diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool3d_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool3d_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5bf4162543bebb2c77e29048cfac4d1c93cf2359
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool3d_backward_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _adaptive_avg_pool3d_backward_out(const at::Tensor & grad_output, const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor adaptive_avg_pool3d_backward_cpu(const at::Tensor & grad_output, const at::Tensor & self);
+TORCH_API at::Tensor adaptive_avg_pool3d_backward_cuda(const at::Tensor & grad_output, const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_add_relu_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_add_relu_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..6901d1f657e771c9660b58338b6fcafaed5f9c64
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_add_relu_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 _add_relu_Tensor {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_add_relu";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "_add_relu.Tensor(Tensor self, Tensor other, *, Scalar alpha=1) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha);
+};
+
+struct TORCH_API _add_relu__Tensor {
+  using schema = at::Tensor & (at::Tensor &, const at::Tensor &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_add_relu_";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "_add_relu_.Tensor(Tensor(a!) self, Tensor other, *, Scalar alpha=1) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha);
+};
+
+struct TORCH_API _add_relu_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const at::Scalar &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_add_relu";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_add_relu.out(Tensor self, Tensor other, *, Scalar alpha=1, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha, at::Tensor & out);
+};
+
+struct TORCH_API _add_relu_Scalar {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_add_relu";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "_add_relu.Scalar(Tensor self, Scalar other, Scalar alpha=1) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha);
+};
+
+struct TORCH_API _add_relu__Scalar {
+  using schema = at::Tensor & (at::Tensor &, const at::Scalar &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_add_relu_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "_add_relu_.Scalar(Tensor(a!) self, Scalar other, Scalar alpha=1) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha);
+};
+
+struct TORCH_API _add_relu_Scalar_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, const at::Scalar &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_add_relu";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "_add_relu.Scalar_out(Tensor self, Scalar other, Scalar alpha=1, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_amp_foreach_non_finite_check_and_unscale_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_amp_foreach_non_finite_check_and_unscale_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2540f4df92e1ba255b4ac36cce81a2e58500d4cc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_amp_foreach_non_finite_check_and_unscale_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 ::std::tuple<::std::vector<at::Tensor>,at::Tensor> _amp_foreach_non_finite_check_and_unscale(at::TensorList self, const at::Tensor & found_inf, const at::Tensor & inv_scale);
+TORCH_API void _amp_foreach_non_finite_check_and_unscale_out(at::TensorList out, at::TensorList self, at::Tensor & found_inf, const at::Tensor & inv_scale);
+TORCH_API void _amp_foreach_non_finite_check_and_unscale_outf(at::TensorList self, at::Tensor & found_inf, const at::Tensor & inv_scale, at::TensorList out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_tensor_metadata.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_tensor_metadata.h
new file mode 100644
index 0000000000000000000000000000000000000000..d8f830635eb4a40721a37c5df83f45e2edc5f72a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_tensor_metadata.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/_assert_tensor_metadata_ops.h>
+
+namespace at {
+
+
+// aten::_assert_tensor_metadata(Tensor a, SymInt[]? size=None, SymInt[]? stride=None, ScalarType? dtype=None, *, Device? device=None, Layout? layout=None) -> ()
+inline void _assert_tensor_metadata(const at::Tensor & a, at::OptionalIntArrayRef size=::std::nullopt, at::OptionalIntArrayRef stride=::std::nullopt, ::std::optional<at::ScalarType> dtype=::std::nullopt, ::std::optional<at::Device> device=::std::nullopt, ::std::optional<at::Layout> layout=::std::nullopt) {
+    return at::_ops::_assert_tensor_metadata::call(a, size.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*size)) : ::std::nullopt, stride.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*stride)) : ::std::nullopt, dtype, device, layout);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  void _assert_tensor_metadata(const at::Tensor & a, at::OptionalIntArrayRef size=::std::nullopt, at::OptionalIntArrayRef stride=::std::nullopt, ::std::optional<at::ScalarType> dtype=::std::nullopt, ::std::optional<at::Device> device=::std::nullopt, ::std::optional<at::Layout> layout=::std::nullopt) {
+    return at::_ops::_assert_tensor_metadata::call(a, size.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*size)) : ::std::nullopt, stride.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*stride)) : ::std::nullopt, dtype, device, layout);
+  }
+}
+
+// aten::_assert_tensor_metadata(Tensor a, SymInt[]? size=None, SymInt[]? stride=None, ScalarType? dtype=None, *, Device? device=None, Layout? layout=None) -> ()
+inline void _assert_tensor_metadata_symint(const at::Tensor & a, at::OptionalSymIntArrayRef size=::std::nullopt, at::OptionalSymIntArrayRef stride=::std::nullopt, ::std::optional<at::ScalarType> dtype=::std::nullopt, ::std::optional<at::Device> device=::std::nullopt, ::std::optional<at::Layout> layout=::std::nullopt) {
+    return at::_ops::_assert_tensor_metadata::call(a, size, stride, dtype, device, layout);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  void _assert_tensor_metadata(const at::Tensor & a, at::OptionalSymIntArrayRef size=::std::nullopt, at::OptionalSymIntArrayRef stride=::std::nullopt, ::std::optional<at::ScalarType> dtype=::std::nullopt, ::std::optional<at::Device> device=::std::nullopt, ::std::optional<at::Layout> layout=::std::nullopt) {
+    return at::_ops::_assert_tensor_metadata::call(a, size, stride, dtype, device, layout);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_tensor_metadata_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_tensor_metadata_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..07f5f4268713e275e9b34537868b351437cbaa28
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_tensor_metadata_meta_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 meta {
+
+TORCH_API void _assert_tensor_metadata(const at::Tensor & a, at::OptionalIntArrayRef size=::std::nullopt, at::OptionalIntArrayRef stride=::std::nullopt, ::std::optional<at::ScalarType> dtype=::std::nullopt, ::std::optional<at::Device> device=::std::nullopt, ::std::optional<at::Layout> layout=::std::nullopt);
+TORCH_API void _assert_tensor_metadata_symint(const at::Tensor & a, at::OptionalSymIntArrayRef size=::std::nullopt, at::OptionalSymIntArrayRef stride=::std::nullopt, ::std::optional<at::ScalarType> dtype=::std::nullopt, ::std::optional<at::Device> device=::std::nullopt, ::std::optional<at::Layout> layout=::std::nullopt);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index.h
new file mode 100644
index 0000000000000000000000000000000000000000..10c41f5432cae17d9aad6e6b48bead8af6981ae6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index.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/_batch_norm_impl_index_ops.h>
+
+namespace at {
+
+
+// aten::_batch_norm_impl_index(Tensor input, Tensor? weight, Tensor? bias, Tensor? running_mean, Tensor? running_var, bool training, float momentum, float eps, bool cudnn_enabled) -> (Tensor, Tensor, Tensor, Tensor, int)
+inline ::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) {
+    return at::_ops::_batch_norm_impl_index::call(input, weight, bias, running_mean, running_var, training, momentum, eps, cudnn_enabled);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..3549f5b41ed3647c23e8be1780324bd8ae0dc0be
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_impl_index_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/_batch_norm_impl_index_backward_ops.h>
+
+namespace at {
+
+
+// aten::_batch_norm_impl_index_backward(int impl_index, Tensor input, Tensor grad_output, Tensor? weight, Tensor? running_mean, Tensor? running_var, Tensor? save_mean, Tensor? save_var_transform, bool train, float eps, bool[3] output_mask, Tensor reservedSpace) -> (Tensor, Tensor, Tensor)
+inline ::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) {
+    return at::_ops::_batch_norm_impl_index_backward::call(impl_index, input, grad_output, weight, running_mean, running_var, save_mean, save_var_transform, train, eps, output_mask, reservedSpace);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Char_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Char_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..65388193206ff3e2c9425335637b62c8f8a0effe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Char_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_Char(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_Short.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Short.h
new file mode 100644
index 0000000000000000000000000000000000000000..af6396f3b73e2c9eb52b449df0ad942c963576dc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Short.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_Short_ops.h>
+
+namespace at {
+
+
+// aten::_cast_Short(Tensor self, bool non_blocking=False) -> Tensor
+inline at::Tensor _cast_Short(const at::Tensor & self, bool non_blocking=false) {
+    return at::_ops::_cast_Short::call(self, non_blocking);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cdist_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cdist_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d9ad166ce6cf8fee25ac15cb50342a487f4f1d76
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cdist_backward_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor _cdist_backward(const at::Tensor & grad, const at::Tensor & x1, const at::Tensor & x2, double p, const at::Tensor & cdist);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f14662f72bb0939558a2d9f76d8b799d686451fb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr_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 _convert_indices_from_coo_to_csr(const at::Tensor & self, int64_t size, bool out_int32=false);
+TORCH_API at::Tensor & _convert_indices_from_coo_to_csr_out(at::Tensor & out, const at::Tensor & self, int64_t size, bool out_int32=false);
+TORCH_API at::Tensor & _convert_indices_from_coo_to_csr_outf(const at::Tensor & self, int64_t size, bool out_int32, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..40a72f8599b66bf7d83dfa0c51ac22446429a15a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr_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/_convert_indices_from_coo_to_csr_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured__convert_indices_from_coo_to_csr_structured_cpu : public at::meta::structured__convert_indices_from_coo_to_csr {
+void impl(const at::Tensor & self, int64_t size, bool out_int32, const at::Tensor & out);
+};
+struct TORCH_API structured__convert_indices_from_coo_to_csr_structured_cuda : public at::meta::structured__convert_indices_from_coo_to_csr {
+void impl(const at::Tensor & self, int64_t size, bool out_int32, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..0cda84cb83d134cc327d9e280ac462f757db101b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr_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 _convert_indices_from_coo_to_csr {
+  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::_convert_indices_from_coo_to_csr";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_convert_indices_from_coo_to_csr(Tensor self, int size, *, bool out_int32=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t size, bool out_int32);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t size, bool out_int32);
+};
+
+struct TORCH_API _convert_indices_from_coo_to_csr_out {
+  using schema = at::Tensor & (const at::Tensor &, int64_t, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_convert_indices_from_coo_to_csr";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_convert_indices_from_coo_to_csr.out(Tensor self, int size, *, bool out_int32=False, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, int64_t size, bool out_int32, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t size, bool out_int32, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_csr_to_coo_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_csr_to_coo_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..628c62691e07802581d8dd42c59cc851810a94f7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_indices_from_csr_to_coo_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 _convert_indices_from_csr_to_coo(const at::Tensor & crow_indices, const at::Tensor & col_indices, bool out_int32=false, bool transpose=false);
+TORCH_API at::Tensor & _convert_indices_from_csr_to_coo_out(at::Tensor & out, const at::Tensor & crow_indices, const at::Tensor & col_indices, bool out_int32=false, bool transpose=false);
+TORCH_API at::Tensor & _convert_indices_from_csr_to_coo_outf(const at::Tensor & crow_indices, const at::Tensor & col_indices, bool out_int32, bool transpose, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..61c7460efdb3c1d4af119daeb6bc67a86105d208
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_backward_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API void _cudnn_rnn_backward_out_symint(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, const at::Tensor & output, const ::std::optional<at::Tensor> & grad_output, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, int64_t mode, c10::SymInt hidden_size, c10::SymInt proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, c10::SymIntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::TensorList out3);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,::std::vector<at::Tensor>> _cudnn_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 proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, at::IntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, const at::Tensor & reserve, ::std::array<bool,4> output_mask);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1267b7894c52ef4de0d0f7751802854f57cdd5d5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_rnn_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _cudnn_rnn_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, at::Tensor & out4, const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const ::std::optional<at::Tensor> & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, int64_t mode, int64_t hidden_size, int64_t proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, at::IntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _cudnn_rnn_outf(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const ::std::optional<at::Tensor> & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, int64_t mode, int64_t hidden_size, int64_t proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, at::IntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, at::Tensor & out4);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _cudnn_rnn_symint_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, at::Tensor & out4, const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const ::std::optional<at::Tensor> & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, int64_t mode, c10::SymInt hidden_size, c10::SymInt proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, c10::SymIntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _cudnn_rnn_symint_outf(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const ::std::optional<at::Tensor> & weight_buf, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, int64_t mode, c10::SymInt hidden_size, c10::SymInt proj_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, c10::SymIntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, at::Tensor & out4);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_get_plan_cache_max_size_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_get_plan_cache_max_size_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..65f1ebe28d3c1781ddec8f77539724fb1f9a1b14
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_get_plan_cache_max_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 _cufft_get_plan_cache_max_size {
+  using schema = int64_t (at::DeviceIndex);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_cufft_get_plan_cache_max_size";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_cufft_get_plan_cache_max_size(DeviceIndex device_index) -> int";
+  static int64_t call(at::DeviceIndex device_index);
+  static int64_t redispatch(c10::DispatchKeySet dispatchKeySet, at::DeviceIndex device_index);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_set_plan_cache_max_size.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_set_plan_cache_max_size.h
new file mode 100644
index 0000000000000000000000000000000000000000..78db7ef4a817779e3ff1d8e5f3f93ad8b79f56b7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_set_plan_cache_max_size.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/_cufft_set_plan_cache_max_size_ops.h>
+
+namespace at {
+
+
+// aten::_cufft_set_plan_cache_max_size(DeviceIndex device_index, int max_size) -> ()
+inline void _cufft_set_plan_cache_max_size(at::DeviceIndex device_index, int64_t max_size) {
+    return at::_ops::_cufft_set_plan_cache_max_size::call(device_index, max_size);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_set_plan_cache_max_size_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_set_plan_cache_max_size_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..397fda7a53b84a0949a7f72352fc931e963db3fc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_set_plan_cache_max_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 _cufft_set_plan_cache_max_size {
+  using schema = void (at::DeviceIndex, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_cufft_set_plan_cache_max_size";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_cufft_set_plan_cache_max_size(DeviceIndex device_index, int max_size) -> ()";
+  static void call(at::DeviceIndex device_index, int64_t max_size);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::DeviceIndex device_index, int64_t max_size);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cummax_helper.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cummax_helper.h
new file mode 100644
index 0000000000000000000000000000000000000000..bdffb81a2eb7760aa1dc7d820d15eb9f4714d6f0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cummax_helper.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/_cummax_helper_ops.h>
+
+namespace at {
+
+
+// aten::_cummax_helper(Tensor self, Tensor(a!) values, Tensor(b!) indices, int dim) -> ()
+inline void _cummax_helper(const at::Tensor & self, at::Tensor & values, at::Tensor & indices, int64_t dim) {
+    return at::_ops::_cummax_helper::call(self, values, indices, dim);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_efficient_attention_forward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_efficient_attention_forward.h
new file mode 100644
index 0000000000000000000000000000000000000000..f139164e7a1d18cceb3ea496230b6967bdec06ae
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_efficient_attention_forward.h
@@ -0,0 +1,47 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_efficient_attention_forward_ops.h>
+
+namespace at {
+
+
+// aten::_efficient_attention_forward(Tensor query, Tensor key, Tensor value, Tensor? bias, Tensor? cu_seqlens_q, Tensor? cu_seqlens_k, SymInt? max_seqlen_q, SymInt? max_seqlen_k, float dropout_p, int custom_mask_type, bool compute_log_sumexp=False, *, float? scale=None, Tensor? seqlen_k=None, int? window_size=None) -> (Tensor output, Tensor logsumexp, Tensor philox_seed, Tensor philox_offset, SymInt max_seqlen_batch_q, SymInt max_seqlen_batch_k)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt> _efficient_attention_forward(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, ::std::optional<int64_t> max_seqlen_q, ::std::optional<int64_t> max_seqlen_k, double dropout_p, int64_t custom_mask_type, bool compute_log_sumexp=false, ::std::optional<double> scale=::std::nullopt, const ::std::optional<at::Tensor> & seqlen_k={}, ::std::optional<int64_t> window_size=::std::nullopt) {
+    return at::_ops::_efficient_attention_forward::call(query, key, value, bias, cu_seqlens_q, cu_seqlens_k, max_seqlen_q.has_value() ? ::std::make_optional(c10::SymInt(*max_seqlen_q)) : ::std::nullopt, max_seqlen_k.has_value() ? ::std::make_optional(c10::SymInt(*max_seqlen_k)) : ::std::nullopt, dropout_p, custom_mask_type, compute_log_sumexp, scale, seqlen_k, window_size);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt> _efficient_attention_forward(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, ::std::optional<int64_t> max_seqlen_q, ::std::optional<int64_t> max_seqlen_k, double dropout_p, int64_t custom_mask_type, bool compute_log_sumexp=false, ::std::optional<double> scale=::std::nullopt, const ::std::optional<at::Tensor> & seqlen_k={}, ::std::optional<int64_t> window_size=::std::nullopt) {
+    return at::_ops::_efficient_attention_forward::call(query, key, value, bias, cu_seqlens_q, cu_seqlens_k, max_seqlen_q.has_value() ? ::std::make_optional(c10::SymInt(*max_seqlen_q)) : ::std::nullopt, max_seqlen_k.has_value() ? ::std::make_optional(c10::SymInt(*max_seqlen_k)) : ::std::nullopt, dropout_p, custom_mask_type, compute_log_sumexp, scale, seqlen_k, window_size);
+  }
+}
+
+// aten::_efficient_attention_forward(Tensor query, Tensor key, Tensor value, Tensor? bias, Tensor? cu_seqlens_q, Tensor? cu_seqlens_k, SymInt? max_seqlen_q, SymInt? max_seqlen_k, float dropout_p, int custom_mask_type, bool compute_log_sumexp=False, *, float? scale=None, Tensor? seqlen_k=None, int? window_size=None) -> (Tensor output, Tensor logsumexp, Tensor philox_seed, Tensor philox_offset, SymInt max_seqlen_batch_q, SymInt max_seqlen_batch_k)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt> _efficient_attention_forward_symint(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, ::std::optional<c10::SymInt> max_seqlen_q, ::std::optional<c10::SymInt> max_seqlen_k, double dropout_p, int64_t custom_mask_type, bool compute_log_sumexp=false, ::std::optional<double> scale=::std::nullopt, const ::std::optional<at::Tensor> & seqlen_k={}, ::std::optional<int64_t> window_size=::std::nullopt) {
+    return at::_ops::_efficient_attention_forward::call(query, key, value, bias, cu_seqlens_q, cu_seqlens_k, max_seqlen_q, max_seqlen_k, dropout_p, custom_mask_type, compute_log_sumexp, scale, seqlen_k, window_size);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt> _efficient_attention_forward(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, ::std::optional<c10::SymInt> max_seqlen_q, ::std::optional<c10::SymInt> max_seqlen_k, double dropout_p, int64_t custom_mask_type, bool compute_log_sumexp=false, ::std::optional<double> scale=::std::nullopt, const ::std::optional<at::Tensor> & seqlen_k={}, ::std::optional<int64_t> window_size=::std::nullopt) {
+    return at::_ops::_efficient_attention_forward::call(query, key, value, bias, cu_seqlens_q, cu_seqlens_k, max_seqlen_q, max_seqlen_k, dropout_p, custom_mask_type, compute_log_sumexp, scale, seqlen_k, window_size);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_efficient_attention_forward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_efficient_attention_forward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2690e00ec786c48d587e5e26527acf3684cf24aa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_efficient_attention_forward_cuda_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt> _efficient_attention_forward(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, ::std::optional<int64_t> max_seqlen_q, ::std::optional<int64_t> max_seqlen_k, double dropout_p, int64_t custom_mask_type, bool compute_log_sumexp=false, ::std::optional<double> scale=::std::nullopt, const ::std::optional<at::Tensor> & seqlen_k={}, ::std::optional<int64_t> window_size=::std::nullopt);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt> _efficient_attention_forward_symint(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, ::std::optional<c10::SymInt> max_seqlen_q, ::std::optional<c10::SymInt> max_seqlen_k, double dropout_p, int64_t custom_mask_type, bool compute_log_sumexp=false, ::std::optional<double> scale=::std::nullopt, const ::std::optional<at::Tensor> & seqlen_k={}, ::std::optional<int64_t> window_size=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_empty_per_channel_affine_quantized_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_empty_per_channel_affine_quantized_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..3ae737a61247953142032733689e0dca0416702c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_empty_per_channel_affine_quantized_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 & _empty_per_channel_affine_quantized_out_symint(c10::SymIntArrayRef size, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+TORCH_API at::Tensor empty_per_channel_affine_quantized_other_backends_stub(at::IntArrayRef size, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, ::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=c10::MemoryFormat::Contiguous);
+TORCH_API at::Tensor empty_per_channel_affine_quantized(at::IntArrayRef size, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, ::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=c10::MemoryFormat::Contiguous);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_euclidean_dist_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_euclidean_dist_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..3bdf109b8724dbe6ff7c4fee18486ae082144f0f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_euclidean_dist_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 _euclidean_dist(const at::Tensor & x1, const at::Tensor & x2);
+TORCH_API at::Tensor & _euclidean_dist_out(const at::Tensor & x1, const at::Tensor & x2, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1689c40e3072c3a333053fb7cb6e2f380361fd08
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _fake_quantize_per_tensor_affine_cachemask_tensor_qparams {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const 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::_fake_quantize_per_tensor_affine_cachemask_tensor_qparams";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_fake_quantize_per_tensor_affine_cachemask_tensor_qparams(Tensor self, Tensor scale, Tensor zero_point, Tensor fake_quant_enabled, int quant_min, int quant_max) -> (Tensor output, Tensor mask)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, const at::Tensor & fake_quant_enabled, int64_t quant_min, int64_t quant_max);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, const at::Tensor & fake_quant_enabled, int64_t quant_min, int64_t quant_max);
+};
+
+struct TORCH_API _fake_quantize_per_tensor_affine_cachemask_tensor_qparams_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, int64_t, int64_t, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fake_quantize_per_tensor_affine_cachemask_tensor_qparams";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_fake_quantize_per_tensor_affine_cachemask_tensor_qparams.out(Tensor self, Tensor scale, Tensor zero_point, Tensor fake_quant_enabled, int quant_min, int quant_max, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, const at::Tensor & fake_quant_enabled, int64_t quant_min, int64_t quant_max, at::Tensor & out0, at::Tensor & out1);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, const at::Tensor & fake_quant_enabled, int64_t quant_min, int64_t quant_max, at::Tensor & out0, at::Tensor & out1);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_acos_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_acos_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..3e17d8108b4ff683034bb48bcb80f04191a43725
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_acos_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_acos {
+  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_acos";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_acos(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_acos_ {
+  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_acos_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_acos_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_acos_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_acos";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_acos.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_asin_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_asin_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..75f6798daa999c62a366b0c6077223b216c65a74
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_asin_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_asin(at::TensorList self);
+TORCH_API void _foreach_asin_(at::TensorList self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_cos_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_cos_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5747fa171a120bf669755a1a7d7b64408289d6a1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_cos_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_cos(at::TensorList self);
+TORCH_API void _foreach_cos_(at::TensorList self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_erfc_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_erfc_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..3304446b3e0577d3a37e1a986300c2ed8f01ca33
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_erfc_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_erfc_slow(at::TensorList self);
+TORCH_API void _foreach_erfc_out(at::TensorList self, at::TensorList out);
+TORCH_API void foreach_tensor_erfc_slow_(at::TensorList self);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_erfc_cuda(at::TensorList self);
+TORCH_API void foreach_tensor_erfc_cuda_(at::TensorList self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_lerp.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_lerp.h
new file mode 100644
index 0000000000000000000000000000000000000000..188176b833736d4c49832f98e8f4eeb087bb5abc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_lerp.h
@@ -0,0 +1,82 @@
+#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_lerp_ops.h>
+
+namespace at {
+
+
+// aten::_foreach_lerp.List(Tensor[] self, Tensor[] tensors1, Tensor[] weights) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_lerp(at::TensorList self, at::TensorList tensors1, at::TensorList weights) {
+    return at::_ops::_foreach_lerp_List::call(self, tensors1, weights);
+}
+
+// aten::_foreach_lerp_.List(Tensor(a!)[] self, Tensor[] tensors1, Tensor[] weights) -> ()
+inline void _foreach_lerp_(at::TensorList self, at::TensorList tensors1, at::TensorList weights) {
+    return at::_ops::_foreach_lerp__List::call(self, tensors1, weights);
+}
+
+// aten::_foreach_lerp.Scalar(Tensor[] self, Tensor[] tensors1, Scalar weight) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_lerp(at::TensorList self, at::TensorList tensors1, const at::Scalar & weight) {
+    return at::_ops::_foreach_lerp_Scalar::call(self, tensors1, weight);
+}
+
+// aten::_foreach_lerp_.Scalar(Tensor(a!)[] self, Tensor[] tensors1, Scalar weight) -> ()
+inline void _foreach_lerp_(at::TensorList self, at::TensorList tensors1, const at::Scalar & weight) {
+    return at::_ops::_foreach_lerp__Scalar::call(self, tensors1, weight);
+}
+
+// aten::_foreach_lerp.ScalarList(Tensor[] self, Tensor[] tensors1, Scalar[] weight) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_lerp(at::TensorList self, at::TensorList tensors1, at::ArrayRef<at::Scalar> weight) {
+    return at::_ops::_foreach_lerp_ScalarList::call(self, tensors1, weight);
+}
+
+// aten::_foreach_lerp_.ScalarList(Tensor(a!)[] self, Tensor[] tensors1, Scalar[] weight) -> ()
+inline void _foreach_lerp_(at::TensorList self, at::TensorList tensors1, at::ArrayRef<at::Scalar> weight) {
+    return at::_ops::_foreach_lerp__ScalarList::call(self, tensors1, weight);
+}
+
+// aten::_foreach_lerp.List_out(Tensor[] self, Tensor[] tensors1, Tensor[] weights, *, Tensor(a!)[] out) -> ()
+inline void _foreach_lerp_out(at::TensorList out, at::TensorList self, at::TensorList tensors1, at::TensorList weights) {
+    return at::_ops::_foreach_lerp_List_out::call(self, tensors1, weights, out);
+}
+// aten::_foreach_lerp.List_out(Tensor[] self, Tensor[] tensors1, Tensor[] weights, *, Tensor(a!)[] out) -> ()
+inline void _foreach_lerp_outf(at::TensorList self, at::TensorList tensors1, at::TensorList weights, at::TensorList out) {
+    return at::_ops::_foreach_lerp_List_out::call(self, tensors1, weights, out);
+}
+
+// aten::_foreach_lerp.Scalar_out(Tensor[] self, Tensor[] tensors1, Scalar weight, *, Tensor(a!)[] out) -> ()
+inline void _foreach_lerp_out(at::TensorList out, at::TensorList self, at::TensorList tensors1, const at::Scalar & weight) {
+    return at::_ops::_foreach_lerp_Scalar_out::call(self, tensors1, weight, out);
+}
+// aten::_foreach_lerp.Scalar_out(Tensor[] self, Tensor[] tensors1, Scalar weight, *, Tensor(a!)[] out) -> ()
+inline void _foreach_lerp_outf(at::TensorList self, at::TensorList tensors1, const at::Scalar & weight, at::TensorList out) {
+    return at::_ops::_foreach_lerp_Scalar_out::call(self, tensors1, weight, out);
