(const at::Tensor & grad_output, const at::Tensor & output, at::Tensor & grad_input);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..67a8f5d3e2a1a54d502e235b67c5e5d82da7eda4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_native.h
@@ -0,0 +1,34 @@
+#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/tanh_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_tanh_out : public at::meta::structured_tanh {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+TORCH_API at::Tensor NestedTensor_tanh(const at::Tensor & self);
+TORCH_API at::Tensor & NestedTensor_tanh_(at::Tensor & self);
+TORCH_API at::Tensor tanh_sparse(const at::Tensor & self);
+TORCH_API at::Tensor & tanh_sparse_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & tanh_sparse_(at::Tensor & self);
+TORCH_API at::Tensor tanh_sparse_csr(const at::Tensor & self);
+TORCH_API at::Tensor & tanh_sparse_csr_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & tanh_sparse_csr_(at::Tensor & self);
+TORCH_API at::Tensor mkldnn_tanh(const at::Tensor & self);
+TORCH_API at::Tensor & mkldnn_tanh_(at::Tensor & self);
+TORCH_API at::Tensor tanh_quantized_cpu(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..866ee5a122e71937482d2244afbfe61532efa278
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_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/threshold_backward_ops.h>
+
+namespace at {
+
+
+// aten::threshold_backward.grad_input(Tensor grad_output, Tensor self, Scalar threshold, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & threshold_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold) {
+    return at::_ops::threshold_backward_grad_input::call(grad_output, self, threshold, grad_input);
+}
+// aten::threshold_backward.grad_input(Tensor grad_output, Tensor self, Scalar threshold, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & threshold_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold, at::Tensor & grad_input) {
+    return at::_ops::threshold_backward_grad_input::call(grad_output, self, threshold, grad_input);
+}
+
+// aten::threshold_backward(Tensor grad_output, Tensor self, Scalar threshold) -> Tensor
+inline at::Tensor threshold_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold) {
+    return at::_ops::threshold_backward::call(grad_output, self, threshold);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..73b0386d74677a2ac9e33e880a00de6fdd8d786d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_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 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 cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/tile.h b/lib/python3.12/site-packages/torch/include/ATen/ops/tile.h
new file mode 100644
index 0000000000000000000000000000000000000000..4ee6cc8f5d15c81ef7b0d643e59967ba74da1d1a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/tile.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/tile_ops.h>
+
+namespace at {
+
+
+// aten::tile(Tensor self, SymInt[] dims) -> Tensor
+inline at::Tensor tile(const at::Tensor & self, at::IntArrayRef dims) {
+    return at::_ops::tile::call(self, c10::fromIntArrayRefSlow(dims));
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor tile(const at::Tensor & self, at::IntArrayRef dims) {
+    return at::_ops::tile::call(self, c10::fromIntArrayRefSlow(dims));
+  }
+}
+
+// aten::tile(Tensor self, SymInt[] dims) -> Tensor
+inline at::Tensor tile_symint(const at::Tensor & self, c10::SymIntArrayRef dims) {
+    return at::_ops::tile::call(self, dims);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor tile(const at::Tensor & self, c10::SymIntArrayRef dims) {
+    return at::_ops::tile::call(self, dims);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/to_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/to_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..8022f194f6f1d49f523f5134ee61b5e2d8bdc12c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/to_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 to_dtype_layout {
+  using schema = at::Tensor (const at::Tensor &, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>, bool, bool, ::std::optional<at::MemoryFormat>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::to";
+  static constexpr const char* overload_name = "dtype_layout";
+  static constexpr const char* schema_str = "to.dtype_layout(Tensor(a) self, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, bool non_blocking=False, bool copy=False, MemoryFormat? memory_format=None) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, bool non_blocking, bool copy, ::std::optional<at::MemoryFormat> memory_format);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, bool non_blocking, bool copy, ::std::optional<at::MemoryFormat> memory_format);
+};
+
+struct TORCH_API to_device {
+  using schema = at::Tensor (const at::Tensor &, at::Device, at::ScalarType, bool, bool, ::std::optional<at::MemoryFormat>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::to";
+  static constexpr const char* overload_name = "device";
+  static constexpr const char* schema_str = "to.device(Tensor(a) self, Device device, ScalarType dtype, bool non_blocking=False, bool copy=False, MemoryFormat? memory_format=None) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, at::Device device, at::ScalarType dtype, bool non_blocking, bool copy, ::std::optional<at::MemoryFormat> memory_format);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Device device, at::ScalarType dtype, bool non_blocking, bool copy, ::std::optional<at::MemoryFormat> memory_format);
+};
+
+struct TORCH_API to_dtype {
+  using schema = at::Tensor (const at::Tensor &, at::ScalarType, bool, bool, ::std::optional<at::MemoryFormat>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::to";
+  static constexpr const char* overload_name = "dtype";
+  static constexpr const char* schema_str = "to.dtype(Tensor(a) self, ScalarType dtype, bool non_blocking=False, bool copy=False, MemoryFormat? memory_format=None) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, at::ScalarType dtype, bool non_blocking, bool copy, ::std::optional<at::MemoryFormat> memory_format);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::ScalarType dtype, bool non_blocking, bool copy, ::std::optional<at::MemoryFormat> memory_format);
+};
+