+}
+
+// aten::_foreach_lerp.ScalarList_out(Tensor[] self, Tensor[] tensors1, Scalar[] weight, *, Tensor(a!)[] out) -> ()
+inline void _foreach_lerp_out(at::TensorList out, at::TensorList self, at::TensorList tensors1, at::ArrayRef<at::Scalar> weight) {
+    return at::_ops::_foreach_lerp_ScalarList_out::call(self, tensors1, weight, out);
+}
+// aten::_foreach_lerp.ScalarList_out(Tensor[] self, Tensor[] tensors1, Scalar[] weight, *, Tensor(a!)[] out) -> ()
+inline void _foreach_lerp_outf(at::TensorList self, at::TensorList tensors1, at::ArrayRef<at::Scalar> weight, at::TensorList out) {
+    return at::_ops::_foreach_lerp_ScalarList_out::call(self, tensors1, weight, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_lerp_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_lerp_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f9d7f553ba407fa2b9a1b89eb91eb17d7cfac396
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_lerp_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 ::std::vector<at::Tensor> _foreach_lerp(at::TensorList self, at::TensorList tensors1, at::TensorList weights);
+TORCH_API void _foreach_lerp_out(at::TensorList out, at::TensorList self, at::TensorList tensors1, at::TensorList weights);
+TORCH_API void _foreach_lerp_outf(at::TensorList self, at::TensorList tensors1, at::TensorList weights, at::TensorList out);
+TORCH_API void _foreach_lerp_(at::TensorList self, at::TensorList tensors1, at::TensorList weights);
+TORCH_API ::std::vector<at::Tensor> _foreach_lerp(at::TensorList self, at::TensorList tensors1, const at::Scalar & weight);
+TORCH_API void _foreach_lerp_out(at::TensorList out, at::TensorList self, at::TensorList tensors1, const at::Scalar & weight);
+TORCH_API void _foreach_lerp_outf(at::TensorList self, at::TensorList tensors1, const at::Scalar & weight, at::TensorList out);
+TORCH_API void _foreach_lerp_(at::TensorList self, at::TensorList tensors1, const at::Scalar & weight);
+TORCH_API ::std::vector<at::Tensor> _foreach_lerp(at::TensorList self, at::TensorList tensors1, at::ArrayRef<at::Scalar> weight);
+TORCH_API void _foreach_lerp_out(at::TensorList out, at::TensorList self, at::TensorList tensors1, at::ArrayRef<at::Scalar> weight);
+TORCH_API void _foreach_lerp_outf(at::TensorList self, at::TensorList tensors1, at::ArrayRef<at::Scalar> weight, at::TensorList out);
+TORCH_API void _foreach_lerp_(at::TensorList self, at::TensorList tensors1, at::ArrayRef<at::Scalar> weight);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_lgamma_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_lgamma_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..352e488b43cd8b7f2c7d0b3e653320dd549debd9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_lgamma_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_lgamma(at::TensorList self);
+TORCH_API void _foreach_lgamma_out(at::TensorList out, at::TensorList self);
+TORCH_API void _foreach_lgamma_outf(at::TensorList self, at::TensorList out);
+TORCH_API void _foreach_lgamma_(at::TensorList self);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log10_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log10_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5cf8e7b30c965bd6e40b95cde9afd66e48f3334f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log10_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_log10(at::TensorList self);
+TORCH_API void _foreach_log10_out(at::TensorList out, at::TensorList self);
+TORCH_API void _foreach_log10_outf(at::TensorList self, at::TensorList out);
+TORCH_API void _foreach_log10_(at::TensorList self);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_maximum_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_maximum_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..76ce0eaae371d4d9a423ac0c9376ba6936839f9c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_maximum_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 ::std::vector<at::Tensor> _foreach_maximum(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_maximum_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> _foreach_maximum(at::TensorList self, at::TensorList other);
+TORCH_API void _foreach_maximum_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> _foreach_maximum(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_maximum_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..07ae157c5eb4811c046f59c4534cdd7abc9ad992
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum_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 ::std::vector<at::Tensor> _foreach_minimum(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_minimum_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> _foreach_minimum(at::TensorList self, at::TensorList other);
+TORCH_API void _foreach_minimum_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> _foreach_minimum(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_minimum_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..76168652209f053f71311862b0f13f20faf5dd46
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum_native.h
@@ -0,0 +1,35 @@
+#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_clamp_max_scalar_kernel_slow(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_minimum_Scalar_out(at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+TORCH_API void foreach_tensor_clamp_max_scalar_kernel_slow_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_max_scalar_kernel_cuda(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void foreach_tensor_clamp_max_scalar_kernel_cuda_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_max_list_kernel_slow(at::TensorList self, at::TensorList other);
+TORCH_API void _foreach_minimum_List_out(at::TensorList self, at::TensorList other, at::TensorList out);
+TORCH_API void foreach_tensor_clamp_max_list_kernel_slow_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_max_list_kernel_cuda(at::TensorList self, at::TensorList other);
+TORCH_API void foreach_tensor_clamp_max_list_kernel_cuda_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_max_scalarlist_kernel_slow(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_minimum_ScalarList_out(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+TORCH_API void foreach_tensor_clamp_max_scalarlist_kernel_slow_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_max_scalarlist_kernel_cuda(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void foreach_tensor_clamp_max_scalarlist_kernel_cuda_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_mul_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_mul_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..620aa683133f352e7684193de7844490f8fcff2f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_mul_native.h
@@ -0,0 +1,40 @@
+#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_mul_scalar_kernel_slow(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_mul_Scalar_out(at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+TORCH_API void foreach_tensor_mul_scalar_kernel_slow_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_mul_scalar_kernel_cuda(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void foreach_tensor_mul_scalar_kernel_cuda_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_mul_list_kernel_slow(at::TensorList self, at::TensorList other);
+TORCH_API void _foreach_mul_List_out(at::TensorList self, at::TensorList other, at::TensorList out);
+TORCH_API void foreach_tensor_mul_list_kernel_slow_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_mul_list_kernel_cuda(at::TensorList self, at::TensorList other);
+TORCH_API void foreach_tensor_mul_list_kernel_cuda_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_mul_scalarlist_kernel_slow(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_mul_ScalarList_out(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+TORCH_API void foreach_tensor_mul_scalarlist_kernel_slow_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_mul_scalarlist_kernel_cuda(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void foreach_tensor_mul_scalarlist_kernel_cuda_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_mul_tensor_kernel_slow(at::TensorList self, const at::Tensor & other);
+TORCH_API void _foreach_mul_Tensor_out(at::TensorList self, const at::Tensor & other, at::TensorList out);
+TORCH_API void foreach_tensor_mul_tensor_kernel_slow_(at::TensorList self, const at::Tensor & other);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_mul_tensor_kernel_cuda(at::TensorList self, const at::Tensor & other);
+TORCH_API void foreach_tensor_mul_tensor_kernel_cuda_(at::TensorList self, const at::Tensor & other);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_norm_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_norm_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..38340e455a19e770508400df6cbe8433f9545f10
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_norm_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _foreach_norm_Scalar {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, 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::_foreach_norm";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "_foreach_norm.Scalar(Tensor[] self, Scalar ord=2, ScalarType? dtype=None) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(at::TensorList self, const at::Scalar & ord, ::std::optional<at::ScalarType> dtype);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Scalar & ord, ::std::optional<at::ScalarType> dtype);
+};
+
+struct TORCH_API _foreach_norm_Scalar_out {
+  using schema = void (at::TensorList, const at::Scalar &, ::std::optional<at::ScalarType>, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_norm";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "_foreach_norm.Scalar_out(Tensor[] self, Scalar ord=2, ScalarType? dtype=None, *, Tensor(a!)[] out) -> ()";
+  static void call(at::TensorList self, const at::Scalar & ord, ::std::optional<at::ScalarType> dtype, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Scalar & ord, ::std::optional<at::ScalarType> dtype, at::TensorList out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sinh_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sinh_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..00360daac7ed49fbf5745313dcfea321c0fd3a87
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sinh_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_sinh(at::TensorList self);
+TORCH_API void _foreach_sinh_(at::TensorList self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sqrt_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sqrt_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..85daa667775d85bd37be44c603c26703bd486b2a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sqrt_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_sqrt {
+  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_sqrt";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_sqrt(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_sqrt_ {
+  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_sqrt_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_sqrt_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_sqrt_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_sqrt";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_sqrt.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_zero.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_zero.h
new file mode 100644
index 0000000000000000000000000000000000000000..babb15a46726ceb1f6b9622ad44fc1d925306b98
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_zero.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_zero_ops.h>
+
+namespace at {
+
+
+// aten::_foreach_zero_(Tensor(a!)[] self) -> ()
+inline void _foreach_zero_(at::TensorList self) {
+    return at::_ops::_foreach_zero_::call(self);
+}
+
+// aten::_foreach_zero.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_zero_out(at::TensorList out, at::TensorList self) {
+    return at::_ops::_foreach_zero_out::call(self, out);
+}
+// aten::_foreach_zero.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_zero_outf(at::TensorList self, at::TensorList out) {
+    return at::_ops::_foreach_zero_out::call(self, out);
+}
+
+// aten::_foreach_zero(Tensor[] self) -> Tensor[] self_out
+inline ::std::vector<at::Tensor> _foreach_zero(at::TensorList self) {
+    return at::_ops::_foreach_zero::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_assert_scalar.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_assert_scalar.h
new file mode 100644
index 0000000000000000000000000000000000000000..4311a6986417c9fe39fc9addc09284be6acaa9d0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_assert_scalar.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/_functional_assert_scalar_ops.h>
+
+namespace at {
+
+
+// aten::_functional_assert_scalar(Scalar self, str assert_msg, Tensor dep_token) -> Tensor
+inline at::Tensor _functional_assert_scalar(const at::Scalar & self, c10::string_view assert_msg, const at::Tensor & dep_token) {
+    return at::_ops::_functional_assert_scalar::call(self, assert_msg, dep_token);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..82fd19c70f022d615634b61487af8e957aaa0968
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_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_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 cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_index_put_impl_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_index_put_impl_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b0cea2784d68620290551c825d38200b964d7ed6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_index_put_impl_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 & _index_put_impl_(at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false, bool unsafe=false);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_indices_copy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_indices_copy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b7b1613909caadf6122d4fe91e91ccf2581eaf68
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_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 & _indices_copy_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor _indices_copy(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_indices_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_indices_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..697068bf5181d91b847a06e36abfa7e4798defe2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_indices_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _indices {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_indices";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_indices(Tensor(a) self) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_int_mm.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_int_mm.h
new file mode 100644
index 0000000000000000000000000000000000000000..b5c5f62fa1917bcb3f74948d09ee92d2145c87fe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_int_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/_int_mm_ops.h>
+
+namespace at {
+
+
+// aten::_int_mm(Tensor self, Tensor mat2) -> Tensor
+inline at::Tensor _int_mm(const at::Tensor & self, const at::Tensor & mat2) {
+    return at::_ops::_int_mm::call(self, mat2);
+}
+
+// aten::_int_mm.out(Tensor self, Tensor mat2, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _int_mm_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & mat2) {
+    return at::_ops::_int_mm_out::call(self, mat2, out);
+}
+// aten::_int_mm.out(Tensor self, Tensor mat2, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _int_mm_outf(const at::Tensor & self, const at::Tensor & mat2, at::Tensor & out) {
+    return at::_ops::_int_mm_out::call(self, mat2, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_lazy_clone_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_lazy_clone_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ed6be2b9c723b0f88641b8f7012e92e58a9a2b66
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_lazy_clone_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 _lazy_clone(const at::Tensor & self);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_svd_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_svd_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..083c430cb6db0323971d7bfb457d505aab7652e5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_svd_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _linalg_svd(const at::Tensor & A, bool full_matrices=false, bool compute_uv=true, ::std::optional<c10::string_view> driver=::std::nullopt);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _linalg_svd_out(at::Tensor & U, at::Tensor & S, at::Tensor & Vh, const at::Tensor & A, bool full_matrices=false, bool compute_uv=true, ::std::optional<c10::string_view> driver=::std::nullopt);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _linalg_svd_outf(const at::Tensor & A, bool full_matrices, bool compute_uv, ::std::optional<c10::string_view> driver, at::Tensor & U, at::Tensor & S, at::Tensor & Vh);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ec852764e1b940f141eee456d0bfc73133987ffa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_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 _log_softmax(const at::Tensor & self, int64_t dim, bool half_to_float);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..96f7ce727e0c095dc2c018d3b0d5f9048fb01344
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_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 : public at::impl::MetaBase {
+    
+    
+    void meta(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/_log_softmax_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5d8f2ddf4e2880798942e88f7e85959b56cf5704
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_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/_log_softmax_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_log_softmax_cpu_out : public at::meta::structured__log_softmax {
+void impl(const at::Tensor & self, int64_t dim, bool half_to_float, const at::Tensor & out);
+};
+struct TORCH_API structured_log_softmax_cuda_out : public at::meta::structured__log_softmax {
+void impl(const at::Tensor & self, int64_t dim, bool half_to_float, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_make_dual.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_make_dual.h
new file mode 100644
index 0000000000000000000000000000000000000000..1c7316d1f1b479ab3b4ca36064f760dad92afd87
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_make_dual.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/_make_dual_ops.h>
+
+namespace at {
+
+
+// aten::_make_dual(Tensor(a) primal, Tensor tangent, int level) -> Tensor(a)
+inline at::Tensor _make_dual(const at::Tensor & primal, const at::Tensor & tangent, int64_t level) {
+    return at::_ops::_make_dual::call(primal, tangent, level);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_make_dual_copy_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_make_dual_copy_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1bf9c9afe0d70ec375370b59d597a5a18a0eb2c9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_make_dual_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 _make_dual_copy {
+  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::_make_dual_copy";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_make_dual_copy(Tensor primal, Tensor tangent, int level) -> Tensor";
+  static at::Tensor call(const at::Tensor & primal, const at::Tensor & tangent, int64_t level);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & primal, const at::Tensor & tangent, int64_t level);
+};
+
+struct TORCH_API _make_dual_copy_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::_make_dual_copy";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_make_dual_copy.out(Tensor primal, Tensor tangent, int level, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & primal, const at::Tensor & tangent, int64_t level, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & primal, const at::Tensor & tangent, int64_t level, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_make_per_tensor_quantized_tensor_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_make_per_tensor_quantized_tensor_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f9fd0f9ba350eef414b477217538093f52167c82
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_make_per_tensor_quantized_tensor_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 _make_per_tensor_quantized_tensor(const at::Tensor & self, double scale, int64_t zero_point);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_make_per_tensor_quantized_tensor_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_make_per_tensor_quantized_tensor_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0c9c785ec308156e3881226a2d8e39e6b70659df
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_make_per_tensor_quantized_tensor_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 _make_per_tensor_quantized_tensor(const at::Tensor & self, double scale, int64_t zero_point);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f668278755c6b21e2ff373965900b49417cc268a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_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 _native_batch_norm_legit {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, at::Tensor &, at::Tensor &, bool, 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";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_native_batch_norm_legit(Tensor input, Tensor? weight, Tensor? bias, Tensor(a!) running_mean, Tensor(b!) running_var, bool training, 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, at::Tensor & running_mean, at::Tensor & running_var, bool training, 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, at::Tensor & running_mean, at::Tensor & running_var, bool training, double momentum, double eps);
+};
+
+struct TORCH_API _native_batch_norm_legit_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> (const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, at::Tensor &, at::Tensor &, bool, double, double, at::Tensor &, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_native_batch_norm_legit";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_native_batch_norm_legit.out(Tensor input, Tensor? weight, Tensor? bias, Tensor(a!) running_mean, Tensor(b!) running_var, bool training, float momentum, float eps, *, Tensor(d!) out, Tensor(e!) save_mean, Tensor(f!) save_invstd) -> (Tensor(d!), Tensor(e!), Tensor(f!))";
+  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, 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);
+  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, 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);
+};
+
+struct TORCH_API _native_batch_norm_legit_no_stats {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, bool, double, double);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_native_batch_norm_legit";
+  static constexpr const char* overload_name = "no_stats";
+  static constexpr const char* schema_str = "_native_batch_norm_legit.no_stats(Tensor input, Tensor? weight, Tensor? bias, bool training, 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, bool training, 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, bool training, double momentum, double eps);
+};
+
+struct TORCH_API _native_batch_norm_legit_no_stats_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> (const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, bool, double, double, at::Tensor &, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_native_batch_norm_legit";
+  static constexpr const char* overload_name = "no_stats_out";
+  static constexpr const char* schema_str = "_native_batch_norm_legit.no_stats_out(Tensor input, Tensor? weight, Tensor? bias, bool training, float momentum, float eps, *, Tensor(a!) out, Tensor(b!) save_mean, Tensor(c!) save_invstd) -> (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, bool training, double momentum, double eps, at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd);
+  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, bool training, double momentum, double eps, at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd);
+};
+
+struct TORCH_API _native_batch_norm_legit_functional {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,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 &, bool, 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_functional";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_native_batch_norm_legit_functional(Tensor input, Tensor? weight, Tensor? bias, Tensor running_mean, Tensor running_var, bool training, float momentum, float eps) -> (Tensor, Tensor, Tensor, Tensor running_mean_out, Tensor running_var_out)";
+  static ::std::tuple<at::Tensor,at::Tensor,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, bool training, double momentum, double eps);
+  static ::std::tuple<at::Tensor,at::Tensor,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, bool training, double momentum, double eps);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_jagged_dummy_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_jagged_dummy_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..0a766dc98a87fbe28f976211f10dc136ce9961aa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_jagged_dummy_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_jagged_dummy {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_nested_get_jagged_dummy";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_nested_get_jagged_dummy(Tensor any) -> Tensor";
+  static at::Tensor call(const at::Tensor & any);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & any);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_min_seqlen_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_min_seqlen_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b9d2e5fd5be542077158871f57ae9c937e4f7895
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_min_seqlen_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/_nested_tensor_from_mask_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_from_mask_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..3ab3666eafd01f5179d47f767f8ea6a404a4c0f1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_from_mask_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _nested_tensor_from_mask {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_nested_tensor_from_mask";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_nested_tensor_from_mask(Tensor t, Tensor mask, bool mask_check=True) -> Tensor";
+  static at::Tensor call(const at::Tensor & t, const at::Tensor & mask, bool mask_check);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & t, const at::Tensor & mask, bool mask_check);
+};
+
+struct TORCH_API _nested_tensor_from_mask_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_nested_tensor_from_mask";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_nested_tensor_from_mask.out(Tensor t, Tensor mask, bool mask_check=True, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & t, const at::Tensor & mask, bool mask_check, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & t, const at::Tensor & mask, bool mask_check, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_storage_offsets.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_storage_offsets.h
new file mode 100644
index 0000000000000000000000000000000000000000..523e961d811a05e3a018be2400cbe9cec74953ef
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_storage_offsets.h
@@ -0,0 +1,34 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_nested_tensor_storage_offsets_ops.h>
+
+namespace at {
+
+
+// aten::_nested_tensor_storage_offsets.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _nested_tensor_storage_offsets_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::_nested_tensor_storage_offsets_out::call(self, out);
+}
+// aten::_nested_tensor_storage_offsets.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _nested_tensor_storage_offsets_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::_nested_tensor_storage_offsets_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_pack_padded_sequence.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_pack_padded_sequence.h
new file mode 100644
index 0000000000000000000000000000000000000000..524c8994a3468ef6766c72f6602c5d5827988bce
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_pack_padded_sequence.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/_pack_padded_sequence_ops.h>
+
+namespace at {
+
+
+// aten::_pack_padded_sequence(Tensor input, Tensor lengths, bool batch_first) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _pack_padded_sequence(const at::Tensor & input, const at::Tensor & lengths, bool batch_first) {
+    return at::_ops::_pack_padded_sequence::call(input, lengths, batch_first);
+}
+
+// aten::_pack_padded_sequence.out(Tensor input, Tensor lengths, bool batch_first, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _pack_padded_sequence_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & input, const at::Tensor & lengths, bool batch_first) {
+    return at::_ops::_pack_padded_sequence_out::call(input, lengths, batch_first, out0, out1);
+}
+// aten::_pack_padded_sequence.out(Tensor input, Tensor lengths, bool batch_first, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _pack_padded_sequence_outf(const at::Tensor & input, const at::Tensor & lengths, bool batch_first, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::_pack_padded_sequence_out::call(input, lengths, batch_first, out0, out1);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_pack_padded_sequence_backward_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_pack_padded_sequence_backward_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..faecc96ae809434eb3ec6514a70c64fbb2ef803c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_pack_padded_sequence_backward_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 _pack_padded_sequence_backward(const at::Tensor & grad, at::IntArrayRef input_size, const at::Tensor & batch_sizes, bool batch_first);
+TORCH_API at::Tensor _pack_padded_sequence_backward_symint(const at::Tensor & grad, c10::SymIntArrayRef input_size, const at::Tensor & batch_sizes, bool batch_first);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_print_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_print_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c956ce8abf198faa36c4d2427e9de99468b9db08
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_print_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 void _print(c10::string_view s);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_print_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_print_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..d7091acab41803c06cc607a8e9f5fd86edca4451
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_print_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 _print {
+  using schema = void (c10::string_view);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_print";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_print(str s) -> ()";
+  static void call(c10::string_view s);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, c10::string_view s);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_reshape_alias.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_reshape_alias.h
new file mode 100644
index 0000000000000000000000000000000000000000..4509572719dc5af9c9b62bde4c36291c4f594cec
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_reshape_alias.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/_reshape_alias_ops.h>
+
+namespace at {
+
+
+// aten::_reshape_alias(Tensor(a) self, SymInt[] size, SymInt[] stride) -> Tensor(a)
+inline at::Tensor _reshape_alias(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride) {
+    return at::_ops::_reshape_alias::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride));
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor _reshape_alias(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride) {
+    return at::_ops::_reshape_alias::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride));
+  }
+}
+
+// aten::_reshape_alias(Tensor(a) self, SymInt[] size, SymInt[] stride) -> Tensor(a)
+inline at::Tensor _reshape_alias_symint(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride) {
+    return at::_ops::_reshape_alias::call(self, size, stride);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor _reshape_alias(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride) {
+    return at::_ops::_reshape_alias::call(self, size, stride);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_reshape_alias_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_reshape_alias_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..4407d9c96c0743af80e4bee7b8c30cc36c98bb11
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_reshape_alias_copy.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_reshape_alias_copy_ops.h>
+
+namespace at {
+
+
+// aten::_reshape_alias_copy(Tensor self, SymInt[] size, SymInt[] stride) -> Tensor
+inline at::Tensor _reshape_alias_copy(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride) {
+    return at::_ops::_reshape_alias_copy::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride));
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor _reshape_alias_copy(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride) {
+    return at::_ops::_reshape_alias_copy::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride));
+  }
+}
+
+// aten::_reshape_alias_copy(Tensor self, SymInt[] size, SymInt[] stride) -> Tensor
+inline at::Tensor _reshape_alias_copy_symint(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride) {
+    return at::_ops::_reshape_alias_copy::call(self, size, stride);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor _reshape_alias_copy(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride) {
+    return at::_ops::_reshape_alias_copy::call(self, size, stride);
+  }
+}
+
+// aten::_reshape_alias_copy.out(Tensor self, SymInt[] size, SymInt[] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _reshape_alias_copy_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride) {
+    return at::_ops::_reshape_alias_copy_out::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride), out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & _reshape_alias_copy_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride) {
+    return at::_ops::_reshape_alias_copy_out::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride), out);
+  }
+}
+
+// aten::_reshape_alias_copy.out(Tensor self, SymInt[] size, SymInt[] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _reshape_alias_copy_outf(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride, at::Tensor & out) {
+    return at::_ops::_reshape_alias_copy_out::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride), out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & _reshape_alias_copy_outf(const at::Tensor & self, at::IntArrayRef size, at::IntArrayRef stride, at::Tensor & out) {
+    return at::_ops::_reshape_alias_copy_out::call(self, c10::fromIntArrayRefSlow(size), c10::fromIntArrayRefSlow(stride), out);
+  }
+}
+
+// aten::_reshape_alias_copy.out(Tensor self, SymInt[] size, SymInt[] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _reshape_alias_copy_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride) {
+    return at::_ops::_reshape_alias_copy_out::call(self, size, stride, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & _reshape_alias_copy_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride) {
+    return at::_ops::_reshape_alias_copy_out::call(self, size, stride, out);
+  }
+}
+
+// aten::_reshape_alias_copy.out(Tensor self, SymInt[] size, SymInt[] stride, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _reshape_alias_copy_symint_outf(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, at::Tensor & out) {
+    return at::_ops::_reshape_alias_copy_out::call(self, size, stride, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & _reshape_alias_copy_outf(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, at::Tensor & out) {
+    return at::_ops::_reshape_alias_copy_out::call(self, size, stride, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math.h
new file mode 100644
index 0000000000000000000000000000000000000000..554e49088ad3ac5a1a9255ac9d6a3a4068eb892c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math.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_attention_math_ops.h>
+
+namespace at {
+
+
+// aten::_scaled_dot_product_attention_math(Tensor query, Tensor key, Tensor value, Tensor? attn_mask=None, float dropout_p=0.0, bool is_causal=False, Tensor? dropout_mask=None, *, float? scale=None, bool enable_gqa=False) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _scaled_dot_product_attention_math(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, const ::std::optional<at::Tensor> & dropout_mask={}, ::std::optional<double> scale=::std::nullopt, bool enable_gqa=false) {
+    return at::_ops::_scaled_dot_product_attention_math::call(query, key, value, attn_mask, dropout_p, is_causal, dropout_mask, scale, enable_gqa);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..531ce1c3323b7e4ac86f93949865e302037bd746
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_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> _scaled_dot_product_attention_math(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, const ::std::optional<at::Tensor> & dropout_mask={}, ::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/_scaled_dot_product_efficient_attention_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_efficient_attention_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..8c26501c211fc847e3912c44a8286c449a878cfd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_efficient_attention_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _scaled_dot_product_efficient_attention {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, bool, double, bool, ::std::optional<double>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_scaled_dot_product_efficient_attention";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_scaled_dot_product_efficient_attention(Tensor query, Tensor key, Tensor value, Tensor? attn_bias, bool compute_log_sumexp, float dropout_p=0.0, bool is_causal=False, *, float? scale=None) -> (Tensor output, Tensor log_sumexp, Tensor philox_seed, Tensor philox_offset)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_bias, bool compute_log_sumexp, double dropout_p, bool is_causal, ::std::optional<double> scale);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_bias, bool compute_log_sumexp, double dropout_p, bool is_causal, ::std::optional<double> scale);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_flash_attention.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_flash_attention.h
new file mode 100644
index 0000000000000000000000000000000000000000..13b81f4a7dfce0fea842640fb6693e63b414b8ed
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_flash_attention.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_flash_attention_ops.h>
+
+namespace at {
+
+
+// aten::_scaled_dot_product_flash_attention(Tensor query, Tensor key, Tensor value, float dropout_p=0.0, bool is_causal=False, bool return_debug_mask=False, *, float? scale=None) -> (Tensor output, Tensor logsumexp, Tensor cum_seq_q, Tensor cum_seq_k, SymInt max_q, SymInt max_k, Tensor philox_seed, Tensor philox_offset, Tensor debug_attn_mask)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt,at::Tensor,at::Tensor,at::Tensor> _scaled_dot_product_flash_attention(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, double dropout_p=0.0, bool is_causal=false, bool return_debug_mask=false, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_flash_attention::call(query, key, value, dropout_p, is_causal, return_debug_mask, scale);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_segment_reduce_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_segment_reduce_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..d39c880d222533576a9990162e87164987e77a47
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_segment_reduce_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/_segment_reduce_backward_ops.h>
+
+namespace at {
+
+
+// aten::_segment_reduce_backward(Tensor grad, Tensor output, Tensor data, str reduce, *, Tensor? lengths=None, Tensor? offsets=None, int axis=0, Scalar? initial=None) -> Tensor
+inline at::Tensor _segment_reduce_backward(const at::Tensor & grad, const at::Tensor & output, const at::Tensor & data, c10::string_view reduce, const ::std::optional<at::Tensor> & lengths={}, const ::std::optional<at::Tensor> & offsets={}, int64_t axis=0, const ::std::optional<at::Scalar> & initial=::std::nullopt) {
+    return at::_ops::_segment_reduce_backward::call(grad, output, data, reduce, lengths, offsets, axis, initial);
+}
+
+// aten::_segment_reduce_backward.out(Tensor grad, Tensor output, Tensor data, str reduce, *, Tensor? lengths=None, Tensor? offsets=None, int axis=0, Scalar? initial=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _segment_reduce_backward_out(at::Tensor & out, const at::Tensor & grad, const at::Tensor & output, const at::Tensor & data, c10::string_view reduce, const ::std::optional<at::Tensor> & lengths={}, const ::std::optional<at::Tensor> & offsets={}, int64_t axis=0, const ::std::optional<at::Scalar> & initial=::std::nullopt) {
+    return at::_ops::_segment_reduce_backward_out::call(grad, output, data, reduce, lengths, offsets, axis, initial, out);
+}
+// aten::_segment_reduce_backward.out(Tensor grad, Tensor output, Tensor data, str reduce, *, Tensor? lengths=None, Tensor? offsets=None, int axis=0, Scalar? initial=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _segment_reduce_backward_outf(const at::Tensor & grad, const at::Tensor & output, const at::Tensor & data, c10::string_view reduce, const ::std::optional<at::Tensor> & lengths, const ::std::optional<at::Tensor> & offsets, int64_t axis, const ::std::optional<at::Scalar> & initial, at::Tensor & out) {
+    return at::_ops::_segment_reduce_backward_out::call(grad, output, data, reduce, lengths, offsets, axis, initial, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_segment_reduce_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_segment_reduce_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5b58e45b49184badfc26ac658282ac17d120f57f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_segment_reduce_backward_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor _segment_reduce_backward(const at::Tensor & grad, const at::Tensor & output, const at::Tensor & data, c10::string_view reduce, const ::std::optional<at::Tensor> & lengths={}, const ::std::optional<at::Tensor> & offsets={}, int64_t axis=0, const ::std::optional<at::Scalar> & initial=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_slow_conv2d_backward_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_slow_conv2d_backward_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6844c06089055c7a9d828963dd67cf7c01cee983
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_slow_conv2d_backward_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _slow_conv2d_backward_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _slow_conv2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _slow_conv2d_backward_symint_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _slow_conv2d_backward_symint_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_draw_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_draw_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3b1d51c88ce564156b690bc5131cca98ec083d53
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_draw_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> _sobol_engine_draw(const at::Tensor & quasi, int64_t n, const at::Tensor & sobolstate, int64_t dimension, int64_t num_generated, ::std::optional<at::ScalarType> dtype);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_draw_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_draw_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9904118a3bdf71d5055be1cdbf24a1feb7cd9a8b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sobol_engine_draw_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_draw {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, int64_t, const at::Tensor &, int64_t, int64_t, ::std::optional<at::ScalarType>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sobol_engine_draw";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_sobol_engine_draw(Tensor quasi, int n, Tensor sobolstate, int dimension, int num_generated, ScalarType? dtype) -> (Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & quasi, int64_t n, const at::Tensor & sobolstate, int64_t dimension, int64_t num_generated, ::std::optional<at::ScalarType> dtype);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & quasi, int64_t n, const at::Tensor & sobolstate, int64_t dimension, int64_t num_generated, ::std::optional<at::ScalarType> dtype);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_softmax_backward_data_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_softmax_backward_data_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..90c9b73e02689ec145f196305bc143f58797362d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_softmax_backward_data_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 _softmax_backward_data(const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, at::ScalarType input_dtype);
+TORCH_API at::Tensor & _softmax_backward_data_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, at::ScalarType input_dtype);
+TORCH_API at::Tensor & _softmax_backward_data_outf(const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, at::ScalarType input_dtype, at::Tensor & grad_input);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_broadcast_to_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_broadcast_to_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..bc2eae8ed53fc7e80a3f0a20d08490c3fd3f94cb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_broadcast_to_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor sparse_broadcast_to(const at::Tensor & self, at::IntArrayRef size);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_bsr_tensor_unsafe.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_bsr_tensor_unsafe.h
new file mode 100644
index 0000000000000000000000000000000000000000..104ccac7d2e0ac8d60e875c80ac555dbf6f570f7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_bsr_tensor_unsafe.h
@@ -0,0 +1,34 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_sparse_bsr_tensor_unsafe_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_bsr_tensor_unsafe(Tensor crow_indices, Tensor col_indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor _sparse_bsr_tensor_unsafe(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, at::IntArrayRef size, at::TensorOptions options={}) {
+    return at::_ops::_sparse_bsr_tensor_unsafe::call(crow_indices, col_indices, values, size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::_sparse_bsr_tensor_unsafe(Tensor crow_indices, Tensor col_indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor _sparse_bsr_tensor_unsafe(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::_sparse_bsr_tensor_unsafe::call(crow_indices, col_indices, values, size, dtype, layout, device, pin_memory);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_compressed_tensor_with_dims_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_compressed_tensor_with_dims_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8691072344de61b11c3f58a00befb2c5e67f630b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_compressed_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_compressed_tensor_with_dims(int64_t nnz, int64_t dense_dim, at::IntArrayRef size, at::IntArrayRef blocksize, at::ScalarType index_dtype, at::TensorOptions options);
+TORCH_API at::Tensor _sparse_compressed_tensor_with_dims(int64_t nnz, int64_t dense_dim, at::IntArrayRef size, at::IntArrayRef blocksize, at::ScalarType index_dtype, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm.h
new file mode 100644
index 0000000000000000000000000000000000000000..05167846116363b3d6554cd21ec7f8e067e2277e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm.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/_sparse_mm_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_mm(Tensor sparse, Tensor dense) -> Tensor
+inline at::Tensor _sparse_mm(const at::Tensor & sparse, const at::Tensor & dense) {
+    return at::_ops::_sparse_mm::call(sparse, dense);
+}
+
+// aten::_sparse_mm.reduce(Tensor sparse, Tensor dense, str reduce) -> Tensor
+inline at::Tensor _sparse_mm(const at::Tensor & sparse, const at::Tensor & dense, c10::string_view reduce) {
+    return at::_ops::_sparse_mm_reduce::call(sparse, dense, reduce);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_sum_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_sum_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..5ee633375d871f916e015c1d804ca0908f345d27
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_sum_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/_sparse_sum_backward_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_sum_backward(Tensor grad, Tensor self, int[] dim) -> Tensor
+inline at::Tensor _sparse_sum_backward(const at::Tensor & grad, const at::Tensor & self, at::IntArrayRef dim) {
+    return at::_ops::_sparse_sum_backward::call(grad, self, dim);
+}
+
+// aten::_sparse_sum_backward.out(Tensor grad, Tensor self, int[] dim, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_sum_backward_out(at::Tensor & out, const at::Tensor & grad, const at::Tensor & self, at::IntArrayRef dim) {
+    return at::_ops::_sparse_sum_backward_out::call(grad, self, dim, out);
+}
+// aten::_sparse_sum_backward.out(Tensor grad, Tensor self, int[] dim, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_sum_backward_outf(const at::Tensor & grad, const at::Tensor & self, at::IntArrayRef dim, at::Tensor & out) {
+    return at::_ops::_sparse_sum_backward_out::call(grad, self, dim, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_stack_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_stack_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..e2a5b713e7778f197a70b42be3c7e11cb3886310
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_stack_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 _stack {
+  using schema = at::Tensor (at::TensorList, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_stack";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_stack(Tensor[] tensors, int dim=0) -> Tensor";
+  static at::Tensor call(at::TensorList tensors, int64_t dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList tensors, int64_t dim);
+};
+
+struct TORCH_API _stack_out {
+  using schema = at::Tensor & (at::TensorList, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_stack";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_stack.out(Tensor[] tensors, int dim=0, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(at::TensorList tensors, int64_t dim, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList tensors, int64_t dim, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_standard_gamma_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_standard_gamma_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..2580f40b980a24cd15971a8a7f2257f8b93a697b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_standard_gamma_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 _standard_gamma {
+  using schema = at::Tensor (const at::Tensor &, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_standard_gamma";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_standard_gamma(Tensor self, Generator? generator=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, ::std::optional<at::Generator> generator);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, ::std::optional<at::Generator> generator);
+};
+
+struct TORCH_API _standard_gamma_out {
+  using schema = at::Tensor & (const at::Tensor &, ::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::_standard_gamma";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_standard_gamma.out(Tensor self, Generator? generator=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, ::std::optional<at::Generator> generator, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, ::std::optional<at::Generator> generator, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0672af2c7d3653a9824658c6c5d5fb23014f12b3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch.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/_test_autograd_multiple_dispatch_ops.h>
+
+namespace at {
+
+
+// aten::_test_autograd_multiple_dispatch.fullcoverage(Tensor self) -> Tensor
+inline at::Tensor _test_autograd_multiple_dispatch(const at::Tensor & self) {
+    return at::_ops::_test_autograd_multiple_dispatch_fullcoverage::call(self);
+}
+
+// aten::_test_autograd_multiple_dispatch.ntonly(Tensor self, bool b) -> Tensor
+inline at::Tensor _test_autograd_multiple_dispatch(const at::Tensor & self, bool b) {
+    return at::_ops::_test_autograd_multiple_dispatch_ntonly::call(self, b);
+}
+
+// aten::_test_autograd_multiple_dispatch.fullcoverage_out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _test_autograd_multiple_dispatch_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::_test_autograd_multiple_dispatch_fullcoverage_out::call(self, out);
+}
+// aten::_test_autograd_multiple_dispatch.fullcoverage_out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _test_autograd_multiple_dispatch_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::_test_autograd_multiple_dispatch_fullcoverage_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch_view_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch_view_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..f6e872a4aa837b7bcd731980833ae579426b8ec8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_autograd_multiple_dispatch_view_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor _test_autograd_multiple_dispatch_view(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_functorch_fallback_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_functorch_fallback_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..ff283c74984129f99992411cc3d0d4ab0e701a12
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_functorch_fallback_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_functorch_fallback {
+  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::_test_functorch_fallback";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_test_functorch_fallback(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 _test_functorch_fallback_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::_test_functorch_fallback";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_test_functorch_fallback.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_filled_intlist_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_filled_intlist_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..bd14fb9bdf65650e772ede370fa91dd38be217ee
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_filled_intlist_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_filled_intlist {
+  using schema = at::Tensor (const at::Tensor &, at::OptionalIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_test_optional_filled_intlist";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_test_optional_filled_intlist(Tensor values, int[2]? addends) -> Tensor";
+  static at::Tensor call(const at::Tensor & values, at::OptionalIntArrayRef addends);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & values, at::OptionalIntArrayRef addends);
+};
+
+struct TORCH_API _test_optional_filled_intlist_out {
+  using schema = at::Tensor & (const at::Tensor &, at::OptionalIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_test_optional_filled_intlist";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_test_optional_filled_intlist.out(Tensor values, int[2]? addends, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & values, at::OptionalIntArrayRef addends, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & values, at::OptionalIntArrayRef addends, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_intlist_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_intlist_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b32cd22a77a6511f6994fa19fc647414e9cf32d1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_test_optional_intlist_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 _test_optional_intlist(const at::Tensor & values, at::OptionalIntArrayRef addends);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_gru_cell_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_gru_cell_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e83ae863b90eba0ed17e59f5bc13a8881cf414e2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_gru_cell_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> _thnn_fused_gru_cell(const at::Tensor & input_gates, const at::Tensor & hidden_gates, const at::Tensor & hx, const ::std::optional<at::Tensor> & input_bias={}, const ::std::optional<at::Tensor> & hidden_bias={});
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_impl_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_impl_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2c6e6d7a6210f485c04ef2c6e017e420b65085dc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_impl_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 &> _thnn_fused_lstm_cell_backward_impl_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, 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);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _thnn_fused_lstm_cell_backward_impl_outf(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, 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/_to_sparse_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4d13f60fa94538d837ef4b87508a6ce9573d78ed
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_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 & _to_sparse_out(at::Tensor & out, const at::Tensor & self, int64_t sparse_dim);
+TORCH_API at::Tensor & _to_sparse_outf(const at::Tensor & self, int64_t sparse_dim, at::Tensor & out);
+TORCH_API at::Tensor & _to_sparse_out(at::Tensor & out, const at::Tensor & self, ::std::optional<at::Layout> layout=::std::nullopt, at::OptionalIntArrayRef blocksize=::std::nullopt, ::std::optional<int64_t> dense_dim=::std::nullopt);
+TORCH_API at::Tensor & _to_sparse_outf(const at::Tensor & self, ::std::optional<at::Layout> layout, at::OptionalIntArrayRef 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/_transformer_encoder_layer_fwd_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_transformer_encoder_layer_fwd_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f57c06bfd59ac248fa8ba3e178cb5e7a8b6afeea
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_transformer_encoder_layer_fwd_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 _transformer_encoder_layer_fwd(const at::Tensor & src, int64_t embed_dim, int64_t num_heads, const at::Tensor & qkv_weight, const at::Tensor & qkv_bias, const at::Tensor & proj_weight, const at::Tensor & proj_bias, bool use_gelu, bool norm_first, double eps, const at::Tensor & norm_weight_1, const at::Tensor & norm_bias_1, const at::Tensor & norm_weight_2, const at::Tensor & norm_bias_2, const at::Tensor & ffn_weight_1, const at::Tensor & ffn_bias_1, const at::Tensor & ffn_weight_2, const at::Tensor & ffn_bias_2, const ::std::optional<at::Tensor> & mask={}, ::std::optional<int64_t> mask_type=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_transformer_encoder_layer_fwd_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_transformer_encoder_layer_fwd_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c33179ad4cf0257337254666ebda06a4dc0a26ba
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_transformer_encoder_layer_fwd_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 & _transformer_encoder_layer_fwd_out(const at::Tensor & src, int64_t embed_dim, int64_t num_heads, const at::Tensor & qkv_weight, const at::Tensor & qkv_bias, const at::Tensor & proj_weight, const at::Tensor & proj_bias, bool use_gelu, bool norm_first, double eps, const at::Tensor & norm_weight_1, const at::Tensor & norm_bias_1, const at::Tensor & norm_weight_2, const at::Tensor & norm_bias_2, const at::Tensor & ffn_weight_1, const at::Tensor & ffn_bias_1, const at::Tensor & ffn_weight_2, const at::Tensor & ffn_bias_2, const ::std::optional<at::Tensor> & mask, ::std::optional<int64_t> mask_type, at::Tensor & out);
+TORCH_API at::Tensor transformer_encoder_layer_forward(const at::Tensor & src, int64_t embed_dim, int64_t num_heads, const at::Tensor & qkv_weight, const at::Tensor & qkv_bias, const at::Tensor & proj_weight, const at::Tensor & proj_bias, bool use_gelu, bool norm_first, double eps, const at::Tensor & norm_weight_1, const at::Tensor & norm_bias_1, const at::Tensor & norm_weight_2, const at::Tensor & norm_bias_2, const at::Tensor & ffn_weight_1, const at::Tensor & ffn_bias_1, const at::Tensor & ffn_weight_2, const at::Tensor & ffn_bias_2, const ::std::optional<at::Tensor> & mask={}, ::std::optional<int64_t> mask_type=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_transformer_encoder_layer_fwd_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_transformer_encoder_layer_fwd_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..d83c7082bc649e7074c08b78a8c6f5890c97e7d1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_transformer_encoder_layer_fwd_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 _transformer_encoder_layer_fwd {
+  using schema = at::Tensor (const at::Tensor &, int64_t, int64_t, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, bool, bool, double, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, ::std::optional<int64_t>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_transformer_encoder_layer_fwd";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_transformer_encoder_layer_fwd(Tensor src, int embed_dim, int num_heads, Tensor qkv_weight, Tensor qkv_bias, Tensor proj_weight, Tensor proj_bias, bool use_gelu, bool norm_first, float eps, Tensor norm_weight_1, Tensor norm_bias_1, Tensor norm_weight_2, Tensor norm_bias_2, Tensor ffn_weight_1, Tensor ffn_bias_1, Tensor ffn_weight_2, Tensor ffn_bias_2, Tensor? mask=None, int? mask_type=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & src, int64_t embed_dim, int64_t num_heads, const at::Tensor & qkv_weight, const at::Tensor & qkv_bias, const at::Tensor & proj_weight, const at::Tensor & proj_bias, bool use_gelu, bool norm_first, double eps, const at::Tensor & norm_weight_1, const at::Tensor & norm_bias_1, const at::Tensor & norm_weight_2, const at::Tensor & norm_bias_2, const at::Tensor & ffn_weight_1, const at::Tensor & ffn_bias_1, const at::Tensor & ffn_weight_2, const at::Tensor & ffn_bias_2, const ::std::optional<at::Tensor> & mask, ::std::optional<int64_t> mask_type);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & src, int64_t embed_dim, int64_t num_heads, const at::Tensor & qkv_weight, const at::Tensor & qkv_bias, const at::Tensor & proj_weight, const at::Tensor & proj_bias, bool use_gelu, bool norm_first, double eps, const at::Tensor & norm_weight_1, const at::Tensor & norm_bias_1, const at::Tensor & norm_weight_2, const at::Tensor & norm_bias_2, const at::Tensor & ffn_weight_1, const at::Tensor & ffn_bias_1, const at::Tensor & ffn_weight_2, const at::Tensor & ffn_bias_2, const ::std::optional<at::Tensor> & mask, ::std::optional<int64_t> mask_type);
+};
+
+struct TORCH_API _transformer_encoder_layer_fwd_out {
+  using schema = at::Tensor & (const at::Tensor &, int64_t, int64_t, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, bool, bool, double, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, ::std::optional<int64_t>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_transformer_encoder_layer_fwd";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_transformer_encoder_layer_fwd.out(Tensor src, int embed_dim, int num_heads, Tensor qkv_weight, Tensor qkv_bias, Tensor proj_weight, Tensor proj_bias, bool use_gelu, bool norm_first, float eps, Tensor norm_weight_1, Tensor norm_bias_1, Tensor norm_weight_2, Tensor norm_bias_2, Tensor ffn_weight_1, Tensor ffn_bias_1, Tensor ffn_weight_2, Tensor ffn_bias_2, Tensor? mask=None, int? mask_type=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & src, int64_t embed_dim, int64_t num_heads, const at::Tensor & qkv_weight, const at::Tensor & qkv_bias, const at::Tensor & proj_weight, const at::Tensor & proj_bias, bool use_gelu, bool norm_first, double eps, const at::Tensor & norm_weight_1, const at::Tensor & norm_bias_1, const at::Tensor & norm_weight_2, const at::Tensor & norm_bias_2, const at::Tensor & ffn_weight_1, const at::Tensor & ffn_bias_1, const at::Tensor & ffn_weight_2, const at::Tensor & ffn_bias_2, const ::std::optional<at::Tensor> & mask, ::std::optional<int64_t> mask_type, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & src, int64_t embed_dim, int64_t num_heads, const at::Tensor & qkv_weight, const at::Tensor & qkv_bias, const at::Tensor & proj_weight, const at::Tensor & proj_bias, bool use_gelu, bool norm_first, double eps, const at::Tensor & norm_weight_1, const at::Tensor & norm_bias_1, const at::Tensor & norm_weight_2, const at::Tensor & norm_bias_2, const at::Tensor & ffn_weight_1, const at::Tensor & ffn_bias_1, const at::Tensor & ffn_weight_2, const at::Tensor & ffn_bias_2, const ::std::optional<at::Tensor> & mask, ::std::optional<int64_t> mask_type, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_trilinear.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_trilinear.h
new file mode 100644
index 0000000000000000000000000000000000000000..6942f26f43d8c15d7f1bd80d8165fe056556b969
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_trilinear.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/_trilinear_ops.h>
+
+namespace at {
+
+
+// aten::_trilinear(Tensor i1, Tensor i2, Tensor i3, int[] expand1, int[] expand2, int[] expand3, int[] sumdim, int unroll_dim=1) -> Tensor
+inline at::Tensor _trilinear(const at::Tensor & i1, const at::Tensor & i2, const at::Tensor & i3, at::IntArrayRef expand1, at::IntArrayRef expand2, at::IntArrayRef expand3, at::IntArrayRef sumdim, int64_t unroll_dim=1) {
+    return at::_ops::_trilinear::call(i1, i2, i3, expand1, expand2, expand3, sumdim, unroll_dim);
+}
+
+// aten::_trilinear.out(Tensor i1, Tensor i2, Tensor i3, int[] expand1, int[] expand2, int[] expand3, int[] sumdim, int unroll_dim=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _trilinear_out(at::Tensor & out, const at::Tensor & i1, const at::Tensor & i2, const at::Tensor & i3, at::IntArrayRef expand1, at::IntArrayRef expand2, at::IntArrayRef expand3, at::IntArrayRef sumdim, int64_t unroll_dim=1) {
+    return at::_ops::_trilinear_out::call(i1, i2, i3, expand1, expand2, expand3, sumdim, unroll_dim, out);
+}
+// aten::_trilinear.out(Tensor i1, Tensor i2, Tensor i3, int[] expand1, int[] expand2, int[] expand3, int[] sumdim, int unroll_dim=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _trilinear_outf(const at::Tensor & i1, const at::Tensor & i2, const at::Tensor & i3, at::IntArrayRef expand1, at::IntArrayRef expand2, at::IntArrayRef expand3, at::IntArrayRef sumdim, int64_t unroll_dim, at::Tensor & out) {
+    return at::_ops::_trilinear_out::call(i1, i2, i3, expand1, expand2, expand3, sumdim, unroll_dim, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_unique2_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_unique2_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..411eb56ba422e990cb56314f0b04bbcac97f43ed