+struct TORCH_API to_other {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, bool, bool, ::std::optional<at::MemoryFormat>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::to";
+  static constexpr const char* overload_name = "other";
+  static constexpr const char* schema_str = "to.other(Tensor(a) self, Tensor other, bool non_blocking=False, bool copy=False, MemoryFormat? memory_format=None) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & other, bool non_blocking, bool copy, ::std::optional<at::MemoryFormat> memory_format);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, bool non_blocking, bool copy, ::std::optional<at::MemoryFormat> memory_format);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_bsc_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_bsc_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..dd3bb6e65b2e298a0f63bdbe56eab239485cbcc7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_bsc_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_bsc(const at::Tensor & self, at::IntArrayRef blocksize, ::std::optional<int64_t> dense_dim=::std::nullopt);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_csr.h b/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_csr.h
new file mode 100644
index 0000000000000000000000000000000000000000..f946a169b2b4cf105df93d6d6ac5c7648c07a402
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_csr.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_sparse_csr_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/trace_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/trace_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f8bf1881b7732ce8b991d71d615bc8aece997ff6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/trace_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 trace {
+  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::trace";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "trace(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 trace_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::trace";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "trace.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/trapezoid_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/trapezoid_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..b5f5a562b677f4aaa7197e4e5ff0ec754fb473ea
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/trapezoid_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 trapezoid_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::trapezoid";
+  static constexpr const char* overload_name = "x";
+  static constexpr const char* schema_str = "trapezoid.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 trapezoid_dx {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::trapezoid";
+  static constexpr const char* overload_name = "dx";
+  static constexpr const char* schema_str = "trapezoid.dx(Tensor y, *, Scalar dx=1, int dim=-1) -> Tensor";
+  static at::Tensor call(const at::Tensor & y, const at::Scalar & dx, int64_t dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & y, const at::Scalar & dx, int64_t dim);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/triplet_margin_loss.h b/lib/python3.12/site-packages/torch/include/ATen/ops/triplet_margin_loss.h
new file mode 100644
index 0000000000000000000000000000000000000000..3f0e42367a2f4cdf4d49cce535470a897102cff0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/triplet_margin_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/triplet_margin_loss_ops.h>
+
+namespace at {
+
+
+// aten::triplet_margin_loss(Tensor anchor, Tensor positive, Tensor negative, float margin=1.0, float p=2, float eps=1e-06, bool swap=False, int reduction=Mean) -> Tensor
+inline at::Tensor triplet_margin_loss(const at::Tensor & anchor, const at::Tensor & positive, const at::Tensor & negative, double margin=1.0, double p=2, double eps=1e-06, bool swap=false, int64_t reduction=at::Reduction::Mean) {
+    return at::_ops::triplet_margin_loss::call(anchor, positive, negative, margin, p, eps, swap, reduction);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/triu_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/triu_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3a1d23e298a1571c122ee3400c39e54d17f55167
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/triu_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 triu(const at::Tensor & self, int64_t diagonal=0);
+TORCH_API at::Tensor & triu_out(at::Tensor & out, const at::Tensor & self, int64_t diagonal=0);
+TORCH_API at::Tensor & triu_outf(const at::Tensor & self, int64_t diagonal, at::Tensor & out);
+TORCH_API at::Tensor & triu_(at::Tensor & self, int64_t diagonal=0);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/triu_indices_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/triu_indices_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..3e0ab2480c20f92cca01b0e84b3aa44b23a25a51
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/triu_indices_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 triu_indices {
+  using schema = at::Tensor (int64_t, int64_t, 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::triu_indices";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "triu_indices(int row, int col, int offset=0, *, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(int64_t row, int64_t col, int64_t offset, ::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 row, int64_t col, int64_t offset, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API triu_indices_out {
+  using schema = at::Tensor & (int64_t, int64_t, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::triu_indices";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "triu_indices.out(int row, int col, int offset=0, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t row, int64_t col, int64_t offset, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t row, int64_t col, int64_t offset, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/triu_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/triu_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..c0195c7af485a1226c4dc05e5408e066a7aae49b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/triu_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_triu : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, int64_t diagonal);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/true_divide.h b/lib/python3.12/site-packages/torch/include/ATen/ops/true_divide.h