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_unique2_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 &,at::Tensor &> _unique2_out(const at::Tensor & self, bool sorted, bool return_inverse, bool return_counts, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _unique2_cpu(const at::Tensor & self, bool sorted=true, bool return_inverse=false, bool return_counts=false);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _unique2_cuda(const at::Tensor & self, bool sorted=true, bool return_inverse=false, bool return_counts=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_unpack_dual.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_unpack_dual.h
new file mode 100644
index 0000000000000000000000000000000000000000..47ac084248ef3c51e2289d40b25c2da6f36f621b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_unpack_dual.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/_unpack_dual_ops.h>
+
+namespace at {
+
+
+// aten::_unpack_dual(Tensor(a) dual, int level) -> (Tensor(a) primal, Tensor tangent)
+inline ::std::tuple<at::Tensor,at::Tensor> _unpack_dual(const at::Tensor & dual, int64_t level) {
+    return at::_ops::_unpack_dual::call(dual, level);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f59253f640e54b1f0187cf11acc46f070bff1a9c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_backward_cpu_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor _upsample_nearest_exact1d_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_nearest_exact1d_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_nearest_exact1d_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_nearest_exact1d_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_nearest_exact1d_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_nearest_exact1d_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 cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact2d_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact2d_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a20631e1888ce0e6c409f24451cbbc2e3eb76b9b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact2d_cpu_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor _upsample_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 cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact3d_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact3d_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..be75b46f0860d1d04518119e0ca5e2fd34c72a4c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact3d_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/_upsample_nearest_exact3d_backward_ops.h>
+
+namespace at {
+
+
+// aten::_upsample_nearest_exact3d_backward.grad_input(Tensor grad_output, SymInt[3] output_size, SymInt[5] input_size, float? scales_d=None, float? scales_h=None, float? scales_w=None, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & _upsample_nearest_exact3d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_nearest_exact3d_backward_grad_input::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), scales_d, scales_h, scales_w, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & _upsample_nearest_exact3d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_nearest_exact3d_backward_grad_input::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), scales_d, scales_h, scales_w, grad_input);
+  }
+}
+
+// aten::_upsample_nearest_exact3d_backward.grad_input(Tensor grad_output, SymInt[3] output_size, SymInt[5] input_size, float? scales_d=None, float? scales_h=None, float? scales_w=None, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & _upsample_nearest_exact3d_backward_outf(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input) {
+    return at::_ops::_upsample_nearest_exact3d_backward_grad_input::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), scales_d, scales_h, scales_w, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & _upsample_nearest_exact3d_backward_outf(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input) {
+    return at::_ops::_upsample_nearest_exact3d_backward_grad_input::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), scales_d, scales_h, scales_w, grad_input);
+  }
+}
+
+// aten::_upsample_nearest_exact3d_backward.grad_input(Tensor grad_output, SymInt[3] output_size, SymInt[5] input_size, float? scales_d=None, float? scales_h=None, float? scales_w=None, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & _upsample_nearest_exact3d_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_nearest_exact3d_backward_grad_input::call(grad_output, output_size, input_size, scales_d, scales_h, scales_w, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & _upsample_nearest_exact3d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_nearest_exact3d_backward_grad_input::call(grad_output, output_size, input_size, scales_d, scales_h, scales_w, grad_input);
+  }
+}
+
+// aten::_upsample_nearest_exact3d_backward.grad_input(Tensor grad_output, SymInt[3] output_size, SymInt[5] input_size, float? scales_d=None, float? scales_h=None, float? scales_w=None, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & _upsample_nearest_exact3d_backward_symint_outf(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input) {
+    return at::_ops::_upsample_nearest_exact3d_backward_grad_input::call(grad_output, output_size, input_size, scales_d, scales_h, scales_w, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & _upsample_nearest_exact3d_backward_outf(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input) {
+    return at::_ops::_upsample_nearest_exact3d_backward_grad_input::call(grad_output, output_size, input_size, scales_d, scales_h, scales_w, grad_input);
+  }
+}
+
+// aten::_upsample_nearest_exact3d_backward(Tensor grad_output, SymInt[3] output_size, SymInt[5] input_size, float? scales_d=None, float? scales_h=None, float? scales_w=None) -> Tensor
+inline at::Tensor _upsample_nearest_exact3d_backward(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_nearest_exact3d_backward::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), scales_d, scales_h, scales_w);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor _upsample_nearest_exact3d_backward(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_nearest_exact3d_backward::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), scales_d, scales_h, scales_w);
+  }
+}
+
+// aten::_upsample_nearest_exact3d_backward(Tensor grad_output, SymInt[3] output_size, SymInt[5] input_size, float? scales_d=None, float? scales_h=None, float? scales_w=None) -> Tensor
+inline at::Tensor _upsample_nearest_exact3d_backward_symint(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_nearest_exact3d_backward::call(grad_output, output_size, input_size, scales_d, scales_h, scales_w);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor _upsample_nearest_exact3d_backward(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_nearest_exact3d_backward::call(grad_output, output_size, input_size, scales_d, scales_h, scales_w);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_compressed_sparse_indices.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_compressed_sparse_indices.h
new file mode 100644
index 0000000000000000000000000000000000000000..fb2c9538a517042ca00dfafcf90d003d7f1daf6d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_compressed_sparse_indices.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/_validate_compressed_sparse_indices_ops.h>
+
+namespace at {
+
+
+// aten::_validate_compressed_sparse_indices(bool is_crow, Tensor compressed_idx, Tensor plain_idx, int cdim, int dim, int nnz) -> ()
+inline void _validate_compressed_sparse_indices(bool is_crow, const at::Tensor & compressed_idx, const at::Tensor & plain_idx, int64_t cdim, int64_t dim, int64_t nnz) {
+    return at::_ops::_validate_compressed_sparse_indices::call(is_crow, compressed_idx, plain_idx, cdim, dim, nnz);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_compressed_tensor_args_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_compressed_tensor_args_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c67b87f598cbfabfd646534ef974be8a02dedea3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_validate_sparse_compressed_tensor_args_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 void _validate_sparse_compressed_tensor_args(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, at::IntArrayRef size, at::Layout layout);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_differentiable_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_differentiable_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..434db4f8bdd75459a08fd3e9ebc8a202c3a89bef
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_differentiable_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> _weight_norm_differentiable_backward(const at::Tensor & grad_w, const at::Tensor & saved_v, const at::Tensor & saved_g, const at::Tensor & saved_norms, int64_t dim);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_backward_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_backward_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3f1af0abc5a4f70ea3dc04da71e70922a8c9ef47
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_backward_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> _weight_norm_interface_backward_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & grad_w, const at::Tensor & saved_v, const at::Tensor & saved_g, const at::Tensor & saved_norms, int64_t dim);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> _weight_norm_interface_backward_outf(const at::Tensor & grad_w, const at::Tensor & saved_v, const at::Tensor & saved_g, const at::Tensor & saved_norms, int64_t dim, at::Tensor & out0, at::Tensor & out1);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b8a54daf60841981383e559fdfbd50d8c49ef71a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_backward_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor adaptive_max_pool3d_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & indices);
+TORCH_API at::Tensor & adaptive_max_pool3d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & indices);
+TORCH_API at::Tensor & adaptive_max_pool3d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & indices, at::Tensor & grad_input);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_backward_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_backward_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..d87c3b38df111bc0866920d02973381a38c69438
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_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_adaptive_max_pool3d_backward : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & indices);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/add_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/add_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ce96b40c3dc3ece19c59c1be2ea5ebaed063678e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/add_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 add(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & add_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & add_outf(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha, at::Tensor & out);
+TORCH_API at::Tensor & add_(at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addbmm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addbmm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8ec26da7e68b21f5bdaefad422bdc9c9ad3e4afa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addbmm_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 addbmm(const at::Tensor & self, const at::Tensor & batch1, const at::Tensor & batch2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & addbmm_out(const at::Tensor & self, const at::Tensor & batch1, const at::Tensor & batch2, const at::Scalar & beta, const at::Scalar & alpha, at::Tensor & out);
+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 native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addmv_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addmv_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c1b3a1107a9e84f68c97ff2a4d95a5d5b7c1d609
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addmv_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 addmv(const at::Tensor & self, const at::Tensor & mat, const at::Tensor & vec, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & addmv_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & mat, const at::Tensor & vec, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & addmv_outf(const at::Tensor & self, const at::Tensor & mat, const at::Tensor & vec, const at::Scalar & beta, const at::Scalar & alpha, at::Tensor & out);
+TORCH_API at::Tensor & addmv_(at::Tensor & self, const at::Tensor & mat, const at::Tensor & vec, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/align_to_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/align_to_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..06bddcb5e223516726dd75a63d716714fcb17a5e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/align_to_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 align_to(const at::Tensor & self, at::DimnameList names);
+TORCH_API at::Tensor align_to(const at::Tensor & self, at::DimnameList order, int64_t ellipsis_idx);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/alpha_dropout.h b/lib/python3.12/site-packages/torch/include/ATen/ops/alpha_dropout.h
new file mode 100644
index 0000000000000000000000000000000000000000..55e9e061eea7ba1b943edb0c6b9282438fdcbeb4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/alpha_dropout.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/alpha_dropout_ops.h>
+
+namespace at {
+
+
+// aten::alpha_dropout(Tensor input, float p, bool train) -> Tensor
+inline at::Tensor alpha_dropout(const at::Tensor & input, double p, bool train) {
+    return at::_ops::alpha_dropout::call(input, p, train);
+}
+
+// aten::alpha_dropout_(Tensor(a!) self, float p, bool train) -> Tensor(a!)
+inline at::Tensor & alpha_dropout_(at::Tensor & self, double p, bool train) {
+    return at::_ops::alpha_dropout_::call(self, p, train);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/amax_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/amax_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..15333c3c84b3b1c4ecade8802c529033ae9a7697
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/amax_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 amax(const at::Tensor & self, at::IntArrayRef dim={}, bool keepdim=false);
+TORCH_API at::Tensor & amax_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim={}, bool keepdim=false);
+TORCH_API at::Tensor & amax_outf(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/and_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/and_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..adc2ea9f1dc3375395900d7061b0e529643d218c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/and_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 __and__(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & __iand__(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor __and__(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & __iand__(at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arccos.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arccos.h
new file mode 100644
index 0000000000000000000000000000000000000000..c7bb87a17fdcda558cb013c63c3824ff1124e526
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arccos.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/arccos_ops.h>
+
+namespace at {
+
+
+// aten::arccos(Tensor self) -> Tensor
+inline at::Tensor arccos(const at::Tensor & self) {
+    return at::_ops::arccos::call(self);
+}
+
+// aten::arccos_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & arccos_(at::Tensor & self) {
+    return at::_ops::arccos_::call(self);
+}
+
+// aten::arccos.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arccos_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::arccos_out::call(self, out);
+}
+// aten::arccos.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arccos_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::arccos_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arcsin.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arcsin.h
new file mode 100644
index 0000000000000000000000000000000000000000..d5e7ea54575db54f3f6a082650c53acddf9aa33e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arcsin.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/arcsin_ops.h>
+
+namespace at {
+
+
+// aten::arcsin(Tensor self) -> Tensor
+inline at::Tensor arcsin(const at::Tensor & self) {
+    return at::_ops::arcsin::call(self);
+}
+
+// aten::arcsin_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & arcsin_(at::Tensor & self) {
+    return at::_ops::arcsin_::call(self);
+}
+
+// aten::arcsin.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arcsin_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::arcsin_out::call(self, out);
+}
+// aten::arcsin.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arcsin_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::arcsin_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arctan2_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arctan2_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..65389006739b18b17e3c362c84a6172295e98005
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arctan2_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 arctan2 {
+  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::arctan2";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "arctan2(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 arctan2_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::arctan2";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "arctan2.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 arctan2_ {
+  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::arctan2_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "arctan2_(Tensor(a!) self, Tensor other) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & other);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & other);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arctanh_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arctanh_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..6b5bd42bf4e66278a70f5e8d0dd32504d56179bf
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arctanh_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 arctanh(const at::Tensor & self);
+TORCH_API at::Tensor & arctanh_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & arctanh_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/argmax.h b/lib/python3.12/site-packages/torch/include/ATen/ops/argmax.h
new file mode 100644
index 0000000000000000000000000000000000000000..d1709ab691e621c44c703365eacec6e023f8429f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/argmax.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/argmax_ops.h>
+
+namespace at {
+
+
+// aten::argmax(Tensor self, int? dim=None, bool keepdim=False) -> Tensor
+inline at::Tensor argmax(const at::Tensor & self, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false) {
+    return at::_ops::argmax::call(self, dim, keepdim);
+}
+
+// aten::argmax.out(Tensor self, int? dim=None, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & argmax_out(at::Tensor & out, const at::Tensor & self, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false) {
+    return at::_ops::argmax_out::call(self, dim, keepdim, out);
+}
+// aten::argmax.out(Tensor self, int? dim=None, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & argmax_outf(const at::Tensor & self, ::std::optional<int64_t> dim, bool keepdim, at::Tensor & out) {
+    return at::_ops::argmax_out::call(self, dim, keepdim, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/argmax_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/argmax_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..3d48bb1ec74cc7946d45939d47bfff16ab4ca955
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/argmax_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_argmax : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, ::std::optional<int64_t> dim, bool keepdim);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/argsort_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/argsort_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..560d0f2cd5e0a161d4639f80c8ce566d217dc496
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/argsort_compositeimplicitautograd_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 compositeimplicitautograd {
+
+TORCH_API at::Tensor argsort(const at::Tensor & self, int64_t dim=-1, bool descending=false);
+TORCH_API at::Tensor argsort(const at::Tensor & self, bool stable, int64_t dim=-1, bool descending=false);
+TORCH_API at::Tensor & argsort_out(at::Tensor & out, const at::Tensor & self, bool stable, int64_t dim=-1, bool descending=false);
+TORCH_API at::Tensor & argsort_outf(const at::Tensor & self, bool stable, int64_t dim, bool descending, at::Tensor & out);
+TORCH_API at::Tensor argsort(const at::Tensor & self, at::Dimname dim, bool descending=false);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/as_strided_scatter_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/as_strided_scatter_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..c44de6dd92d41fb4ac1b40cf2d360d5e726a7281
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/as_strided_scatter_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API as_strided_scatter {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, c10::SymIntArrayRef, c10::SymIntArrayRef, ::std::optional<c10::SymInt>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::as_strided_scatter";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "as_strided_scatter(Tensor self, Tensor src, SymInt[] size, SymInt[] stride, SymInt? storage_offset=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & src, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<c10::SymInt> storage_offset);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & src, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<c10::SymInt> storage_offset);
+};
+
+struct TORCH_API as_strided_scatter_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, c10::SymIntArrayRef, c10::SymIntArrayRef, ::std::optional<c10::SymInt>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::as_strided_scatter";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "as_strided_scatter.out(Tensor self, Tensor src, SymInt[] size, SymInt[] stride, SymInt? storage_offset=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & src, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<c10::SymInt> storage_offset, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & src, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<c10::SymInt> storage_offset, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/asinh_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/asinh_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..44e3ec7c5c7848bdc2e93db75911380a1a65c102
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/asinh_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 asinh(const at::Tensor & self);
+TORCH_API at::Tensor & asinh_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & asinh_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & asinh_(at::Tensor & self);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool2d_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool2d_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f2a0ac53d4712eb7693d65fb18abee131e63cca8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool2d_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 avg_pool2d(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_pool2d_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_pool2d_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 cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool3d_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool3d_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..4ae39947615552dbd4aa777ecc3c65b654fc68ab
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool3d_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/avg_pool3d_backward_ops.h>
+
+namespace at {
+
+
+// aten::avg_pool3d_backward.grad_input(Tensor grad_output, Tensor self, int[3] kernel_size, int[3] stride, int[3] padding, bool ceil_mode, bool count_include_pad, int? divisor_override, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & avg_pool3d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override) {
+    return at::_ops::avg_pool3d_backward_grad_input::call(grad_output, self, kernel_size, stride, padding, ceil_mode, count_include_pad, divisor_override, grad_input);
+}
+// aten::avg_pool3d_backward.grad_input(Tensor grad_output, Tensor self, int[3] kernel_size, int[3] stride, int[3] padding, bool ceil_mode, bool count_include_pad, int? divisor_override, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & avg_pool3d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override, at::Tensor & grad_input) {
+    return at::_ops::avg_pool3d_backward_grad_input::call(grad_output, self, kernel_size, stride, padding, ceil_mode, count_include_pad, divisor_override, grad_input);
+}
+
+// aten::avg_pool3d_backward(Tensor grad_output, Tensor self, int[3] kernel_size, int[3] stride, int[3] padding, bool ceil_mode, bool count_include_pad, int? divisor_override) -> Tensor
+inline at::Tensor avg_pool3d_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override) {
+    return at::_ops::avg_pool3d_backward::call(grad_output, self, kernel_size, stride, padding, ceil_mode, count_include_pad, divisor_override);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool3d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool3d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..3cdf1a3117062331b758869a36bdaa3e4667634d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool3d_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API avg_pool3d_out {
+  using schema = at::Tensor & (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, bool, ::std::optional<int64_t>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::avg_pool3d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "avg_pool3d.out(Tensor self, int[3] kernel_size, int[3] stride=[], int[3] padding=0, bool ceil_mode=False, bool count_include_pad=True, int? divisor_override=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(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);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+struct TORCH_API avg_pool3d {
+  using schema = at::Tensor (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, bool, ::std::optional<int64_t>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::avg_pool3d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "avg_pool3d(Tensor self, int[3] kernel_size, int[3] stride=[], int[3] padding=0, bool ceil_mode=False, bool count_include_pad=True, int? divisor_override=None) -> Tensor";
+  static at::Tensor call(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);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_elemt.h b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_elemt.h
new file mode 100644
index 0000000000000000000000000000000000000000..cf56886fd447b413adf1bac09d67e75c6a24bff2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_elemt.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/batch_norm_elemt_ops.h>
+
+namespace at {
+
+
+// aten::batch_norm_elemt(Tensor input, Tensor? weight, Tensor? bias, Tensor mean, Tensor invstd, float eps) -> Tensor
+inline at::Tensor batch_norm_elemt(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & mean, const at::Tensor & invstd, double eps) {
+    return at::_ops::batch_norm_elemt::call(input, weight, bias, mean, invstd, eps);
+}
+
+// aten::batch_norm_elemt.out(Tensor input, Tensor? weight, Tensor? bias, Tensor mean, Tensor invstd, float eps, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & batch_norm_elemt_out(at::Tensor & out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & mean, const at::Tensor & invstd, double eps) {
+    return at::_ops::batch_norm_elemt_out::call(input, weight, bias, mean, invstd, eps, out);
+}
+// aten::batch_norm_elemt.out(Tensor input, Tensor? weight, Tensor? bias, Tensor mean, Tensor invstd, float eps, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & batch_norm_elemt_outf(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & mean, const at::Tensor & invstd, double eps, at::Tensor & out) {
+    return at::_ops::batch_norm_elemt_out::call(input, weight, bias, mean, invstd, eps, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bilinear_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bilinear_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..72982391a915acbcefd93a573db180385653538a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bilinear_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 bilinear(const at::Tensor & input1, const at::Tensor & input2, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={});
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/binary_cross_entropy_with_logits_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/binary_cross_entropy_with_logits_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..bce6b86ce11e02596d401bd1542a325d6a831a7f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/binary_cross_entropy_with_logits_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 binary_cross_entropy_with_logits {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::binary_cross_entropy_with_logits";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "binary_cross_entropy_with_logits(Tensor self, Tensor target, Tensor? weight=None, Tensor? pos_weight=None, int reduction=Mean) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & pos_weight, int64_t reduction);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & pos_weight, int64_t reduction);
+};
+
+struct TORCH_API binary_cross_entropy_with_logits_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::binary_cross_entropy_with_logits";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "binary_cross_entropy_with_logits.out(Tensor self, Tensor target, Tensor? weight=None, Tensor? pos_weight=None, int reduction=Mean, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & pos_weight, int64_t reduction, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & pos_weight, int64_t reduction, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/binomial_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/binomial_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..2d9c6fcae13f4ac0cb4fa4d46cf09298bf6bcfa6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/binomial_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 & binomial_out(const at::Tensor & count, const at::Tensor & prob, ::std::optional<at::Generator> generator, at::Tensor & out);
+TORCH_API at::Tensor _s_binomial_cpu(const at::Tensor & count, const at::Tensor & prob, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor _s_binomial_cuda(const at::Tensor & count, const at::Tensor & prob, ::std::optional<at::Generator> generator=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_or_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_or_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..ccf036fb6714c61b3160f8dfd10db617701c3aba
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_or_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_bitwise_or_Tensor : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Tensor & other);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_xor_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_xor_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..667635e3ef5fd5980d07a491d9e20fb822e07ccc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_xor_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 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 cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/broadcast_tensors_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/broadcast_tensors_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0b1cef3bfb94db8d0cf24c245b3f99348a968e89
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/broadcast_tensors_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API ::std::vector<at::Tensor> broadcast_tensors(at::TensorList tensors);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/can_cast_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/can_cast_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a6c34a7e0dc998f75b8fc0ac67e5a13f6b9a67c8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/can_cast_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 bool can_cast(at::ScalarType from_, at::ScalarType to);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/can_cast_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/can_cast_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..1eb114eb3c0f4a2c408e6c13dd480e2fc7a4e66e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/can_cast_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 can_cast(at::ScalarType from_, at::ScalarType to);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/celu.h b/lib/python3.12/site-packages/torch/include/ATen/ops/celu.h
new file mode 100644
index 0000000000000000000000000000000000000000..578effaafbec0540ed0fd95b4a2b940f71951460
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/celu.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/celu_ops.h>
+
+namespace at {
+
+
+// aten::celu(Tensor self, Scalar alpha=1.0) -> Tensor
+inline at::Tensor celu(const at::Tensor & self, const at::Scalar & alpha=1.0) {
+    return at::_ops::celu::call(self, alpha);
+}
+
+// aten::celu_(Tensor(a!) self, Scalar alpha=1.0) -> Tensor(a!)