new file mode 100644
index 0000000000000000000000000000000000000000..88967493dc2b0bb22de83b30d380ae96211600f8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/true_divide.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/true_divide_ops.h>
+
+namespace at {
+
+
+// aten::true_divide.Tensor(Tensor self, Tensor other) -> Tensor
+inline at::Tensor true_divide(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::true_divide_Tensor::call(self, other);
+}
+
+// aten::true_divide.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & true_divide_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::true_divide_out::call(self, other, out);
+}
+// aten::true_divide.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & true_divide_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::true_divide_out::call(self, other, out);
+}
+
+// aten::true_divide.Scalar(Tensor self, Scalar other) -> Tensor
+inline at::Tensor true_divide(const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::true_divide_Scalar::call(self, other);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/trunc_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/trunc_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..81f92307c4aead7aa1f466dd68962c87a2383794
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/trunc_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 trunc(const at::Tensor & self);
+TORCH_API at::Tensor & trunc_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & trunc_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & trunc_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/type_as_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/type_as_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7c3f58b82ca0ad2b82bb4589277ab1d4711729c7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/type_as_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 type_as(const at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unbind_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unbind_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..0ede026e9553cd66daa88b1d6fc57d0f8375f250
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unbind_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 ::std::vector<at::Tensor> unbind(const at::Tensor & self, int64_t dim=0);
+TORCH_API ::std::vector<at::Tensor> NestedTensor_unbind(const at::Tensor & self, int64_t dim=0);
+TORCH_API ::std::vector<at::Tensor> unbind(const at::Tensor & self, at::Dimname dim);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unfold_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unfold_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..d23fd54e44f2aa28d99878e7dd94fe523373ecec
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unfold_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 unfold_backward {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, 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::unfold_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "unfold_backward(Tensor grad_in, SymInt[] input_sizes, int dim, int size, int step) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad_in, c10::SymIntArrayRef input_sizes, int64_t dim, int64_t size, int64_t step);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_in, c10::SymIntArrayRef input_sizes, int64_t dim, int64_t size, int64_t step);
+};
+
+struct TORCH_API unfold_backward_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, int64_t, int64_t, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::unfold_backward";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "unfold_backward.out(Tensor grad_in, SymInt[] input_sizes, int dim, int size, int step, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & grad_in, c10::SymIntArrayRef input_sizes, int64_t dim, int64_t size, int64_t step, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_in, c10::SymIntArrayRef input_sizes, int64_t dim, int64_t size, int64_t step, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unfold_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unfold_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..568495b7acbb85a6c03bdacc70da5259f14368ef
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unfold_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 unfold(const at::Tensor & self, int64_t dimension, int64_t size, int64_t step);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unique_dim_consecutive_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unique_dim_consecutive_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8ac20e4a3b77b9b5e2ff35de21e45b84ea0c24c1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unique_dim_consecutive_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 &> 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);
+TORCH_API ::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);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_split.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_split.h
new file mode 100644
index 0000000000000000000000000000000000000000..dd6572434a67d1d7bd17db3a37f1d9cf8c00d581
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_split.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/unsafe_split_ops.h>
+
+namespace at {
+
+
+// aten::unsafe_split.Tensor(Tensor self, SymInt split_size, int dim=0) -> Tensor[]
+inline ::std::vector<at::Tensor> unsafe_split(const at::Tensor & self, int64_t split_size, int64_t dim=0) {
+    return at::_ops::unsafe_split_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> unsafe_split(const at::Tensor & self, int64_t split_size, int64_t dim=0) {
+    return at::_ops::unsafe_split_Tensor::call(self, split_size, dim);
+  }
+}
+
+// aten::unsafe_split.Tensor(Tensor self, SymInt split_size, int dim=0) -> Tensor[]
+inline ::std::vector<at::Tensor> unsafe_split_symint(const at::Tensor & self, c10::SymInt split_size, int64_t dim=0) {
+    return at::_ops::unsafe_split_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> unsafe_split(const at::Tensor & self, c10::SymInt split_size, int64_t dim=0) {
+    return at::_ops::unsafe_split_Tensor::call(self, split_size, dim);
+  }
+}
+
+// aten::unsafe_split.Tensor_out(Tensor self, SymInt split_size, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void unsafe_split_out(at::TensorList out, const at::Tensor & self, int64_t split_size, int64_t dim=0) {
+    return at::_ops::unsafe_split_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 unsafe_split_out(at::TensorList out, const at::Tensor & self, int64_t split_size, int64_t dim=0) {
+    return at::_ops::unsafe_split_Tensor_out::call(self, split_size, dim, out);
+  }
+}
+
+// aten::unsafe_split.Tensor_out(Tensor self, SymInt split_size, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void unsafe_split_outf(const at::Tensor & self, int64_t split_size, int64_t dim, at::TensorList out) {
+    return at::_ops::unsafe_split_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 unsafe_split_outf(const at::Tensor & self, int64_t split_size, int64_t dim, at::TensorList out) {
+    return at::_ops::unsafe_split_Tensor_out::call(self, split_size, dim, out);
+  }
+}
+
+// aten::unsafe_split.Tensor_out(Tensor self, SymInt split_size, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void unsafe_split_symint_out(at::TensorList out, const at::Tensor & self, c10::SymInt split_size, int64_t dim=0) {
+    return at::_ops::unsafe_split_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 unsafe_split_out(at::TensorList out, const at::Tensor & self, c10::SymInt split_size, int64_t dim=0) {
+    return at::_ops::unsafe_split_Tensor_out::call(self, split_size, dim, out);
+  }
+}
+
+// aten::unsafe_split.Tensor_out(Tensor self, SymInt split_size, int dim=0, *, Tensor(a!)[] out) -> ()
+inline void unsafe_split_symint_outf(const at::Tensor & self, c10::SymInt split_size, int64_t dim, at::TensorList out) {
+    return at::_ops::unsafe_split_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 unsafe_split_outf(const at::Tensor & self, c10::SymInt split_size, int64_t dim, at::TensorList out) {
+    return at::_ops::unsafe_split_Tensor_out::call(self, split_size, dim, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unsqueeze_copy_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unsqueeze_copy_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1792b2da9fcb5c12b812b19eba1a0ec119781413
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unsqueeze_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 & unsqueeze_copy_out(at::Tensor & out, const at::Tensor & self, int64_t dim);
+TORCH_API at::Tensor & unsqueeze_copy_outf(const at::Tensor & self, int64_t dim, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c413a2b881a99980203f6827ddc498d2fb48ece9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_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/upsample_linear1d_meta.h>
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor upsample_linear1d(const at::Tensor & input, at::OptionalIntArrayRef output_size, bool align_corners, ::std::optional<at::ArrayRef<double>> scale_factors);
+struct TORCH_API structured_upsample_linear1d_out_cpu : public at::meta::structured_upsample_linear1d {
+void impl(const at::Tensor & self, at::ArrayRef<int64_t> output_size, bool align_corners, ::std::optional<double> scales, const at::Tensor & out);
+};
+struct TORCH_API structured_upsample_linear1d_out_cuda : public at::meta::structured_upsample_linear1d {
+void impl(const at::Tensor & self, at::ArrayRef<int64_t> output_size, bool align_corners, ::std::optional<double> scales, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest1d_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest1d_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3abf0e46b6abb69f195a51eafcfe1f25f3891814
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest1d_backward_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_nearest1d_backward(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor upsample_nearest1d_backward_symint(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor & upsample_nearest1d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor & upsample_nearest1d_backward_outf(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, ::std::optional<double> scales, at::Tensor & grad_input);
+TORCH_API at::Tensor & upsample_nearest1d_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor & upsample_nearest1d_backward_symint_outf(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::std::optional<double> scales, at::Tensor & grad_input);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..421c1b3d0586382ecdbb5399dd2e612324fca9a7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_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_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);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_trilinear3d_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_trilinear3d_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..267212712c74ecd8181030f46345db862714699e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_trilinear3d_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_trilinear3d(const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::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_trilinear3d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::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_trilinear3d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::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_trilinear3d_outf(const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+TORCH_API at::Tensor & upsample_trilinear3d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::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_trilinear3d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::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/upsample_trilinear3d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_trilinear3d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..b7c4c3c2d4e7de0c97cf38a1304c46fb769404a5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_trilinear3d_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_trilinear3d_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_trilinear3d";
+  static constexpr const char* overload_name = "vec";