+inline at::Tensor & celu_(at::Tensor & self, const at::Scalar & alpha=1.0) {
+    return at::_ops::celu_::call(self, alpha);
+}
+
+// aten::celu.out(Tensor self, Scalar alpha=1.0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & celu_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & alpha=1.0) {
+    return at::_ops::celu_out::call(self, alpha, out);
+}
+// aten::celu.out(Tensor self, Scalar alpha=1.0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & celu_outf(const at::Tensor & self, const at::Scalar & alpha, at::Tensor & out) {
+    return at::_ops::celu_out::call(self, alpha, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/chain_matmul_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/chain_matmul_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..379d1b7a5afe03d1bff0118f71418db7c0d00e89
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/chain_matmul_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor chain_matmul(at::TensorList matrices);
+TORCH_API at::Tensor & chain_matmul_out(at::Tensor & out, at::TensorList matrices);
+TORCH_API at::Tensor & chain_matmul_outf(at::TensorList matrices, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/chalf_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/chalf_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c2b0d79fbe50c0f9aba006c9495b07e3a5b29095
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/chalf_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 chalf(const at::Tensor & self, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/channel_shuffle_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/channel_shuffle_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..ffcd0a9fcf139e51b9c48b88344abd1b4498c48f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/channel_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 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::channel_shuffle";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "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);
+};
+
+struct TORCH_API channel_shuffle_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymInt, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::channel_shuffle";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "channel_shuffle.out(Tensor self, SymInt groups, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, c10::SymInt groups, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymInt groups, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/clamp_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/clamp_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..477a42efa16f4354fc31b2b9d4abf383c8cdf471
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/clamp_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/clamp_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_clamp_out : public at::meta::structured_clamp {
+void impl(const at::Tensor & self, at::OptionalScalarRef min, at::OptionalScalarRef max, const at::Tensor & out);
+};
+TORCH_API at::Tensor clamp_quantized_cpu(const at::Tensor & self, const ::std::optional<at::Scalar> & min=::std::nullopt, const ::std::optional<at::Scalar> & max=::std::nullopt);
+struct TORCH_API structured_clamp_Tensor_out : public at::meta::structured_clamp_Tensor {
+void impl(const at::Tensor & self, at::OptionalTensorRef min, at::OptionalTensorRef max, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/concat_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/concat_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..56880b67295a34732f2aca23b5f6536b4f5d79fc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/concat_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 concat(at::TensorList tensors, int64_t dim=0);
+TORCH_API at::Tensor & concat_out(at::TensorList tensors, int64_t dim, at::Tensor & out);
+TORCH_API at::Tensor concat(at::TensorList tensors, at::Dimname dim);
+TORCH_API at::Tensor & concat_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/concatenate_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/concatenate_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..852550ccd376df10bde41f0aa9104c957238ef01
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/concatenate_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 concatenate(at::TensorList tensors, int64_t dim=0);
+TORCH_API at::Tensor & concatenate_out(at::TensorList tensors, int64_t dim, at::Tensor & out);
+TORCH_API at::Tensor concatenate(at::TensorList tensors, at::Dimname dim);
+TORCH_API at::Tensor & concatenate_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/conv1d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/conv1d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8783db0a485627344c365449fe97d6adc8e1e4f4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/conv1d_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 conv1d_symint(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef padding=c10::SymInt(0), c10::SymIntArrayRef dilation=c10::SymInt(1), c10::SymInt groups=1);
+TORCH_API at::Tensor conv1d_padding_symint(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::string_view padding="valid", c10::SymIntArrayRef dilation=c10::SymInt(1), c10::SymInt groups=1);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/conv2d.h b/lib/python3.12/site-packages/torch/include/ATen/ops/conv2d.h
new file mode 100644
index 0000000000000000000000000000000000000000..f0c75d93124538a37ab3afa86ef2a7f6b04cd944
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/conv2d.h
@@ -0,0 +1,69 @@
+#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/conv2d_ops.h>
+
+namespace at {
+
+
+// aten::conv2d(Tensor input, Tensor weight, Tensor? bias=None, SymInt[2] stride=1, SymInt[2] padding=0, SymInt[2] dilation=1, SymInt groups=1) -> Tensor
+inline at::Tensor conv2d(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={}, at::IntArrayRef stride=1, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, int64_t groups=1) {
+    return at::_ops::conv2d::call(input, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), groups);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor conv2d(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={}, at::IntArrayRef stride=1, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, int64_t groups=1) {
+    return at::_ops::conv2d::call(input, weight, bias, c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(dilation), groups);
+  }
+}
+
+// aten::conv2d(Tensor input, Tensor weight, Tensor? bias=None, SymInt[2] stride=1, SymInt[2] padding=0, SymInt[2] dilation=1, SymInt groups=1) -> Tensor
+inline at::Tensor conv2d_symint(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef padding=c10::SymInt(0), c10::SymIntArrayRef dilation=c10::SymInt(1), c10::SymInt groups=1) {
+    return at::_ops::conv2d::call(input, weight, bias, stride, padding, dilation, groups);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor conv2d(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef padding=c10::SymInt(0), c10::SymIntArrayRef dilation=c10::SymInt(1), c10::SymInt groups=1) {
+    return at::_ops::conv2d::call(input, weight, bias, stride, padding, dilation, groups);
+  }
+}
+
+// aten::conv2d.padding(Tensor input, Tensor weight, Tensor? bias=None, SymInt[2] stride=1, str padding="valid", SymInt[2] dilation=1, SymInt groups=1) -> Tensor
+inline at::Tensor conv2d(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, c10::string_view padding, at::IntArrayRef dilation=1, int64_t groups=1) {
+    return at::_ops::conv2d_padding::call(input, weight, bias, c10::fromIntArrayRefSlow(stride), padding, c10::fromIntArrayRefSlow(dilation), groups);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor conv2d(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, c10::string_view padding, at::IntArrayRef dilation=1, int64_t groups=1) {
+    return at::_ops::conv2d_padding::call(input, weight, bias, c10::fromIntArrayRefSlow(stride), padding, c10::fromIntArrayRefSlow(dilation), groups);
+  }
+}
+
+// aten::conv2d.padding(Tensor input, Tensor weight, Tensor? bias=None, SymInt[2] stride=1, str padding="valid", SymInt[2] dilation=1, SymInt groups=1) -> Tensor
+inline at::Tensor conv2d_symint(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::string_view padding, c10::SymIntArrayRef dilation=c10::SymInt(1), c10::SymInt groups=1) {
+    return at::_ops::conv2d_padding::call(input, weight, bias, stride, padding, dilation, groups);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor conv2d(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::string_view padding, c10::SymIntArrayRef dilation=c10::SymInt(1), c10::SymInt groups=1) {
+    return at::_ops::conv2d_padding::call(input, weight, bias, stride, padding, dilation, groups);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/conv_transpose1d_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/conv_transpose1d_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b8e96219f08395a7786a4eb1350672d59b58fffe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/conv_transpose1d_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 conv_transpose1d(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={}, at::IntArrayRef stride=1, at::IntArrayRef padding=0, at::IntArrayRef output_padding=0, int64_t groups=1, at::IntArrayRef dilation=1);
+TORCH_API at::Tensor conv_transpose1d_symint(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef padding=c10::SymInt(0), c10::SymIntArrayRef output_padding=c10::SymInt(0), c10::SymInt groups=1, c10::SymIntArrayRef dilation=c10::SymInt(1));
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5dbe31da7a0800878c5b75c854a6b5c3c447a963
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/convolution_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor convolution(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups);
+TORCH_API at::Tensor convolution_symint(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups);
+TORCH_API at::Tensor & convolution_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups);
+TORCH_API at::Tensor & convolution_outf(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups, at::Tensor & out);
+TORCH_API at::Tensor & convolution_symint_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups);
+TORCH_API at::Tensor & convolution_symint_outf(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/copy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/copy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9a004b1bd1ec92730cb81db3e1565b30b9bcbf85
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/copy_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 & copy_out(const at::Tensor & self, const at::Tensor & src, bool non_blocking, at::Tensor & out);
+TORCH_API at::Tensor & copy_(at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+TORCH_API at::Tensor copy_meta(const at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+TORCH_API at::Tensor & copy_nested_(at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+TORCH_API at::Tensor & copy_sparse_wrapper_(at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+TORCH_API at::Tensor & copy_sparse_compressed_(at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+TORCH_API at::Tensor & copy_mkldnn_(at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+TORCH_API at::Tensor copy(const at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/copy_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/copy_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a675b1ea8ac8c12048def017905756f1b2866e85
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/copy_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 copy {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::copy";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "copy(Tensor self, Tensor src, bool non_blocking=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & src, bool non_blocking);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & src, bool non_blocking);
+};
+
+struct TORCH_API copy_ {
+  using schema = 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::copy_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "copy_(Tensor(a!) self, Tensor src, bool non_blocking=False) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & src, bool non_blocking);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & src, bool non_blocking);
+};
+
+struct TORCH_API copy_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::copy";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "copy.out(Tensor self, Tensor src, bool non_blocking=False, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & src, bool non_blocking, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & src, bool non_blocking, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/copy_sparse_to_sparse_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/copy_sparse_to_sparse_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b9eac2148a5387735e2806603a491ea7b5f03be9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/copy_sparse_to_sparse_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 copy_sparse_to_sparse(const at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+TORCH_API at::Tensor & copy_sparse_to_sparse_out(const at::Tensor & self, const at::Tensor & src, bool non_blocking, at::Tensor & out);
+TORCH_API at::Tensor & copy_sparse_(at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/copysign_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/copysign_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..549275fceee5fd8b179f30846aeaca13f3f451f9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/copysign_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_copysign_Tensor : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Tensor & other);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cos_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cos_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..54a8622cdc21fc0012a0429901130d948fe83c39
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cos_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_cos : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cos_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cos_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..aa28fa2c890955ec10992d214fbeba8f0d14b1e9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cos_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>
+#include <ATen/ops/cos_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_cos_out : public at::meta::structured_cos {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+TORCH_API at::Tensor NestedTensor_cos(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cosine_embedding_loss_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cosine_embedding_loss_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..dde85bcfc683c38afee0ff0c9a3aab6feb8ebf93
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cosine_embedding_loss_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 cosine_embedding_loss(const at::Tensor & input1, const at::Tensor & input2, const at::Tensor & target, double margin=0.0, int64_t reduction=at::Reduction::Mean);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_affine_grid_generator_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_affine_grid_generator_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9295acd941060b30a068c441a993ad7dc4dfa27d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_affine_grid_generator_backward_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API cudnn_affine_grid_generator_backward {
+  using schema = at::Tensor (const at::Tensor &, int64_t, 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::cudnn_affine_grid_generator_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "cudnn_affine_grid_generator_backward(Tensor grad, int N, int C, int H, int W) -> Tensor grad_theta";
+  static at::Tensor call(const at::Tensor & grad, int64_t N, int64_t C, int64_t H, int64_t W);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad, int64_t N, int64_t C, int64_t H, int64_t W);
+};
+
+struct TORCH_API cudnn_affine_grid_generator_backward_out {
+  using schema = at::Tensor & (const at::Tensor &, int64_t, 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::cudnn_affine_grid_generator_backward";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "cudnn_affine_grid_generator_backward.out(Tensor grad, int N, int C, int H, int W, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & grad, int64_t N, int64_t C, int64_t H, int64_t W, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad, int64_t N, int64_t C, int64_t H, int64_t W, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_affine_grid_generator_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_affine_grid_generator_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c1310c1d2760c1e0c474227d44b7f660f1363dcd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_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 & cudnn_affine_grid_generator_out(const at::Tensor & theta, int64_t N, int64_t C, int64_t H, int64_t W, at::Tensor & out);
+TORCH_API at::Tensor cudnn_affine_grid_generator_forward(const at::Tensor & theta, int64_t N, int64_t C, int64_t H, int64_t W);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_batch_norm_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_batch_norm_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..8ea257018c38893d51be5b888d409e14665384f0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_batch_norm_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/cudnn_batch_norm_backward_ops.h>
+
+namespace at {
+
+
+// aten::cudnn_batch_norm_backward(Tensor input, Tensor grad_output, Tensor weight, Tensor? running_mean, Tensor? running_var, Tensor? save_mean, Tensor? save_var, float epsilon, Tensor reserveSpace) -> (Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> cudnn_batch_norm_backward(const at::Tensor & input, const at::Tensor & grad_output, const at::Tensor & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_var, double epsilon, const at::Tensor & reserveSpace) {
+    return at::_ops::cudnn_batch_norm_backward::call(input, grad_output, weight, running_mean, running_var, save_mean, save_var, epsilon, reserveSpace);
+}
+
+// aten::cudnn_batch_norm_backward.out(Tensor input, Tensor grad_output, Tensor weight, Tensor? running_mean, Tensor? running_var, Tensor? save_mean, Tensor? save_var, float epsilon, Tensor reserveSpace, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> cudnn_batch_norm_backward_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & input, const at::Tensor & grad_output, const at::Tensor & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_var, double epsilon, const at::Tensor & reserveSpace) {
+    return at::_ops::cudnn_batch_norm_backward_out::call(input, grad_output, weight, running_mean, running_var, save_mean, save_var, epsilon, reserveSpace, out0, out1, out2);
+}
+// aten::cudnn_batch_norm_backward.out(Tensor input, Tensor grad_output, Tensor weight, Tensor? running_mean, Tensor? running_var, Tensor? save_mean, Tensor? save_var, float epsilon, Tensor reserveSpace, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> cudnn_batch_norm_backward_outf(const at::Tensor & input, const at::Tensor & grad_output, const at::Tensor & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_var, double epsilon, const at::Tensor & reserveSpace, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2) {
+    return at::_ops::cudnn_batch_norm_backward_out::call(input, grad_output, weight, running_mean, running_var, save_mean, save_var, epsilon, reserveSpace, out0, out1, out2);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_batch_norm_backward_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_batch_norm_backward_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1c824d76d063f2c569c77078fddc0f21eb6a5270
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_batch_norm_backward_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> cudnn_batch_norm_backward_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & input, const at::Tensor & grad_output, const at::Tensor & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_var, double epsilon, const at::Tensor & reserveSpace);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> cudnn_batch_norm_backward_outf(const at::Tensor & input, const at::Tensor & grad_output, const at::Tensor & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_var, double epsilon, const at::Tensor & reserveSpace, 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/cudnn_batch_norm_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_batch_norm_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ba02fd83d5283a4de1d7621ab72ec7e503fc65fa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_batch_norm_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> cudnn_batch_norm_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, bool training, double exponential_average_factor, double epsilon);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> cudnn_batch_norm_outf(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, bool training, double exponential_average_factor, double epsilon, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..c262c658f4ede2af8103ab56d3ee546277c2c0e1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API cudnn_convolution {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymInt, bool, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::cudnn_convolution";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "cudnn_convolution(Tensor self, Tensor weight, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, bool allow_tf32) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32);
+};
+
+struct TORCH_API cudnn_convolution_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymInt, bool, bool, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::cudnn_convolution";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "cudnn_convolution.out(Tensor self, Tensor weight, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, bool allow_tf32, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_grid_sampler.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_grid_sampler.h
new file mode 100644
index 0000000000000000000000000000000000000000..9ce0fbaeb590571273b58c5772526c48af4d8b7d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_grid_sampler.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/cudnn_grid_sampler_ops.h>
+
+namespace at {
+
+
+// aten::cudnn_grid_sampler(Tensor self, Tensor grid) -> Tensor output
+inline at::Tensor cudnn_grid_sampler(const at::Tensor & self, const at::Tensor & grid) {
+    return at::_ops::cudnn_grid_sampler::call(self, grid);
+}
+
+// aten::cudnn_grid_sampler.out(Tensor self, Tensor grid, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cudnn_grid_sampler_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & grid) {
+    return at::_ops::cudnn_grid_sampler_out::call(self, grid, out);
+}
+// aten::cudnn_grid_sampler.out(Tensor self, Tensor grid, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cudnn_grid_sampler_outf(const at::Tensor & self, const at::Tensor & grid, at::Tensor & out) {
+    return at::_ops::cudnn_grid_sampler_out::call(self, grid, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cummaxmin_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cummaxmin_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..385bc5fb9e65e388100cabcde81559171e142a64
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cummaxmin_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/cummaxmin_backward_ops.h>
+
+namespace at {
+
+
+// aten::cummaxmin_backward(Tensor grad, Tensor input, Tensor indices, int dim) -> Tensor
+inline at::Tensor cummaxmin_backward(const at::Tensor & grad, const at::Tensor & input, const at::Tensor & indices, int64_t dim) {
+    return at::_ops::cummaxmin_backward::call(grad, input, indices, dim);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cumulative_trapezoid_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cumulative_trapezoid_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0e78ed19fef37909b2a0d7b5f87fecacdf735909
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cumulative_trapezoid_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 cumulative_trapezoid(const at::Tensor & y, const at::Tensor & x, int64_t dim=-1);
+TORCH_API at::Tensor cumulative_trapezoid(const at::Tensor & y, const at::Scalar & dx=1, int64_t dim=-1);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/diff.h b/lib/python3.12/site-packages/torch/include/ATen/ops/diff.h
new file mode 100644
index 0000000000000000000000000000000000000000..b3158f8e0f6000a6f27188547f1c98ce099f9d58
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/diff.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/diff_ops.h>
+
+namespace at {
+
+
+// aten::diff(Tensor self, int n=1, int dim=-1, Tensor? prepend=None, Tensor? append=None) -> Tensor
+inline at::Tensor diff(const at::Tensor & self, int64_t n=1, int64_t dim=-1, const ::std::optional<at::Tensor> & prepend={}, const ::std::optional<at::Tensor> & append={}) {
+    return at::_ops::diff::call(self, n, dim, prepend, append);
+}
+
+// aten::diff.out(Tensor self, int n=1, int dim=-1, Tensor? prepend=None, Tensor? append=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & diff_out(at::Tensor & out, const at::Tensor & self, int64_t n=1, int64_t dim=-1, const ::std::optional<at::Tensor> & prepend={}, const ::std::optional<at::Tensor> & append={}) {
+    return at::_ops::diff_out::call(self, n, dim, prepend, append, out);
+}
+// aten::diff.out(Tensor self, int n=1, int dim=-1, Tensor? prepend=None, Tensor? append=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & diff_outf(const at::Tensor & self, int64_t n, int64_t dim, const ::std::optional<at::Tensor> & prepend, const ::std::optional<at::Tensor> & append, at::Tensor & out) {
+    return at::_ops::diff_out::call(self, n, dim, prepend, append, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/digamma_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/digamma_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..fb125902be2003c62d6c53d252096713ea5f138b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/digamma_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_digamma : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/dist.h b/lib/python3.12/site-packages/torch/include/ATen/ops/dist.h
new file mode 100644
index 0000000000000000000000000000000000000000..198bad40af80a2d60289951c83d39b9dc6067e45
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/dist.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/dist_ops.h>
+
+namespace at {
+
+
+// aten::dist(Tensor self, Tensor other, Scalar p=2) -> Tensor
+inline at::Tensor dist(const at::Tensor & self, const at::Tensor & other, const at::Scalar & p=2) {
+    return at::_ops::dist::call(self, other, p);
+}
+
+// aten::dist.out(Tensor self, Tensor other, Scalar p=2, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & dist_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other, const at::Scalar & p=2) {
+    return at::_ops::dist_out::call(self, other, p, out);
+}
+// aten::dist.out(Tensor self, Tensor other, Scalar p=2, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & dist_outf(const at::Tensor & self, const at::Tensor & other, const at::Scalar & p, at::Tensor & out) {
+    return at::_ops::dist_out::call(self, other, p, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/dist_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/dist_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7742bd4b2f9d27ff6077e29c9098e0770397b32e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/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 dist(const at::Tensor & self, const at::Tensor & other, const at::Scalar & p=2);
+TORCH_API at::Tensor & dist_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other, const at::Scalar & p=2);
+TORCH_API at::Tensor & dist_outf(const at::Tensor & self, const at::Tensor & other, const at::Scalar & p, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/div_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/div_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..0888b13f7fa482fb0a65d881ed7548f64e03cf86
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/div_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_div_Tensor : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Tensor & other);
+};
+struct TORCH_API structured_div_Tensor_mode : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Tensor & other, ::std::optional<c10::string_view> rounding_mode);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/dot_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/dot_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..bc8bb6ec68f4b94f5532ba58a49add50211046f6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/dot_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 dot {
+  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::dot";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "dot(Tensor self, Tensor tensor) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & tensor);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & tensor);
+};
+
+struct TORCH_API dot_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::dot";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "dot.out(Tensor self, Tensor tensor, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & tensor, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & tensor, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/einsum_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/einsum_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..74b577c528f9eb37b3bfc5142013669cdce983a7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/einsum_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 einsum(c10::string_view equation, at::TensorList tensors, at::OptionalIntArrayRef path=::std::nullopt);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/einsum_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/einsum_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..4e2e3613a9cfc5680240d8b942352dc91aea712d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/einsum_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 einsum {
+  using schema = at::Tensor (c10::string_view, at::TensorList, at::OptionalIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::einsum";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "einsum(str equation, Tensor[] tensors, *, int[]? path=None) -> Tensor";
+  static at::Tensor call(c10::string_view equation, at::TensorList tensors, at::OptionalIntArrayRef path);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, c10::string_view equation, at::TensorList tensors, at::OptionalIntArrayRef path);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/elu_backward_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/elu_backward_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..483f603f25b1cf050becf2be52df06a201faa50e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/elu_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 elu_backward(const at::Tensor & grad_output, const at::Scalar & alpha, const at::Scalar & scale, const at::Scalar & input_scale, bool is_result, const at::Tensor & self_or_result);
+TORCH_API at::Tensor & elu_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Scalar & alpha, const at::Scalar & scale, const at::Scalar & input_scale, bool is_result, const at::Tensor & self_or_result);
+TORCH_API at::Tensor & elu_backward_outf(const at::Tensor & grad_output, const at::Scalar & alpha, const at::Scalar & scale, const at::Scalar & input_scale, bool is_result, const at::Tensor & self_or_result, at::Tensor & grad_input);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/elu_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/elu_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..04a76e25109ac742777b3b46fe2d475f8d503b3d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/elu_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 elu(const at::Tensor & self, const at::Scalar & alpha=1, const at::Scalar & scale=1, const at::Scalar & input_scale=1);
+TORCH_API at::Tensor & elu_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & alpha=1, const at::Scalar & scale=1, const at::Scalar & input_scale=1);
+TORCH_API at::Tensor & elu_outf(const at::Tensor & self, const at::Scalar & alpha, const at::Scalar & scale, const at::Scalar & input_scale, at::Tensor & out);
+TORCH_API at::Tensor & elu_(at::Tensor & self, const at::Scalar & alpha=1, const at::Scalar & scale=1, const at::Scalar & input_scale=1);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_dense_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_dense_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..baeb8bd12b318499f66be8c4177c5c384d498be5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_dense_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/embedding_dense_backward_ops.h>
+
+namespace at {
+
+
+// aten::embedding_dense_backward(Tensor grad_output, Tensor indices, SymInt num_weights, SymInt padding_idx, bool scale_grad_by_freq) -> Tensor
+inline at::Tensor embedding_dense_backward(const at::Tensor & grad_output, const at::Tensor & indices, int64_t num_weights, int64_t padding_idx, bool scale_grad_by_freq) {
+    return at::_ops::embedding_dense_backward::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor embedding_dense_backward(const at::Tensor & grad_output, const at::Tensor & indices, int64_t num_weights, int64_t padding_idx, bool scale_grad_by_freq) {
+    return at::_ops::embedding_dense_backward::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq);
+  }
+}
+
+// aten::embedding_dense_backward(Tensor grad_output, Tensor indices, SymInt num_weights, SymInt padding_idx, bool scale_grad_by_freq) -> Tensor
+inline at::Tensor embedding_dense_backward_symint(const at::Tensor & grad_output, const at::Tensor & indices, c10::SymInt num_weights, c10::SymInt padding_idx, bool scale_grad_by_freq) {
+    return at::_ops::embedding_dense_backward::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor embedding_dense_backward(const at::Tensor & grad_output, const at::Tensor & indices, c10::SymInt num_weights, c10::SymInt padding_idx, bool scale_grad_by_freq) {
+    return at::_ops::embedding_dense_backward::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq);
+  }
+}
+
+// aten::embedding_dense_backward.out(Tensor grad_output, Tensor indices, SymInt num_weights, SymInt padding_idx, bool scale_grad_by_freq, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & embedding_dense_backward_out(at::Tensor & out, const at::Tensor & grad_output, const at::Tensor & indices, int64_t num_weights, int64_t padding_idx, bool scale_grad_by_freq) {
+    return at::_ops::embedding_dense_backward_out::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & embedding_dense_backward_out(at::Tensor & out, const at::Tensor & grad_output, const at::Tensor & indices, int64_t num_weights, int64_t padding_idx, bool scale_grad_by_freq) {
+    return at::_ops::embedding_dense_backward_out::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq, out);
+  }
+}
+
+// aten::embedding_dense_backward.out(Tensor grad_output, Tensor indices, SymInt num_weights, SymInt padding_idx, bool scale_grad_by_freq, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & embedding_dense_backward_outf(const at::Tensor & grad_output, const at::Tensor & indices, int64_t num_weights, int64_t padding_idx, bool scale_grad_by_freq, at::Tensor & out) {
+    return at::_ops::embedding_dense_backward_out::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & embedding_dense_backward_outf(const at::Tensor & grad_output, const at::Tensor & indices, int64_t num_weights, int64_t padding_idx, bool scale_grad_by_freq, at::Tensor & out) {
+    return at::_ops::embedding_dense_backward_out::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq, out);
+  }
+}
+
+// aten::embedding_dense_backward.out(Tensor grad_output, Tensor indices, SymInt num_weights, SymInt padding_idx, bool scale_grad_by_freq, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & embedding_dense_backward_symint_out(at::Tensor & out, const at::Tensor & grad_output, const at::Tensor & indices, c10::SymInt num_weights, c10::SymInt padding_idx, bool scale_grad_by_freq) {
+    return at::_ops::embedding_dense_backward_out::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & embedding_dense_backward_out(at::Tensor & out, const at::Tensor & grad_output, const at::Tensor & indices, c10::SymInt num_weights, c10::SymInt padding_idx, bool scale_grad_by_freq) {
+    return at::_ops::embedding_dense_backward_out::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq, out);
+  }
+}
+
+// aten::embedding_dense_backward.out(Tensor grad_output, Tensor indices, SymInt num_weights, SymInt padding_idx, bool scale_grad_by_freq, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & embedding_dense_backward_symint_outf(const at::Tensor & grad_output, const at::Tensor & indices, c10::SymInt num_weights, c10::SymInt padding_idx, bool scale_grad_by_freq, at::Tensor & out) {
+    return at::_ops::embedding_dense_backward_out::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & embedding_dense_backward_outf(const at::Tensor & grad_output, const at::Tensor & indices, c10::SymInt num_weights, c10::SymInt padding_idx, bool scale_grad_by_freq, at::Tensor & out) {
+    return at::_ops::embedding_dense_backward_out::call(grad_output, indices, num_weights, padding_idx, scale_grad_by_freq, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/empty_like_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_like_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0280aadec87191cabeb0542f23ad1c56d8328984
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_like_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_like(const at::Tensor & self, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor empty_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);
+TORCH_API at::Tensor & empty_like_out(at::Tensor & out, const at::Tensor & self, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor & empty_like_outf(const at::Tensor & self, ::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_permuted_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_permuted_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1ef764bb7410f95b009e6db2ffae63a41f80724c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_permuted_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_permuted {
+  using schema = at::Tensor (c10::SymIntArrayRef, at::IntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::empty_permuted";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "empty_permuted(SymInt[] size, int[] physical_layout, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(c10::SymIntArrayRef size, at::IntArrayRef physical_layout, ::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, at::IntArrayRef physical_layout, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API empty_permuted_out {
+  using schema = at::Tensor & (c10::SymIntArrayRef, at::IntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::empty_permuted";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "empty_permuted.out(SymInt[] size, int[] physical_layout, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymIntArrayRef size, at::IntArrayRef physical_layout, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymIntArrayRef size, at::IntArrayRef physical_layout, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/erfc_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/erfc_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..66216294b4a2b8ae7c43534fe54d2070853c8565
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/erfc_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 erfc(const at::Tensor & self);
+TORCH_API at::Tensor & erfc_(at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/erfc_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/erfc_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4c354c0ff55279981692258d623216cdd47653d7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/erfc_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/erfc_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_erfc_out : public at::meta::structured_erfc {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/erfinv_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/erfinv_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c4aea293eb0634655d184f74a54778f32b09d504
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/erfinv_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 erfinv(const at::Tensor & self);
+TORCH_API at::Tensor & erfinv_(at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/expand_copy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/expand_copy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..dc83f53673b7384e6a6f0b632eb59b48a609aac1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/expand_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 & expand_copy_out_symint(const at::Tensor & self, c10::SymIntArrayRef size, bool implicit, at::Tensor & out);
+TORCH_API at::Tensor expand_copy_symint(const at::Tensor & self, c10::SymIntArrayRef size, bool implicit=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/eye_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/eye_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..92f79da163d3c0d4c1cdc9b05e5ec4c0855ad15f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/eye_compositeexplicitautograd_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 compositeexplicitautograd {
+
+TORCH_API at::Tensor eye(int64_t n, at::TensorOptions options={});
+TORCH_API at::Tensor eye(int64_t n, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor eye_symint(c10::SymInt n, at::TensorOptions options={});
+TORCH_API at::Tensor eye_symint(c10::SymInt n, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor eye(int64_t n, int64_t m, at::TensorOptions options={});
+TORCH_API at::Tensor eye(int64_t n, int64_t m, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor eye_symint(c10::SymInt n, c10::SymInt m, at::TensorOptions options={});
+TORCH_API at::Tensor eye_symint(c10::SymInt n, c10::SymInt m, ::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/fake_quantize_per_channel_affine_cachemask_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_cachemask_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..ecf30243b1f3515f5526c0ee820cf02586c6da9d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_cachemask_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API fake_quantize_per_channel_affine_cachemask {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, const 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::fake_quantize_per_channel_affine_cachemask";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "fake_quantize_per_channel_affine_cachemask(Tensor self, Tensor scale, Tensor zero_point, int axis, int quant_min, int quant_max) -> (Tensor output, Tensor mask)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max);
+};
+
+struct TORCH_API fake_quantize_per_channel_affine_cachemask_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const at::Tensor &, const at::Tensor &, const at::Tensor &, int64_t, int64_t, int64_t, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fake_quantize_per_channel_affine_cachemask";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "fake_quantize_per_channel_affine_cachemask.out(Tensor self, Tensor scale, Tensor zero_point, int axis, int quant_min, int quant_max, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max, at::Tensor & out0, at::Tensor & out1);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max, at::Tensor & out0, at::Tensor & out1);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_quantize_weight_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_quantize_weight_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..7dfe49713c2620644acad567443e2cb841681dd2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_quantize_weight_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API fbgemm_linear_quantize_weight {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,double,int64_t> (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fbgemm_linear_quantize_weight";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "fbgemm_linear_quantize_weight(Tensor input) -> (Tensor, Tensor, float, int)";
+  static ::std::tuple<at::Tensor,at::Tensor,double,int64_t> call(const at::Tensor & input);
+  static ::std::tuple<at::Tensor,at::Tensor,double,int64_t> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/feature_dropout.h b/lib/python3.12/site-packages/torch/include/ATen/ops/feature_dropout.h
new file mode 100644
index 0000000000000000000000000000000000000000..96ab123bf5cfecd9f71512c56c7aa82c9a660c53
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/feature_dropout.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/feature_dropout_ops.h>
+
+namespace at {
+
+
+// aten::feature_dropout(Tensor input, float p, bool train) -> Tensor
+inline at::Tensor feature_dropout(const at::Tensor & input, double p, bool train) {
+    return at::_ops::feature_dropout::call(input, p, train);
+}
+
+// aten::feature_dropout_(Tensor(a!) self, float p, bool train) -> Tensor(a!)