+  static constexpr const char* schema_str = "upsample_trilinear3d.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_trilinear3d_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, bool, ::std::optional<double>, ::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_trilinear3d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "upsample_trilinear3d.out(Tensor self, SymInt[3] output_size, bool align_corners, float? scales_d=None, 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_d, ::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_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+};
+
+struct TORCH_API upsample_trilinear3d {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, bool, ::std::optional<double>, ::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_trilinear3d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "upsample_trilinear3d(Tensor self, SymInt[3] output_size, bool align_corners, float? scales_d=None, 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_d, ::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_d, ::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/vander.h b/lib/python3.12/site-packages/torch/include/ATen/ops/vander.h
new file mode 100644
index 0000000000000000000000000000000000000000..86ffc2295a3e0238ba425ed201b979bb606beb29
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/vander.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/vander_ops.h>
+
+namespace at {
+
+
+// aten::vander(Tensor x, int? N=None, bool increasing=False) -> Tensor
+inline at::Tensor vander(const at::Tensor & x, ::std::optional<int64_t> N=::std::nullopt, bool increasing=false) {
+    return at::_ops::vander::call(x, N, increasing);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/view_as_real_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/view_as_real_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..2d8a4d4c170751eae9fbdf7953b9699ed6eddf6a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/view_as_real_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/view_as_real_copy_ops.h>
+
+namespace at {
+
+
+// aten::view_as_real_copy(Tensor self) -> Tensor
+inline at::Tensor view_as_real_copy(const at::Tensor & self) {
+    return at::_ops::view_as_real_copy::call(self);
+}
+
+// aten::view_as_real_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & view_as_real_copy_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::view_as_real_copy_out::call(self, out);
+}
+// aten::view_as_real_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & view_as_real_copy_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::view_as_real_copy_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/view_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/view_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ba8e92f6a12174ee3fbf24376b6d1b8ad81951da
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/view_compositeexplicitautograd_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 compositeexplicitautograd {
+
+TORCH_API at::Tensor view(const at::Tensor & self, at::ScalarType dtype);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/view_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/view_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..130cc1aa1957fab0dd205b2050e92bbf55e36631
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/view_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 view {
+  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::view";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "view(Tensor(a) self, SymInt[] size) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef size);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef size);
+};
+
+struct TORCH_API view_dtype {
+  using schema = at::Tensor (const at::Tensor &, at::ScalarType);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::view";
+  static constexpr const char* overload_name = "dtype";
+  static constexpr const char* schema_str = "view.dtype(Tensor(a) self, ScalarType dtype) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, at::ScalarType dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::ScalarType dtype);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2430dac31d52acedb6f05431b878d10248666252
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,29 @@
+#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 xlogy(const at::Scalar & self, const at::Tensor & other);
+TORCH_API at::Tensor & xlogy_out(at::Tensor & out, const at::Scalar & self, const at::Tensor & other);
+TORCH_API at::Tensor & xlogy_outf(const at::Scalar & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor xlogy(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & xlogy_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & xlogy_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & xlogy_(at::Tensor & self, const at::Scalar & other);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..44c1228a1aeca10c2716814ea609f3fe87f53df6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_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 xlogy(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & xlogy_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/zeros_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/zeros_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..88da4fff1364ed77f89cca67cdd95430a61a22e7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/zeros_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,34 @@
+#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 zeros(at::IntArrayRef size, ::std::optional<at::DimnameList> names, at::TensorOptions options={});
+TORCH_API at::Tensor zeros(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);
+TORCH_API at::Tensor & zeros_out(at::Tensor & out, at::IntArrayRef size, ::std::optional<at::DimnameList> names);
+TORCH_API at::Tensor & zeros_outf(at::IntArrayRef size, ::std::optional<at::DimnameList> names, at::Tensor & out);
+TORCH_API at::Tensor zeros(at::IntArrayRef size, at::TensorOptions options={});
+TORCH_API at::Tensor zeros(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 zeros_symint(c10::SymIntArrayRef size, at::TensorOptions options={});
+TORCH_API at::Tensor zeros_symint(c10::SymIntArrayRef 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 & zeros_out(at::Tensor & out, at::IntArrayRef size);
+TORCH_API at::Tensor & zeros_outf(at::IntArrayRef size, at::Tensor & out);
+TORCH_API at::Tensor & zeros_symint_out(at::Tensor & out, c10::SymIntArrayRef size);
+TORCH_API at::Tensor & zeros_symint_outf(c10::SymIntArrayRef size, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at