+inline at::Tensor & feature_dropout_(at::Tensor & self, double p, bool train) {
+    return at::_ops::feature_dropout_::call(self, p, train);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft2.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft2.h
new file mode 100644
index 0000000000000000000000000000000000000000..328e21a6bb059989bb6d886de2fe130dfd4b0333
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifft2.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_ifft2_ops.h>
+
+namespace at {
+
+
+// aten::fft_ifft2(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None) -> Tensor
+inline at::Tensor fft_ifft2(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifft2::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_ifft2(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_ifft2::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm);
+  }
+}
+
+// aten::fft_ifft2(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None) -> Tensor
+inline at::Tensor fft_ifft2_symint(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifft2::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_ifft2(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_ifft2::call(self, s, dim, norm);
+  }
+}
+
+// aten::fft_ifft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_ifft2_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifft2_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & fft_ifft2_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifft2_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_ifft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_ifft2_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_ifft2_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & fft_ifft2_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_ifft2_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_ifft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_ifft2_symint_out(at::Tensor & out, const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifft2_out::call(self, s, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & fft_ifft2_out(at::Tensor & out, const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_ifft2_out::call(self, s, dim, norm, out);
+  }
+}
+
+// aten::fft_ifft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_ifft2_symint_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_ifft2_out::call(self, s, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & fft_ifft2_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_ifft2_out::call(self, s, dim, norm, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftfreq_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftfreq_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4bf8ac53c30409d9fc06828d9903d1e532d260d8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftfreq_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor fft_rfftfreq(int64_t n, double d=1.0, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={});
+TORCH_API at::Tensor & fft_rfftfreq_out(int64_t n, double d, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftn_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftn_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..15861bedee8e3f3def82c535f8a3eb0485e43471
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftn_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_rfftn {
+  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_rfftn";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "fft_rfftn(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_rfftn_out {
+  using schema = at::Tensor & (const at::Tensor &, at::OptionalSymIntArrayRef, at::OptionalIntArrayRef, ::std::optional<c10::string_view>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fft_rfftn";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "fft_rfftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fill_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fill_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9332df138892de6340f2da918ec6683dae7eef08
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fill_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 & fill_(at::Tensor & self, const at::Scalar & value);
+TORCH_API at::Tensor & fill_(at::Tensor & self, const at::Tensor & value);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fill_diagonal_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fill_diagonal_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..acaa70a8ced133e36b16de07480baee71130098d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fill_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 & fill_diagonal_(at::Tensor & self, const at::Scalar & fill_value, bool wrap=false);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fix_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fix_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..90856c0eea1cb2dccfe9851910e34c33103cc624
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fix_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 fix(const at::Tensor & self);
+TORCH_API at::Tensor & fix_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & fix_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & fix_(at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fliplr_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fliplr_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..68310b01e3e2b4cca9eac4fea5dc8d26c7af2b82
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fliplr_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 fliplr(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..97054aebdf2a440f0f51abf4f003663fc6d05c32
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_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 fmax(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & fmax_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & fmax_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1c51ba2002c4422ab5f3279397efb82e76e312ad
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fmax_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 fmax(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & fmax_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & fmax_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fmod_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fmod_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e5842c4a4bc52301f61dc685e316efc068e63f5d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fmod_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 fmod(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & fmod_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & fmod_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & fmod_(at::Tensor & self, const at::Scalar & other);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fmod_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fmod_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d785c97d7d639e4db2843e430e711b2d45efb6ec
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fmod_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 fmod(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & fmod_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & fmod_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & fmod_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/frac_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/frac_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4b77882d7756f03fd5ebf2b73fc5ef0ba0a25b2d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/frac_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 frac(const at::Tensor & self);
+TORCH_API at::Tensor & frac_(at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool2d_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool2d_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..a0bde9faeffa4cbd265687632a7e24083146f361
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool2d_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_fractional_max_pool2d : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & random_samples);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fdf6534cf41bfba0453d57bb3d9c26cc021ab5a7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fractional_max_pool3d_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> fractional_max_pool3d(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & random_samples);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/from_file.h b/lib/python3.12/site-packages/torch/include/ATen/ops/from_file.h
new file mode 100644
index 0000000000000000000000000000000000000000..d7f275ddbd2c026ab51846af3873643b7469a31d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/from_file.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/from_file_ops.h>
+
+namespace at {
+
+
+// aten::from_file(str filename, bool? shared=None, int? size=0, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor from_file(c10::string_view filename, ::std::optional<bool> shared=::std::nullopt, ::std::optional<int64_t> size=0, at::TensorOptions options={}) {
+    return at::_ops::from_file::call(filename, shared, size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::from_file(str filename, bool? shared=None, int? size=0, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor from_file(c10::string_view filename, ::std::optional<bool> shared, ::std::optional<int64_t> size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::from_file::call(filename, shared, size, dtype, layout, device, pin_memory);
+}
+
+// aten::from_file.out(str filename, bool? shared=None, int? size=0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & from_file_out(at::Tensor & out, c10::string_view filename, ::std::optional<bool> shared=::std::nullopt, ::std::optional<int64_t> size=0) {
+    return at::_ops::from_file_out::call(filename, shared, size, out);
+}
+// aten::from_file.out(str filename, bool? shared=None, int? size=0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & from_file_outf(c10::string_view filename, ::std::optional<bool> shared, ::std::optional<int64_t> size, at::Tensor & out) {
+    return at::_ops::from_file_out::call(filename, shared, size, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gather_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gather_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..828620d2bca938902fd15f1b41be0d003022132b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gather_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 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 meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/geometric_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/geometric_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..fe714db661fa30c019a782bed34c851f2b888bfe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/geometric_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 geometric_ {
+  using schema = at::Tensor & (at::Tensor &, double, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::geometric_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "geometric_(Tensor(a!) self, float p, *, Generator? generator=None) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, double p, ::std::optional<at::Generator> generator);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, double p, ::std::optional<at::Generator> generator);
+};
+
+struct TORCH_API geometric_out {
+  using schema = at::Tensor & (const at::Tensor &, double, ::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::geometric";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "geometric.out(Tensor self, float p, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, double p, ::std::optional<at::Generator> generator, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, double p, ::std::optional<at::Generator> generator, at::Tensor & out);
+};
+
+struct TORCH_API geometric {
+  using schema = at::Tensor (const at::Tensor &, double, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::geometric";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "geometric(Tensor self, float p, *, Generator? generator=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, double p, ::std::optional<at::Generator> generator);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, double p, ::std::optional<at::Generator> generator);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/geqrf_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/geqrf_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c27fb0722080b67b07010e295dbf4dc566f90f59
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/geqrf_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> geqrf(const at::Tensor & self);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> geqrf_out(const at::Tensor & self, at::Tensor & a, at::Tensor & tau);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gradient.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gradient.h
new file mode 100644
index 0000000000000000000000000000000000000000..9b55d1a230197b9a1cfbd6e65aa9a9fa53588964
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gradient.h
@@ -0,0 +1,60 @@
+#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/gradient_ops.h>
+
+namespace at {
+
+
+// aten::gradient.scalarint(Tensor self, *, Scalar? spacing=None, int? dim=None, int edge_order=1) -> Tensor[]
+inline ::std::vector<at::Tensor> gradient(const at::Tensor & self, const ::std::optional<at::Scalar> & spacing=::std::nullopt, ::std::optional<int64_t> dim=::std::nullopt, int64_t edge_order=1) {
+    return at::_ops::gradient_scalarint::call(self, spacing, dim, edge_order);
+}
+
+// aten::gradient.scalararray(Tensor self, *, Scalar spacing, int[] dim, int edge_order=1) -> Tensor[]
+inline ::std::vector<at::Tensor> gradient(const at::Tensor & self, const at::Scalar & spacing, at::IntArrayRef dim, int64_t edge_order=1) {
+    return at::_ops::gradient_scalararray::call(self, spacing, dim, edge_order);
+}
+
+// aten::gradient.array(Tensor self, *, int[] dim, int edge_order=1) -> Tensor[]
+inline ::std::vector<at::Tensor> gradient(const at::Tensor & self, at::IntArrayRef dim, int64_t edge_order=1) {
+    return at::_ops::gradient_array::call(self, dim, edge_order);
+}
+
+// aten::gradient.scalarrayint(Tensor self, *, Scalar[] spacing, int? dim=None, int edge_order=1) -> Tensor[]
+inline ::std::vector<at::Tensor> gradient(const at::Tensor & self, at::ArrayRef<at::Scalar> spacing, ::std::optional<int64_t> dim=::std::nullopt, int64_t edge_order=1) {
+    return at::_ops::gradient_scalarrayint::call(self, spacing, dim, edge_order);
+}
+
+// aten::gradient.scalarrayarray(Tensor self, *, Scalar[] spacing, int[] dim, int edge_order=1) -> Tensor[]
+inline ::std::vector<at::Tensor> gradient(const at::Tensor & self, at::ArrayRef<at::Scalar> spacing, at::IntArrayRef dim, int64_t edge_order=1) {
+    return at::_ops::gradient_scalarrayarray::call(self, spacing, dim, edge_order);
+}
+
+// aten::gradient.tensorarrayint(Tensor self, *, Tensor[] spacing, int? dim=None, int edge_order=1) -> Tensor[]
+inline ::std::vector<at::Tensor> gradient(const at::Tensor & self, at::TensorList spacing, ::std::optional<int64_t> dim=::std::nullopt, int64_t edge_order=1) {
+    return at::_ops::gradient_tensorarrayint::call(self, spacing, dim, edge_order);
+}
+
+// aten::gradient.tensorarray(Tensor self, *, Tensor[] spacing, int[] dim, int edge_order=1) -> Tensor[]
+inline ::std::vector<at::Tensor> gradient(const at::Tensor & self, at::TensorList spacing, at::IntArrayRef dim, int64_t edge_order=1) {
+    return at::_ops::gradient_tensorarray::call(self, spacing, dim, edge_order);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/greater_equal_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/greater_equal_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..325f39baa1a41897d79cb8bc51219933b7346ce7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/greater_equal_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 greater_equal_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::greater_equal";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "greater_equal.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+};
+
+struct TORCH_API greater_equal_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::greater_equal";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "greater_equal.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 greater_equal_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::greater_equal";
+  static constexpr const char* overload_name = "Tensor_out";
+  static constexpr const char* schema_str = "greater_equal.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+};
+
+struct TORCH_API greater_equal_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::greater_equal";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "greater_equal.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 greater_equal__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::greater_equal_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "greater_equal_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Scalar & other);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Scalar & other);
+};
+
+struct TORCH_API greater_equal__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::greater_equal_";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "greater_equal_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & other);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & other);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/greater_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/greater_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..935f5f55a299b899c60095acf00b1671ab4d8392
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/greater_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 greater(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & greater_out(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & greater_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor greater(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & greater_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & greater_(at::Tensor & self, const at::Tensor & other);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gru_cell_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gru_cell_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..0d800106c399ed46da6d48f0297854995b112979
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gru_cell_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 gru_cell {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::gru_cell";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "gru_cell(Tensor input, Tensor hx, Tensor w_ih, Tensor w_hh, Tensor? b_ih=None, Tensor? b_hh=None) -> Tensor";
+  static at::Tensor call(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);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, 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 at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..45da88a4c4d95b747e5615ad26c0ae7c26534cb1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_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 hardshrink(const at::Tensor & self, const at::Scalar & lambd=0.5);
+TORCH_API at::Tensor & hardshrink_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & lambd=0.5);
+TORCH_API at::Tensor & hardshrink_outf(const at::Tensor & self, const at::Scalar & lambd, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ddedaf4cb1de8e242b8313803308e3c73eb75a42
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_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 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 cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b91f096e0a72beb6fc978452c7af98ffd27fd1fd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_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 hardsigmoid(const at::Tensor & self);
+TORCH_API at::Tensor & hardsigmoid_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & hardsigmoid_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & hardsigmoid_(at::Tensor & self);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardtanh_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardtanh_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..90e05ba52039c823736375263b44bd6106bddf98
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardtanh_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 hardtanh(const at::Tensor & self, const at::Scalar & min_val=-1, const at::Scalar & max_val=1);
+TORCH_API at::Tensor & hardtanh_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & min_val=-1, const at::Scalar & max_val=1);
+TORCH_API at::Tensor & hardtanh_outf(const at::Tensor & self, const at::Scalar & min_val, const at::Scalar & max_val, at::Tensor & out);
+TORCH_API at::Tensor & hardtanh_(at::Tensor & self, const at::Scalar & min_val=-1, const at::Scalar & max_val=1);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hinge_embedding_loss.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hinge_embedding_loss.h
new file mode 100644
index 0000000000000000000000000000000000000000..3cc7f671dc8c5dd18eff078076f428061258b382
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hinge_embedding_loss.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/hinge_embedding_loss_ops.h>
+
+namespace at {
+
+
+// aten::hinge_embedding_loss(Tensor self, Tensor target, float margin=1.0, int reduction=Mean) -> Tensor
+inline at::Tensor hinge_embedding_loss(const at::Tensor & self, const at::Tensor & target, double margin=1.0, int64_t reduction=at::Reduction::Mean) {
+    return at::_ops::hinge_embedding_loss::call(self, target, margin, reduction);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hinge_embedding_loss_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hinge_embedding_loss_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e5b29bc9b9f7e3b8ef594bc0295912bb24f88935
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hinge_embedding_loss_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 hinge_embedding_loss(const at::Tensor & self, const at::Tensor & target, double margin=1.0, int64_t reduction=at::Reduction::Mean);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hinge_embedding_loss_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hinge_embedding_loss_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..80b70c1a1a4a27b896fe8e4446bd4895c70bfdb8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hinge_embedding_loss_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 hinge_embedding_loss(const at::Tensor & self, const at::Tensor & target, double margin=1.0, int64_t reduction=at::Reduction::Mean);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hspmm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hspmm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..227afb22e0813bfcd58b596f45d27e3473d6a730
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hspmm_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 hspmm_sparse_cpu(const at::Tensor & mat1, const at::Tensor & mat2);
+TORCH_API at::Tensor & hspmm_out_sparse_cpu(const at::Tensor & mat1, const at::Tensor & mat2, at::Tensor & out);
+TORCH_API at::Tensor hspmm_sparse_cuda(const at::Tensor & mat1, const at::Tensor & mat2);
+TORCH_API at::Tensor & hspmm_out_sparse_cuda(const at::Tensor & mat1, const at::Tensor & mat2, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hstack_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hstack_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6e7d4bccbbbb214b5a16b7d9bcddd7ac8aa7ff65
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hstack_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor hstack(at::TensorList tensors);
+TORCH_API at::Tensor & hstack_out(at::Tensor & out, at::TensorList tensors);
+TORCH_API at::Tensor & hstack_outf(at::TensorList tensors, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/i0_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/i0_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..7c4c85ba0f0773c9cada314ffa34fd8d2822267b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/i0_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_i0 : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..33b5003030f46842945bf85ea5f9948d9b25ab73
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_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 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 cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..fed1db3698657391ec6d1ba0a49931315e4e0d42
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_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/index_select_backward_ops.h>
+
+namespace at {
+
+
+// aten::index_select_backward(Tensor grad, SymInt[] self_sizes, int dim, Tensor index) -> Tensor
+inline at::Tensor index_select_backward(const at::Tensor & grad, at::IntArrayRef self_sizes, int64_t dim, const at::Tensor & index) {
+    return at::_ops::index_select_backward::call(grad, c10::fromIntArrayRefSlow(self_sizes), dim, index);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor index_select_backward(const at::Tensor & grad, at::IntArrayRef self_sizes, int64_t dim, const at::Tensor & index) {
+    return at::_ops::index_select_backward::call(grad, c10::fromIntArrayRefSlow(self_sizes), dim, index);
+  }
+}
+
+// aten::index_select_backward(Tensor grad, SymInt[] self_sizes, int dim, Tensor index) -> Tensor
+inline at::Tensor index_select_backward_symint(const at::Tensor & grad, c10::SymIntArrayRef self_sizes, int64_t dim, const at::Tensor & index) {
+    return at::_ops::index_select_backward::call(grad, self_sizes, dim, index);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor index_select_backward(const at::Tensor & grad, c10::SymIntArrayRef self_sizes, int64_t dim, const at::Tensor & index) {
+    return at::_ops::index_select_backward::call(grad, self_sizes, dim, index);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/indices.h b/lib/python3.12/site-packages/torch/include/ATen/ops/indices.h
new file mode 100644
index 0000000000000000000000000000000000000000..c3fa482dc4c6c52bac62750f7a70b662e933e41c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/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/indices_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_complex_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_complex_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..e33b9448064fb71f5fa690a480b4049c8b010e80
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_complex_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_complex(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_leaf_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_leaf_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..fe540d99ddb30049061435630867619755bf6a21
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_leaf_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_leaf(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_neg_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_neg_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..db3b1f5afd8943186c1e6ec064c1477d0bde62c0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_neg_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_neg {
+  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_neg";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "is_neg(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/isin_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/isin_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..a982f1f0cffe907649217e99f3708756b81dec9f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/isin_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_isin_Tensor_Tensor : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & elements, const at::Tensor & test_elements, bool assume_unique, bool invert);
+};
+struct TORCH_API structured_isin_Tensor_Scalar : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & elements, const at::Scalar & test_element, bool assume_unique, bool invert);
+};
+struct TORCH_API structured_isin_Scalar_Tensor : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Scalar & element, const at::Tensor & test_elements, bool assume_unique, bool invert);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/istft_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/istft_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..eb32c552923bddbbb8c7dbce887de5b1609532cc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/istft_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 istft {
+  using schema = at::Tensor (const at::Tensor &, int64_t, ::std::optional<int64_t>, ::std::optional<int64_t>, const ::std::optional<at::Tensor> &, bool, bool, ::std::optional<bool>, ::std::optional<int64_t>, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::istft";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "istft(Tensor self, int n_fft, int? hop_length=None, int? win_length=None, Tensor? window=None, bool center=True, bool normalized=False, bool? onesided=None, int? length=None, bool return_complex=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t n_fft, ::std::optional<int64_t> hop_length, ::std::optional<int64_t> win_length, const ::std::optional<at::Tensor> & window, bool center, bool normalized, ::std::optional<bool> onesided, ::std::optional<int64_t> length, bool return_complex);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t n_fft, ::std::optional<int64_t> hop_length, ::std::optional<int64_t> win_length, const ::std::optional<at::Tensor> & window, bool center, bool normalized, ::std::optional<bool> onesided, ::std::optional<int64_t> length, bool return_complex);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/item.h b/lib/python3.12/site-packages/torch/include/ATen/ops/item.h
new file mode 100644
index 0000000000000000000000000000000000000000..40a7273d885d2f286bef65eaaf106ee756259482
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/item.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/item_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/kthvalue_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/kthvalue_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5fb7386712a7fe8a0f421f01f2909832ad6481a1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/kthvalue_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> kthvalue(const at::Tensor & self, int64_t k, at::Dimname dim, bool keepdim=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> kthvalue_out(at::Tensor & values, at::Tensor & indices, const at::Tensor & self, int64_t k, at::Dimname dim, bool keepdim=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> kthvalue_outf(const at::Tensor & self, int64_t k, at::Dimname dim, bool keepdim, at::Tensor & values, at::Tensor & indices);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/layer_norm_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/layer_norm_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1de5705b6046894c4b23f9e37ed1781481c80595
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/layer_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 layer_norm {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, double, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::layer_norm";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "layer_norm(Tensor input, SymInt[] normalized_shape, Tensor? weight=None, Tensor? bias=None, float eps=1e-05, bool cudnn_enable=True) -> Tensor";
+  static at::Tensor call(const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps, bool cudnn_enable);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps, bool cudnn_enable);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lerp.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lerp.h
new file mode 100644
index 0000000000000000000000000000000000000000..0f36e9f3fa99ca96eb93b0829fdc702beb80263c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lerp.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/lerp_ops.h>
+
+namespace at {
+
+
+// aten::lerp.Scalar_out(Tensor self, Tensor end, Scalar weight, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & lerp_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & end, const at::Scalar & weight) {
+    return at::_ops::lerp_Scalar_out::call(self, end, weight, out);
+}
+// aten::lerp.Scalar_out(Tensor self, Tensor end, Scalar weight, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & lerp_outf(const at::Tensor & self, const at::Tensor & end, const at::Scalar & weight, at::Tensor & out) {
+    return at::_ops::lerp_Scalar_out::call(self, end, weight, out);
+}
+
+// aten::lerp.Tensor_out(Tensor self, Tensor end, Tensor weight, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & lerp_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & end, const at::Tensor & weight) {
+    return at::_ops::lerp_Tensor_out::call(self, end, weight, out);
+}
+// aten::lerp.Tensor_out(Tensor self, Tensor end, Tensor weight, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & lerp_outf(const at::Tensor & self, const at::Tensor & end, const at::Tensor & weight, at::Tensor & out) {
+    return at::_ops::lerp_Tensor_out::call(self, end, weight, out);
+}
+
+// aten::lerp.Scalar(Tensor self, Tensor end, Scalar weight) -> Tensor
+inline at::Tensor lerp(const at::Tensor & self, const at::Tensor & end, const at::Scalar & weight) {
+    return at::_ops::lerp_Scalar::call(self, end, weight);
+}
+
+// aten::lerp.Tensor(Tensor self, Tensor end, Tensor weight) -> Tensor
+inline at::Tensor lerp(const at::Tensor & self, const at::Tensor & end, const at::Tensor & weight) {
+    return at::_ops::lerp_Tensor::call(self, end, weight);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lerp_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lerp_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a320092df1fdf8c7bf3eef483d5feddf1b83063c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lerp_meta_dispatch.h
@@ -0,0 +1,30 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor lerp(const at::Tensor & self, const at::Tensor & end, const at::Scalar & weight);
+TORCH_API at::Tensor & lerp_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & end, const at::Scalar & weight);
+TORCH_API at::Tensor & lerp_outf(const at::Tensor & self, const at::Tensor & end, const at::Scalar & weight, at::Tensor & out);
+TORCH_API at::Tensor & lerp_(at::Tensor & self, const at::Tensor & end, const at::Scalar & weight);
+TORCH_API at::Tensor lerp(const at::Tensor & self, const at::Tensor & end, const at::Tensor & weight);
+TORCH_API at::Tensor & lerp_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & end, const at::Tensor & weight);
+TORCH_API at::Tensor & lerp_outf(const at::Tensor & self, const at::Tensor & end, const at::Tensor & weight, at::Tensor & out);
+TORCH_API at::Tensor & lerp_(at::Tensor & self, const at::Tensor & end, const at::Tensor & weight);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/less_equal_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/less_equal_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..6c584448f0e92a318edcba610a21a07a4d7966d9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/less_equal_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 less_equal_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::less_equal";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "less_equal.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+};
+
+struct TORCH_API less_equal_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::less_equal";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "less_equal.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 less_equal_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::less_equal";
+  static constexpr const char* overload_name = "Tensor_out";
+  static constexpr const char* schema_str = "less_equal.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+};
+
+struct TORCH_API less_equal_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::less_equal";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "less_equal.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 less_equal__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::less_equal_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "less_equal_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Scalar & other);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Scalar & other);
+};
+
+struct TORCH_API less_equal__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::less_equal_";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "less_equal_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & other);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & other);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_det_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_det_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b7ec419f0b79876a4d1e51bf319baf120fc3ce64
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_det_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor linalg_det(const at::Tensor & A);
+TORCH_API at::Tensor & linalg_det_out(at::Tensor & out, const at::Tensor & A);
+TORCH_API at::Tensor & linalg_det_outf(const at::Tensor & A, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_eigvals_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_eigvals_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9abe253e72220aaac7e73bb330602ac655a3e2b4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_eigvals_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_eigvals {
+  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::linalg_eigvals";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_eigvals(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 linalg_eigvals_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::linalg_eigvals";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "linalg_eigvals.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_inv_ex_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_inv_ex_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..d8134b5e581e763bf120b2725101c233c1028dc8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_inv_ex_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_inv_ex {
+  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_inv_ex";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_inv_ex(Tensor A, *, bool check_errors=False) -> (Tensor inverse, Tensor info)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & A, bool check_errors);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, bool check_errors);
+};
+
+struct TORCH_API linalg_inv_ex_inverse {
+  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_inv_ex";
+  static constexpr const char* overload_name = "inverse";
+  static constexpr const char* schema_str = "linalg_inv_ex.inverse(Tensor A, *, bool check_errors=False, Tensor(a!) inverse, Tensor(b!) info) -> (Tensor(a!) inverse, Tensor(b!) info)";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & A, bool check_errors, at::Tensor & inverse, at::Tensor & info);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, bool check_errors, at::Tensor & inverse, at::Tensor & info);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1141aca785fdcec5456a177fe8ce9e5fafabce2b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_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> linalg_ldl_factor(const at::Tensor & self, bool hermitian=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> linalg_ldl_factor_out(at::Tensor & LD, at::Tensor & pivots, const at::Tensor & self, bool hermitian=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> linalg_ldl_factor_outf(const at::Tensor & self, bool hermitian, at::Tensor & LD, at::Tensor & pivots);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lstsq_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lstsq_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9f77d69a77c87fb062678b16e81020eeda421b72
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lstsq_cuda_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> linalg_lstsq_out(at::Tensor & solution, at::Tensor & residuals, at::Tensor & rank, at::Tensor & singular_values, const at::Tensor & self, const at::Tensor & b, ::std::optional<double> rcond=::std::nullopt, ::std::optional<c10::string_view> driver=::std::nullopt);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> linalg_lstsq_outf(const at::Tensor & self, const at::Tensor & b, ::std::optional<double> rcond, ::std::optional<c10::string_view> driver, at::Tensor & solution, at::Tensor & residuals, at::Tensor & rank, at::Tensor & singular_values);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu.h
new file mode 100644
index 0000000000000000000000000000000000000000..fc6b1a28b34022ade308e6923077bc605525c670
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu.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_lu_ops.h>
+
+namespace at {
+
+
+// aten::linalg_lu(Tensor A, *, bool pivot=True) -> (Tensor P, Tensor L, Tensor U)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> linalg_lu(const at::Tensor & A, bool pivot=true) {
+    return at::_ops::linalg_lu::call(A, pivot);
+}
+
+// aten::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)
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linalg_lu_out(at::Tensor & P, at::Tensor & L, at::Tensor & U, const at::Tensor & A, bool pivot=true) {
+    return at::_ops::linalg_lu_out::call(A, pivot, P, L, U);
+}
+// aten::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)
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linalg_lu_outf(const at::Tensor & A, bool pivot, at::Tensor & P, at::Tensor & L, at::Tensor & U) {
+    return at::_ops::linalg_lu_out::call(A, pivot, P, L, U);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_solve_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_solve_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..deecd795b72a7db02f83fcf879633760898d3b99
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_solve_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor linalg_lu_solve(const at::Tensor & LU, const at::Tensor & pivots, const at::Tensor & B, bool left=true, bool adjoint=false);
+TORCH_API at::Tensor & linalg_lu_solve_out(at::Tensor & out, const at::Tensor & LU, const at::Tensor & pivots, const at::Tensor & B, bool left=true, bool adjoint=false);
+TORCH_API at::Tensor & linalg_lu_solve_outf(const at::Tensor & LU, const at::Tensor & pivots, const at::Tensor & B, bool left, bool adjoint, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_solve_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_solve_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..6e966df7aad2291123656d76440c88c85bbbff26
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_solve_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_solve {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_lu_solve";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_lu_solve(Tensor LU, Tensor pivots, Tensor B, *, bool left=True, bool adjoint=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & LU, const at::Tensor & pivots, const at::Tensor & B, bool left, bool adjoint);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & LU, const at::Tensor & pivots, const at::Tensor & B, bool left, bool adjoint);
+};
+
+struct TORCH_API linalg_lu_solve_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const at::Tensor &, bool, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_lu_solve";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "linalg_lu_solve.out(Tensor LU, Tensor pivots, Tensor B, *, bool left=True, bool adjoint=False, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & LU, const at::Tensor & pivots, const at::Tensor & B, bool left, bool adjoint, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & LU, const at::Tensor & pivots, const at::Tensor & B, bool left, bool adjoint, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matmul.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matmul.h
new file mode 100644
index 0000000000000000000000000000000000000000..8487bee57424867ac5c4e191388568fe8a4cc40d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matmul.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_matmul_ops.h>
+
+namespace at {
+
+
+// aten::linalg_matmul(Tensor self, Tensor other) -> Tensor
+inline at::Tensor linalg_matmul(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::linalg_matmul::call(self, other);
+}
+
+// aten::linalg_matmul.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matmul_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::linalg_matmul_out::call(self, other, out);
+}
+// aten::linalg_matmul.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matmul_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::linalg_matmul_out::call(self, other, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_norm_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_norm_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2339e1dd3024c89cd8552a03319e65a57cf26883
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_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 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(at::Tensor & out, 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_outf(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(at::Tensor & out, 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_outf(const at::Tensor & self, c10::string_view ord, at::IntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_rank.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_rank.h
new file mode 100644
index 0000000000000000000000000000000000000000..53c07c2a4c195407e205fd75f3834950614a2c3c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_rank.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/linalg_matrix_rank_ops.h>
+
+namespace at {
+
+
+// aten::linalg_matrix_rank.atol_rtol_tensor(Tensor input, *, Tensor? atol=None, Tensor? rtol=None, bool hermitian=False) -> Tensor
+inline at::Tensor linalg_matrix_rank(const at::Tensor & input, const ::std::optional<at::Tensor> & atol={}, const ::std::optional<at::Tensor> & rtol={}, bool hermitian=false) {
+    return at::_ops::linalg_matrix_rank_atol_rtol_tensor::call(input, atol, rtol, hermitian);
+}
+
+// aten::linalg_matrix_rank.atol_rtol_tensor_out(Tensor input, *, Tensor? atol=None, Tensor? rtol=None, bool hermitian=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matrix_rank_out(at::Tensor & out, const at::Tensor & input, const ::std::optional<at::Tensor> & atol={}, const ::std::optional<at::Tensor> & rtol={}, bool hermitian=false) {
+    return at::_ops::linalg_matrix_rank_atol_rtol_tensor_out::call(input, atol, rtol, hermitian, out);
+}
+// aten::linalg_matrix_rank.atol_rtol_tensor_out(Tensor input, *, Tensor? atol=None, Tensor? rtol=None, bool hermitian=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matrix_rank_outf(const at::Tensor & input, const ::std::optional<at::Tensor> & atol, const ::std::optional<at::Tensor> & rtol, bool hermitian, at::Tensor & out) {
+    return at::_ops::linalg_matrix_rank_atol_rtol_tensor_out::call(input, atol, rtol, hermitian, out);
+}
+
+// aten::linalg_matrix_rank.atol_rtol_float(Tensor self, *, float? atol=None, float? rtol=None, bool hermitian=False) -> Tensor
+inline at::Tensor linalg_matrix_rank(const at::Tensor & self, ::std::optional<double> atol, ::std::optional<double> rtol, bool hermitian=false) {
+    return at::_ops::linalg_matrix_rank_atol_rtol_float::call(self, atol, rtol, hermitian);
+}
+
+// aten::linalg_matrix_rank.atol_rtol_float_out(Tensor self, *, float? atol=None, float? rtol=None, bool hermitian=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matrix_rank_out(at::Tensor & out, const at::Tensor & self, ::std::optional<double> atol, ::std::optional<double> rtol, bool hermitian=false) {
+    return at::_ops::linalg_matrix_rank_atol_rtol_float_out::call(self, atol, rtol, hermitian, out);
+}
+// aten::linalg_matrix_rank.atol_rtol_float_out(Tensor self, *, float? atol=None, float? rtol=None, bool hermitian=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matrix_rank_outf(const at::Tensor & self, ::std::optional<double> atol, ::std::optional<double> rtol, bool hermitian, at::Tensor & out) {
+    return at::_ops::linalg_matrix_rank_atol_rtol_float_out::call(self, atol, rtol, hermitian, out);
+}
+
+// aten::linalg_matrix_rank(Tensor self, float tol, bool hermitian=False) -> Tensor
+inline at::Tensor linalg_matrix_rank(const at::Tensor & self, double tol, bool hermitian=false) {
+    return at::_ops::linalg_matrix_rank::call(self, tol, hermitian);
+}
+
+// aten::linalg_matrix_rank.out(Tensor self, float tol, bool hermitian=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matrix_rank_out(at::Tensor & out, const at::Tensor & self, double tol, bool hermitian=false) {
+    return at::_ops::linalg_matrix_rank_out::call(self, tol, hermitian, out);
+}
+// aten::linalg_matrix_rank.out(Tensor self, float tol, bool hermitian=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matrix_rank_outf(const at::Tensor & self, double tol, bool hermitian, at::Tensor & out) {
+    return at::_ops::linalg_matrix_rank_out::call(self, tol, hermitian, out);
+}
+
+// aten::linalg_matrix_rank.tol_tensor(Tensor input, Tensor tol, bool hermitian=False) -> Tensor
+inline at::Tensor linalg_matrix_rank(const at::Tensor & input, const at::Tensor & tol, bool hermitian=false) {
+    return at::_ops::linalg_matrix_rank_tol_tensor::call(input, tol, hermitian);
+}
+
+// aten::linalg_matrix_rank.out_tol_tensor(Tensor input, Tensor tol, bool hermitian=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matrix_rank_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & tol, bool hermitian=false) {
+    return at::_ops::linalg_matrix_rank_out_tol_tensor::call(input, tol, hermitian, out);
+}
+// aten::linalg_matrix_rank.out_tol_tensor(Tensor input, Tensor tol, bool hermitian=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_matrix_rank_outf(const at::Tensor & input, const at::Tensor & tol, bool hermitian, at::Tensor & out) {
+    return at::_ops::linalg_matrix_rank_out_tol_tensor::call(input, tol, hermitian, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_multi_dot_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_multi_dot_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fe6e94966dbd632d502f819c4a2a5da2ca9050f2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_multi_dot_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor linalg_multi_dot(at::TensorList tensors);
+TORCH_API at::Tensor & linalg_multi_dot_out(at::Tensor & out, at::TensorList tensors);
+TORCH_API at::Tensor & linalg_multi_dot_outf(at::TensorList tensors, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_solve_triangular_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_solve_triangular_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..4f03e1a727398aebeb8b42459811b2026eafd5d2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_solve_triangular_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_solve_triangular_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, bool, bool, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_solve_triangular";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "linalg_solve_triangular.out(Tensor self, Tensor B, *, bool upper, bool left=True, bool unitriangular=False, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & B, bool upper, bool left, bool unitriangular, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & B, bool upper, bool left, bool unitriangular, at::Tensor & out);
+};
+
+struct TORCH_API linalg_solve_triangular {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, bool, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_solve_triangular";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_solve_triangular(Tensor self, Tensor B, *, bool upper, bool left=True, bool unitriangular=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & B, bool upper, bool left, bool unitriangular);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & B, bool upper, bool left, bool unitriangular);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_tensorinv_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_tensorinv_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b71771dad48bd3997baee02183432c38e43ee1c6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_tensorinv_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 linalg_tensorinv(const at::Tensor & self, int64_t ind=2);
+TORCH_API at::Tensor & linalg_tensorinv_out(const at::Tensor & self, int64_t ind, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_tensorsolve.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_tensorsolve.h
new file mode 100644
index 0000000000000000000000000000000000000000..c0b8d9add3113f90d853a88217c6f91a61366b47
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_tensorsolve.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_tensorsolve_ops.h>
+
+namespace at {
+
+
+// aten::linalg_tensorsolve(Tensor self, Tensor other, int[]? dims=None) -> Tensor
+inline at::Tensor linalg_tensorsolve(const at::Tensor & self, const at::Tensor & other, at::OptionalIntArrayRef dims=::std::nullopt) {
+    return at::_ops::linalg_tensorsolve::call(self, other, dims);
+}
+
+// aten::linalg_tensorsolve.out(Tensor self, Tensor other, int[]? dims=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_tensorsolve_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other, at::OptionalIntArrayRef dims=::std::nullopt) {
+    return at::_ops::linalg_tensorsolve_out::call(self, other, dims, out);
+}
+// aten::linalg_tensorsolve.out(Tensor self, Tensor other, int[]? dims=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_tensorsolve_outf(const at::Tensor & self, const at::Tensor & other, at::OptionalIntArrayRef dims, at::Tensor & out) {
+    return at::_ops::linalg_tensorsolve_out::call(self, other, dims, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_vector_norm_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_vector_norm_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2ff37d95ddb644203006c921aa28a91306f928ed
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_vector_norm_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor linalg_vector_norm(const at::Tensor & self, const at::Scalar & ord=2, at::OptionalIntArrayRef dim=::std::nullopt, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & linalg_vector_norm_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & ord=2, at::OptionalIntArrayRef dim=::std::nullopt, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & linalg_vector_norm_outf(const at::Tensor & self, const at::Scalar & ord, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log2_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log2_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0c3163a4ff9fd5eb0c470e199be0f7aec2fb4f50
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log2_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 log2(const at::Tensor & self);
+TORCH_API at::Tensor & log2_(at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..5be5ee8ff8c7b99f16ce1820141aec9d8edbbf68
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_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/log_sigmoid_backward_ops.h>
+
+namespace at {
+
+
+// aten::log_sigmoid_backward.grad_input(Tensor grad_output, Tensor self, Tensor buffer, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & log_sigmoid_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & buffer) {
+    return at::_ops::log_sigmoid_backward_grad_input::call(grad_output, self, buffer, grad_input);
+}
+// aten::log_sigmoid_backward.grad_input(Tensor grad_output, Tensor self, Tensor buffer, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & log_sigmoid_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & buffer, at::Tensor & grad_input) {
+    return at::_ops::log_sigmoid_backward_grad_input::call(grad_output, self, buffer, grad_input);
+}
+
+// aten::log_sigmoid_backward(Tensor grad_output, Tensor self, Tensor buffer) -> Tensor
+inline at::Tensor log_sigmoid_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & buffer) {
+    return at::_ops::log_sigmoid_backward::call(grad_output, self, buffer);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logaddexp_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logaddexp_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..085c5c9f76013ae71668447cb7834c8a691f1832
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logaddexp_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 logaddexp(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & logaddexp_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & logaddexp_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/logcumsumexp_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logcumsumexp_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a54fa0356c89018bec592729180662ec5ca515c6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logcumsumexp_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 logcumsumexp {
+  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::logcumsumexp";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "logcumsumexp(Tensor self, int dim) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim);
+};
+
+struct TORCH_API logcumsumexp_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::logcumsumexp";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "logcumsumexp.out(Tensor self, int dim, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, int64_t dim, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, at::Tensor & out);
+};
+
+struct TORCH_API logcumsumexp_dimname {
+  using schema = at::Tensor (const at::Tensor &, at::Dimname);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logcumsumexp";
+  static constexpr const char* overload_name = "dimname";
+  static constexpr const char* schema_str = "logcumsumexp.dimname(Tensor self, Dimname dim) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::Dimname dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname dim);
+};
+
+struct TORCH_API logcumsumexp_dimname_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Dimname, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::logcumsumexp";
+  static constexpr const char* overload_name = "dimname_out";
+  static constexpr const char* schema_str = "logcumsumexp.dimname_out(Tensor self, Dimname dim, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Dimname dim, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname dim, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logical_xor_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_xor_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..2e68262ec0d1890cb2580619166f2f06c0c072c1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_xor_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 logical_xor {
+  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::logical_xor";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "logical_xor(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 logical_xor_ {
+  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::logical_xor_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "logical_xor_(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 logical_xor_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::logical_xor";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "logical_xor.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logit_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logit_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c092087da591448ed5c86fd23ec519a15d12b2c4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logit_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 logit_backward(const at::Tensor & grad_output, const at::Tensor & self, ::std::optional<double> eps=::std::nullopt);
+TORCH_API at::Tensor & logit_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, ::std::optional<double> eps=::std::nullopt);
+TORCH_API at::Tensor & logit_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, ::std::optional<double> eps, at::Tensor & grad_input);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logspace_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logspace_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9c9a1134679edc2c02aa0e90c46c9430717ff5da
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logspace_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 & logspace_out(at::Tensor & out, const at::Scalar & start, const at::Scalar & end, int64_t steps, double base=10.0);
+TORCH_API at::Tensor & logspace_outf(const at::Scalar & start, const at::Scalar & end, int64_t steps, double base, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..29432b17165870eb047ad7c50146606efd5919c0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_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 __lshift__(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & __ilshift__(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor __lshift__(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & __ilshift__(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e75d05f6734819a6c04965dabd8bcf4b3b737568
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_meta_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 meta {
+
+TORCH_API at::Tensor & __ilshift__(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & __ilshift__(at::Tensor & self, const at::Tensor & other);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9d97dca8518b638b8cd24a90cddbed95c2decbce
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_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 & __lshift___Scalar_out(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor __lshift__(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & __ilshift__(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & __lshift___Tensor_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor __lshift__(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & __ilshift__(at::Tensor & self, const at::Tensor & other);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lstm_mps_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lstm_mps_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d71fb6b03d19e081867cd074b18cb47ee363e0a4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lstm_mps_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 void lstm_mps_backward_out(const ::std::optional<at::Tensor> & grad_y, const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, const at::Tensor & z_state, const at::Tensor & cell_state_fwd, const at::Tensor & input, const at::Tensor & layersOutputs, at::TensorList hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional, bool batch_first, at::Tensor & out0, at::TensorList out1, at::TensorList out2);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lstm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lstm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..443568916ddcee2e5a61a7010d45c8bb89ad4c82
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lstm_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> lstm(const at::Tensor & input, at::TensorList hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional, bool batch_first);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> lstm(const at::Tensor & data, const at::Tensor & batch_sizes, at::TensorList hx, at::TensorList params, bool has_biases, int64_t num_layers, double dropout, bool train, bool bidirectional);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lt_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lt_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8ee8178d952a2e0e027d12c27368194b465b7d66
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lt_cuda_dispatch.h
@@ -0,0 +1,30 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor lt(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & lt_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & lt_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & lt_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor lt(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & lt_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & lt_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & lt_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lu_unpack_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lu_unpack_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..33a48f21e220762c6041677d52a654dffef51ad9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lu_unpack_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> lu_unpack(const at::Tensor & LU_data, const at::Tensor & LU_pivots, bool unpack_data=true, bool unpack_pivots=true);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> lu_unpack_out(at::Tensor & P, at::Tensor & L, at::Tensor & U, const at::Tensor & LU_data, const at::Tensor & LU_pivots, bool unpack_data=true, bool unpack_pivots=true);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> lu_unpack_outf(const at::Tensor & LU_data, const at::Tensor & LU_pivots, bool unpack_data, bool unpack_pivots, at::Tensor & P, at::Tensor & L, at::Tensor & U);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mH_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mH_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..ffc7885317db1d98258672418f6dd4bea66b9eea
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mH_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 mH(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/masked_scatter_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_scatter_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b30b8328e0b95d3936bd6594ff051be5be27cdcf
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_scatter_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 at::Tensor masked_scatter_backward_symint(const at::Tensor & grad_output, const at::Tensor & mask, c10::SymIntArrayRef sizes);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/matrix_exp_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/matrix_exp_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8b5bffb06369a87d9ff9070a2e87e7b2b9637257
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/matrix_exp_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 matrix_exp(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..c8649c8c0ac7d2b00094df80d65c33759b2fd205
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_backward_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API max_pool2d_backward {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::max_pool2d_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "max_pool2d_backward(Tensor grad_output, Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad_output, const at::Tensor & 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 & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+};
+
+struct TORCH_API max_pool2d_backward_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::max_pool2d_backward";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "max_pool2d_backward.out(Tensor grad_output, 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!)";
+  static at::Tensor & call(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..816c8a69fd2e008b020d74acc25d90fb0307a640
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_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 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);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cb106bc85639df0a72a06927a60a95d2d0990068
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_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 max_pool3d(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, bool ceil_mode=false);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4f8056e945d5ef8cd6dd9f034257b858b59f5f41
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> max_pool3d_with_indices_cpu(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 ::std::tuple<at::Tensor &,at::Tensor &> max_pool3d_with_indices_out_cpu(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out, at::Tensor & indices);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> max_pool3d_with_indices_cuda(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 ::std::tuple<at::Tensor &,at::Tensor &> max_pool3d_with_indices_out_cuda(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out, at::Tensor & indices);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_unpool2d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_unpool2d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..3d456a4c064bccd975c8e29e43538a5ee53235e3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_unpool2d_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 max_unpooling2d_forward_cpu(const at::Tensor & self, const at::Tensor & indices, at::IntArrayRef output_size);
+TORCH_API at::Tensor & max_unpooling2d_forward_out_cpu(const at::Tensor & self, const at::Tensor & indices, at::IntArrayRef output_size, at::Tensor & out);
+TORCH_API at::Tensor max_unpooling2d_forward_cuda(const at::Tensor & self, const at::Tensor & indices, at::IntArrayRef output_size);
+TORCH_API at::Tensor & max_unpooling2d_forward_out_cuda(const at::Tensor & self, const at::Tensor & indices, at::IntArrayRef output_size, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_unpool3d_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_unpool3d_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..24efbc157c4a474569e90b9c1241265e267d5efd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_unpool3d_cpu_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor max_unpool3d(const at::Tensor & self, const at::Tensor & indices, at::IntArrayRef output_size, at::IntArrayRef stride, at::IntArrayRef padding);
+TORCH_API at::Tensor max_unpool3d_symint(const at::Tensor & self, const at::Tensor & indices, c10::SymIntArrayRef output_size, at::IntArrayRef stride, at::IntArrayRef padding);
+TORCH_API at::Tensor & max_unpool3d_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & indices, at::IntArrayRef output_size, at::IntArrayRef stride, at::IntArrayRef padding);
+TORCH_API at::Tensor & max_unpool3d_outf(const at::Tensor & self, const at::Tensor & indices, at::IntArrayRef output_size, at::IntArrayRef stride, at::IntArrayRef padding, at::Tensor & out);
+TORCH_API at::Tensor & max_unpool3d_symint_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & indices, c10::SymIntArrayRef output_size, at::IntArrayRef stride, at::IntArrayRef padding);
+TORCH_API at::Tensor & max_unpool3d_symint_outf(const at::Tensor & self, const at::Tensor & indices, c10::SymIntArrayRef output_size, at::IntArrayRef stride, at::IntArrayRef padding, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/meshgrid.h b/lib/python3.12/site-packages/torch/include/ATen/ops/meshgrid.h
new file mode 100644
index 0000000000000000000000000000000000000000..7267a9804be0e93475faa4d80c6ac6ab6cc7210f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/meshgrid.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/meshgrid_ops.h>
+
+namespace at {
+
+
+// aten::meshgrid(Tensor[] tensors) -> Tensor[]
+inline ::std::vector<at::Tensor> meshgrid(at::TensorList tensors) {
+    return at::_ops::meshgrid::call(tensors);
+}
+
+// aten::meshgrid.indexing(Tensor[] tensors, *, str indexing) -> Tensor[]
+inline ::std::vector<at::Tensor> meshgrid(at::TensorList tensors, c10::string_view indexing) {
+    return at::_ops::meshgrid_indexing::call(tensors, indexing);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_transpose_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_transpose_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..0cb2cf0ac64f0e98d08e9f3eca48ba6ee53ad95a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_transpose_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 & miopen_convolution_transpose_out_symint(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, at::Tensor & out);
+TORCH_API at::Tensor miopen_convolution_transpose(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef padding, at::IntArrayRef output_padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4f9f9ed1a67ed08a6a14dd2180ba72562c580c98
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_rnn_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 &,at::Tensor &,at::Tensor &> miopen_rnn_out(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, 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, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, at::Tensor & out4);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> miopen_rnn(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, 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);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a4789b5d0b14b81407476cbfd7d8aa8d121653bd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_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 mkldnn_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::mkldnn_linear_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "mkldnn_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 mkldnn_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::mkldnn_linear_backward";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "mkldnn_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/mkldnn_linear_backward_weights_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_backward_weights_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..ed16095f398cc8923ddf956e928e115c227d88c8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_backward_weights_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 &> mkldnn_linear_backward_weights_out(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & weight, bool bias_defined, at::Tensor & out0, at::Tensor & out1);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> mkldnn_linear_backward_weights(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & weight, bool bias_defined);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9501203b4a1fb3ef45dc26364da119f3ac570e78
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_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_linear_out(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & out);
+TORCH_API at::Tensor mkldnn_linear(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias={});
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool3d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool3d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8c0753e6eff02aa2b1e0c2046f5139fc65337a93
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool3d_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_max_pool3d_out(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out);
+TORCH_API at::Tensor mkldnn_max_pool3d(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, bool ceil_mode=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_rnn_layer.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_rnn_layer.h
new file mode 100644
index 0000000000000000000000000000000000000000..8c863658412f92eb5b6818a2f2e536c6619697ce
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_rnn_layer.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_rnn_layer_ops.h>
+
+namespace at {
+
+
+// aten::mkldnn_rnn_layer(Tensor input, Tensor weight0, Tensor weight1, Tensor weight2, Tensor weight3, Tensor hx_, Tensor cx_, bool reverse, int[] batch_sizes, int mode, int hidden_size, int num_layers, bool has_biases, bool bidirectional, bool batch_first, bool train) -> (Tensor, Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> mkldnn_rnn_layer(const at::Tensor & input, const at::Tensor & weight0, const at::Tensor & weight1, const at::Tensor & weight2, const at::Tensor & weight3, const at::Tensor & hx_, const at::Tensor & cx_, bool reverse, at::IntArrayRef batch_sizes, int64_t mode, int64_t hidden_size, int64_t num_layers, bool has_biases, bool bidirectional, bool batch_first, bool train) {
+    return at::_ops::mkldnn_rnn_layer::call(input, weight0, weight1, weight2, weight3, hx_, cx_, reverse, batch_sizes, mode, hidden_size, num_layers, has_biases, bidirectional, batch_first, train);
+}
+
+// aten::mkldnn_rnn_layer.out(Tensor input, Tensor weight0, Tensor weight1, Tensor weight2, Tensor weight3, Tensor hx_, Tensor cx_, bool reverse, int[] batch_sizes, int mode, int hidden_size, int num_layers, bool has_biases, bool bidirectional, bool batch_first, bool train, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2, Tensor(d!) out3) -> (Tensor(a!), Tensor(b!), Tensor(c!), Tensor(d!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> mkldnn_rnn_layer_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, const at::Tensor & input, const at::Tensor & weight0, const at::Tensor & weight1, const at::Tensor & weight2, const at::Tensor & weight3, const at::Tensor & hx_, const at::Tensor & cx_, bool reverse, at::IntArrayRef batch_sizes, int64_t mode, int64_t hidden_size, int64_t num_layers, bool has_biases, bool bidirectional, bool batch_first, bool train) {
+    return at::_ops::mkldnn_rnn_layer_out::call(input, weight0, weight1, weight2, weight3, hx_, cx_, reverse, batch_sizes, mode, hidden_size, num_layers, has_biases, bidirectional, batch_first, train, out0, out1, out2, out3);
+}
+// aten::mkldnn_rnn_layer.out(Tensor input, Tensor weight0, Tensor weight1, Tensor weight2, Tensor weight3, Tensor hx_, Tensor cx_, bool reverse, int[] batch_sizes, int mode, int hidden_size, int num_layers, bool has_biases, bool bidirectional, bool batch_first, bool train, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2, Tensor(d!) out3) -> (Tensor(a!), Tensor(b!), Tensor(c!), Tensor(d!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> mkldnn_rnn_layer_outf(const at::Tensor & input, const at::Tensor & weight0, const at::Tensor & weight1, const at::Tensor & weight2, const at::Tensor & weight3, const at::Tensor & hx_, const at::Tensor & cx_, bool reverse, at::IntArrayRef batch_sizes, int64_t mode, int64_t hidden_size, int64_t num_layers, bool has_biases, bool bidirectional, bool batch_first, bool train, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3) {
+    return at::_ops::mkldnn_rnn_layer_out::call(input, weight0, weight1, weight2, weight3, hx_, cx_, reverse, batch_sizes, mode, hidden_size, num_layers, has_biases, bidirectional, batch_first, train, out0, out1, out2, out3);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_rnn_layer_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_rnn_layer_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8eee96e8c6150b72f9cad8299d0b182612559880
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_rnn_layer_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> mkldnn_rnn_layer_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, const at::Tensor & input, const at::Tensor & weight0, const at::Tensor & weight1, const at::Tensor & weight2, const at::Tensor & weight3, const at::Tensor & hx_, const at::Tensor & cx_, bool reverse, at::IntArrayRef batch_sizes, int64_t mode, int64_t hidden_size, int64_t num_layers, bool has_biases, bool bidirectional, bool batch_first, bool train);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> mkldnn_rnn_layer_outf(const at::Tensor & input, const at::Tensor & weight0, const at::Tensor & weight1, const at::Tensor & weight2, const at::Tensor & weight3, const at::Tensor & hx_, const at::Tensor & cx_, bool reverse, at::IntArrayRef batch_sizes, int64_t mode, int64_t hidden_size, int64_t num_layers, bool has_biases, bool bidirectional, bool batch_first, bool train, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mul_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mul_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..620bb8fd62954006b2d7873d261a5fb463cc601f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mul_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 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 meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/multi_margin_loss_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/multi_margin_loss_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b08ab12dc87d74c0c011ede3a06fef717344bc2c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/multi_margin_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 multi_margin_loss(const at::Tensor & self, const at::Tensor & target, const at::Scalar & p=1, const at::Scalar & margin=1, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean);
+TORCH_API at::Tensor & multi_margin_loss_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & target, const at::Scalar & p=1, const at::Scalar & margin=1, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean);
+TORCH_API at::Tensor & multi_margin_loss_outf(const at::Tensor & self, const at::Tensor & target, const at::Scalar & p, const at::Scalar & margin, const ::std::optional<at::Tensor> & weight, int64_t reduction, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_loss_forward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_loss_forward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c3a721f4cab31833c273e9a3d6d7e81cf7a20d13
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/multilabel_margin_loss_forward_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> multilabel_margin_loss_forward(const at::Tensor & self, const at::Tensor & target, int64_t reduction);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> multilabel_margin_loss_forward_out(at::Tensor & output, at::Tensor & is_target, const at::Tensor & self, const at::Tensor & target, int64_t reduction);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> multilabel_margin_loss_forward_outf(const at::Tensor & self, const at::Tensor & target, int64_t reduction, at::Tensor & output, at::Tensor & is_target);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/multinomial.h b/lib/python3.12/site-packages/torch/include/ATen/ops/multinomial.h
new file mode 100644
index 0000000000000000000000000000000000000000..639a0485b033b21d74809d4eb03d2095c43e624b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/multinomial.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/multinomial_ops.h>
+
+namespace at {
+
+
+// aten::multinomial.out(Tensor self, int num_samples, bool replacement=False, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & multinomial_out(at::Tensor & out, const at::Tensor & self, int64_t num_samples, bool replacement=false, ::std::optional<at::Generator> generator=::std::nullopt) {
+    return at::_ops::multinomial_out::call(self, num_samples, replacement, generator, out);
+}
+// aten::multinomial.out(Tensor self, int num_samples, bool replacement=False, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & multinomial_outf(const at::Tensor & self, int64_t num_samples, bool replacement, ::std::optional<at::Generator> generator, at::Tensor & out) {
+    return at::_ops::multinomial_out::call(self, num_samples, replacement, generator, out);
+}
+
+// aten::multinomial(Tensor self, int num_samples, bool replacement=False, *, Generator? generator=None) -> Tensor
+inline at::Tensor multinomial(const at::Tensor & self, int64_t num_samples, bool replacement=false, ::std::optional<at::Generator> generator=::std::nullopt) {
+    return at::_ops::multinomial::call(self, num_samples, replacement, generator);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mv.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mv.h
new file mode 100644
index 0000000000000000000000000000000000000000..dcdfa174b4aa334a03f3712537a1f4b173e3da3f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mv.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/mv_ops.h>
+
+namespace at {
+
+
+// aten::mv(Tensor self, Tensor vec) -> Tensor
+inline at::Tensor mv(const at::Tensor & self, const at::Tensor & vec) {
+    return at::_ops::mv::call(self, vec);
+}
+
+// aten::mv.out(Tensor self, Tensor vec, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & mv_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & vec) {
+    return at::_ops::mv_out::call(self, vec, out);
+}
+// aten::mv.out(Tensor self, Tensor vec, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & mv_outf(const at::Tensor & self, const at::Tensor & vec, at::Tensor & out) {
+    return at::_ops::mv_out::call(self, vec, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mv_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mv_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..e1d8aad0b3786c8e84c7ad1509e446d6b4cc557d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mv_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 mv(const at::Tensor & self, const at::Tensor & vec);
+TORCH_API at::Tensor & mv_out(const at::Tensor & self, const at::Tensor & vec, at::Tensor & out);
+TORCH_API at::Tensor mv_sparse(const at::Tensor & self, const at::Tensor & vec);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nanmedian_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nanmedian_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..82381d86c84ba66ea9bc76758b25ebf52b30c522
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nanmedian_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 nanmedian {
+  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::nanmedian";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "nanmedian(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 nanmedian_dim {
+  using schema = ::std::tuple<at::Tensor,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::nanmedian";
+  static constexpr const char* overload_name = "dim";
+  static constexpr const char* schema_str = "nanmedian.dim(Tensor self, int dim, bool keepdim=False) -> (Tensor values, Tensor indices)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self, int64_t dim, bool keepdim);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, bool keepdim);
+};
+
+struct TORCH_API nanmedian_dim_values {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const at::Tensor &, int64_t, bool, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::nanmedian";
+  static constexpr const char* overload_name = "dim_values";
+  static constexpr const char* schema_str = "nanmedian.dim_values(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) values, Tensor(b!) indices) -> (Tensor(a!) values, Tensor(b!) indices)";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & self, int64_t dim, bool keepdim, at::Tensor & values, at::Tensor & indices);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, bool keepdim, at::Tensor & values, at::Tensor & indices);
+};
+
+struct TORCH_API nanmedian_names_dim {
+  using schema = ::std::tuple<at::Tensor,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::nanmedian";
+  static constexpr const char* overload_name = "names_dim";
+  static constexpr const char* schema_str = "nanmedian.names_dim(Tensor self, Dimname dim, bool keepdim=False) -> (Tensor values, Tensor indices)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self, at::Dimname dim, bool keepdim);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname dim, bool keepdim);
+};
+
+struct TORCH_API nanmedian_names_dim_values {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const at::Tensor &, at::Dimname, bool, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::nanmedian";
+  static constexpr const char* overload_name = "names_dim_values";
+  static constexpr const char* schema_str = "nanmedian.names_dim_values(Tensor self, Dimname dim, bool keepdim=False, *, Tensor(a!) values, Tensor(b!) indices) -> (Tensor(a!) values, Tensor(b!) indices)";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & self, at::Dimname dim, bool keepdim, at::Tensor & values, at::Tensor & indices);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname dim, bool keepdim, at::Tensor & values, at::Tensor & indices);
+};
+
+struct TORCH_API nanmedian_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::nanmedian";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "nanmedian.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/nanquantile.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nanquantile.h
new file mode 100644
index 0000000000000000000000000000000000000000..6f7df6d27d0672f27be4c5b9fb141e96527dcc52
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nanquantile.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/nanquantile_ops.h>
+
+namespace at {
+
+
+// aten::nanquantile(Tensor self, Tensor q, int? dim=None, bool keepdim=False, *, str interpolation='linear') -> Tensor
+inline at::Tensor nanquantile(const at::Tensor & self, const at::Tensor & q, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false, c10::string_view interpolation="linear") {
+    return at::_ops::nanquantile::call(self, q, dim, keepdim, interpolation);
+}
+
+// aten::nanquantile.out(Tensor self, Tensor q, int? dim=None, bool keepdim=False, *, str interpolation='linear', Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & nanquantile_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & q, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false, c10::string_view interpolation="linear") {
+    return at::_ops::nanquantile_out::call(self, q, dim, keepdim, interpolation, out);
+}
+// aten::nanquantile.out(Tensor self, Tensor q, int? dim=None, bool keepdim=False, *, str interpolation='linear', Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & nanquantile_outf(const at::Tensor & self, const at::Tensor & q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation, at::Tensor & out) {
+    return at::_ops::nanquantile_out::call(self, q, dim, keepdim, interpolation, out);
+}
+
+// aten::nanquantile.scalar(Tensor self, float q, int? dim=None, bool keepdim=False, *, str interpolation='linear') -> Tensor
+inline at::Tensor nanquantile(const at::Tensor & self, double q, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false, c10::string_view interpolation="linear") {
+    return at::_ops::nanquantile_scalar::call(self, q, dim, keepdim, interpolation);
+}
+
+// aten::nanquantile.scalar_out(Tensor self, float q, int? dim=None, bool keepdim=False, *, str interpolation='linear', Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & nanquantile_out(at::Tensor & out, const at::Tensor & self, double q, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false, c10::string_view interpolation="linear") {
+    return at::_ops::nanquantile_scalar_out::call(self, q, dim, keepdim, interpolation, out);
+}
+// aten::nanquantile.scalar_out(Tensor self, float q, int? dim=None, bool keepdim=False, *, str interpolation='linear', Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & nanquantile_outf(const at::Tensor & self, double q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation, at::Tensor & out) {
+    return at::_ops::nanquantile_scalar_out::call(self, q, dim, keepdim, interpolation, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nanquantile_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nanquantile_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..60a4cbbf87085e1921f21042957d01a7728ca6b5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nanquantile_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 nanquantile(const at::Tensor & self, const at::Tensor & q, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false, c10::string_view interpolation="linear");
+TORCH_API at::Tensor & nanquantile_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & q, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false, c10::string_view interpolation="linear");
+TORCH_API at::Tensor & nanquantile_outf(const at::Tensor & self, const at::Tensor & q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation, at::Tensor & out);
+TORCH_API at::Tensor nanquantile(const at::Tensor & self, double q, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false, c10::string_view interpolation="linear");
+TORCH_API at::Tensor & nanquantile_out(at::Tensor & out, const at::Tensor & self, double q, ::std::optional<int64_t> dim=::std::nullopt, bool keepdim=false, c10::string_view interpolation="linear");
+TORCH_API at::Tensor & nanquantile_outf(const at::Tensor & self, double q, ::std::optional<int64_t> dim, bool keepdim, c10::string_view interpolation, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nansum_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nansum_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..7d21d327983f5c00fc7df333a48db0e2e20345c1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nansum_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 nansum(const at::Tensor & self, at::OptionalIntArrayRef dim=::std::nullopt, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & nansum_out(const at::Tensor & self, at::OptionalIntArrayRef 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/narrow_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/narrow_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..86de7a20146d6adfe6e637dc6a44db45cc4ee195
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/narrow_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 narrow {
+  using schema = at::Tensor (const at::Tensor &, int64_t, c10::SymInt, c10::SymInt);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::narrow";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "narrow(Tensor(a) self, int dim, SymInt start, SymInt length) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, int64_t dim, c10::SymInt start, c10::SymInt length);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, c10::SymInt start, c10::SymInt length);
+};
+
+struct TORCH_API narrow_Tensor {
+  using schema = at::Tensor (const at::Tensor &, int64_t, const at::Tensor &, c10::SymInt);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::narrow";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "narrow.Tensor(Tensor(a) self, int dim, Tensor start, SymInt length) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, int64_t dim, const at::Tensor & start, c10::SymInt length);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, const at::Tensor & start, c10::SymInt length);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_group_norm_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_group_norm_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..a6f83a76be9397b7aa5e9a6b797e03f4ebcca85d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_group_norm_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 &> native_group_norm_backward_out_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, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+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);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_group_norm_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_group_norm_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1390c212cce5ea4272dc0e5a42e35af3724658b9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_group_norm_cuda_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> native_group_norm(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, int64_t N, int64_t C, int64_t HxW, int64_t group, double eps);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> native_group_norm_symint(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, c10::SymInt N, c10::SymInt C, c10::SymInt HxW, int64_t group, double eps);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nested_to_padded_tensor.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nested_to_padded_tensor.h
new file mode 100644
index 0000000000000000000000000000000000000000..b4b4777bff01b225cfc00ebda2475ef8900998eb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nested_to_padded_tensor.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/nested_to_padded_tensor_ops.h>
+
+namespace at {
+
+
+// aten::nested_to_padded_tensor(Tensor self, float padding, int[]? output_size=None) -> Tensor
+inline at::Tensor nested_to_padded_tensor(const at::Tensor & self, double padding, at::OptionalIntArrayRef output_size=::std::nullopt) {
+    return at::_ops::nested_to_padded_tensor::call(self, padding, output_size);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss.h
new file mode 100644
index 0000000000000000000000000000000000000000..8060eef0dde2b2b94f216ac0e22fc784eec82181
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/nll_loss_ops.h>
+
+namespace at {
+
+
+// aten::nll_loss.out(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & nll_loss_out(at::Tensor & out, 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) {
+    return at::_ops::nll_loss_out::call(self, target, weight, reduction, ignore_index, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & nll_loss_out(at::Tensor & out, 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) {
+    return at::_ops::nll_loss_out::call(self, target, weight, reduction, ignore_index, out);
+  }
+}
+
+// aten::nll_loss.out(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & nll_loss_outf(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, at::Tensor & out) {
+    return at::_ops::nll_loss_out::call(self, target, weight, reduction, ignore_index, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & nll_loss_outf(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, at::Tensor & out) {
+    return at::_ops::nll_loss_out::call(self, target, weight, reduction, ignore_index, out);
+  }
+}
+
+// aten::nll_loss.out(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & nll_loss_symint_out(at::Tensor & out, 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) {
+    return at::_ops::nll_loss_out::call(self, target, weight, reduction, ignore_index, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & nll_loss_out(at::Tensor & out, 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) {
+    return at::_ops::nll_loss_out::call(self, target, weight, reduction, ignore_index, out);
+  }
+}
+
+// aten::nll_loss.out(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & nll_loss_symint_outf(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, at::Tensor & out) {
+    return at::_ops::nll_loss_out::call(self, target, weight, reduction, ignore_index, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & nll_loss_outf(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, at::Tensor & out) {
+    return at::_ops::nll_loss_out::call(self, target, weight, reduction, ignore_index, out);
+  }
+}
+
+// aten::nll_loss(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100) -> Tensor
+inline at::Tensor nll_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) {
+    return at::_ops::nll_loss::call(self, target, weight, reduction, ignore_index);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor nll_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) {
+    return at::_ops::nll_loss::call(self, target, weight, reduction, ignore_index);
+  }
+}
+
+// aten::nll_loss(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100) -> Tensor
+inline at::Tensor nll_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) {
+    return at::_ops::nll_loss::call(self, target, weight, reduction, ignore_index);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor nll_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) {
+    return at::_ops::nll_loss::call(self, target, weight, reduction, ignore_index);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..2892f057b779ca30779fff4f19ce8091ca538d37
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_backward.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/nll_loss_backward_ops.h>
+
+namespace at {
+
+
+// aten::nll_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & nll_loss_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & nll_loss_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+  }
+}
+
+// aten::nll_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & nll_loss_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight, at::Tensor & grad_input) {
+    return at::_ops::nll_loss_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & nll_loss_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight, at::Tensor & grad_input) {
+    return at::_ops::nll_loss_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+  }
+}
+
+// aten::nll_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & nll_loss_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & nll_loss_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+  }
+}
+
+// aten::nll_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & nll_loss_backward_symint_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight, at::Tensor & grad_input) {
+    return at::_ops::nll_loss_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & nll_loss_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight, at::Tensor & grad_input) {
+    return at::_ops::nll_loss_backward_grad_input::call(grad_output, self, target, weight, reduction, ignore_index, total_weight, grad_input);
+  }
+}
+
+// aten::nll_loss_backward(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight) -> Tensor
+inline at::Tensor nll_loss_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss_backward::call(grad_output, self, target, weight, reduction, ignore_index, total_weight);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor nll_loss_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss_backward::call(grad_output, self, target, weight, reduction, ignore_index, total_weight);
+  }
+}
+
+// aten::nll_loss_backward(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight) -> Tensor
+inline at::Tensor nll_loss_backward_symint(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss_backward::call(grad_output, self, target, weight, reduction, ignore_index, total_weight);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor nll_loss_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, const at::Tensor & total_weight) {
+    return at::_ops::nll_loss_backward::call(grad_output, self, target, weight, reduction, ignore_index, total_weight);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..51832a16556644d9a7658dd7b3814aa9a848534a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_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 nll_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);
+TORCH_API at::Tensor nll_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);
+TORCH_API at::Tensor & nll_loss_out(at::Tensor & out, 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);
+TORCH_API at::Tensor & nll_loss_outf(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, at::Tensor & out);
+TORCH_API at::Tensor & nll_loss_symint_out(at::Tensor & out, 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);
+TORCH_API at::Tensor & nll_loss_symint_outf(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_nd.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_nd.h
new file mode 100644
index 0000000000000000000000000000000000000000..acd3252757a9648221d4968af6ba8fe6ea64803b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_nd.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/nll_loss_nd_ops.h>
+
+namespace at {
+
+
+// aten::nll_loss_nd(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100) -> Tensor
+inline at::Tensor nll_loss_nd(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) {
+    return at::_ops::nll_loss_nd::call(self, target, weight, reduction, ignore_index);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor nll_loss_nd(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) {
+    return at::_ops::nll_loss_nd::call(self, target, weight, reduction, ignore_index);
+  }
+}
+
+// aten::nll_loss_nd(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100) -> Tensor
+inline at::Tensor nll_loss_nd_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) {
+    return at::_ops::nll_loss_nd::call(self, target, weight, reduction, ignore_index);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor nll_loss_nd(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) {
+    return at::_ops::nll_loss_nd::call(self, target, weight, reduction, ignore_index);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/not_equal_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/not_equal_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b76d118bdb2042372005fd9e5dcb544344977fee
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/not_equal_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 not_equal(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & not_equal_out(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & not_equal_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor not_equal(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & not_equal_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & not_equal_(at::Tensor & self, const at::Tensor & other);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/pairwise_distance_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/pairwise_distance_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..63dac93e77043f931d9942530e8291a2b2d1f33f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/pairwise_distance_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 pairwise_distance(const at::Tensor & x1, const at::Tensor & x2, double p=2, double eps=1e-06, bool keepdim=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/polar_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/polar_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..850a0bf13ecc2bd81d27653177796901ba475391
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/polar_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 polar {
+  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::polar";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "polar(Tensor abs, Tensor angle) -> Tensor";
+  static at::Tensor call(const at::Tensor & abs, const at::Tensor & angle);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & abs, const at::Tensor & angle);
+};
+
+struct TORCH_API polar_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::polar";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "polar.out(Tensor abs, Tensor angle, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & abs, const at::Tensor & angle, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & abs, const at::Tensor & angle, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/pow_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/pow_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1b3190e6074895d2a6737d98df5535129dfba886
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/pow_meta_dispatch.h
@@ -0,0 +1,33 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause 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 pow(const at::Tensor & self, const at::Tensor & exponent);
+TORCH_API at::Tensor & pow_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & exponent);
+TORCH_API at::Tensor & pow_outf(const at::Tensor & self, const at::Tensor & exponent, at::Tensor & out);
+TORCH_API at::Tensor & pow_(at::Tensor & self, const at::Tensor & exponent);
+TORCH_API at::Tensor pow(const at::Scalar & self, const at::Tensor & exponent);
+TORCH_API at::Tensor & pow_out(at::Tensor & out, const at::Scalar & self, const at::Tensor & exponent);
+TORCH_API at::Tensor & pow_outf(const at::Scalar & self, const at::Tensor & exponent, at::Tensor & out);
+TORCH_API at::Tensor pow(const at::Tensor & self, const at::Scalar & exponent);
+TORCH_API at::Tensor & pow_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & exponent);
+TORCH_API at::Tensor & pow_outf(const at::Tensor & self, const at::Scalar & exponent, at::Tensor & out);
+TORCH_API at::Tensor & pow_(at::Tensor & self, const at::Scalar & exponent);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/prod_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/prod_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..836707acf2992d7c2e2038e891327f67e8f40178
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/prod_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 prod(const at::Tensor & self, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor prod(const at::Tensor & self, int64_t dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & prod_out(at::Tensor & out, const at::Tensor & self, int64_t dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & prod_outf(const at::Tensor & self, int64_t dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor_dynamic_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor_dynamic_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a61f68c5c6a68185cbe1e1c2b4056ea8091045b3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantize_per_tensor_dynamic_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_tensor_dynamic {
+  using schema = at::Tensor (const at::Tensor &, at::ScalarType, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::quantize_per_tensor_dynamic";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "quantize_per_tensor_dynamic(Tensor self, ScalarType dtype, bool reduce_range) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::ScalarType dtype, bool reduce_range);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::ScalarType dtype, bool reduce_range);
+};
+
+struct TORCH_API quantize_per_tensor_dynamic_out {
+  using schema = at::Tensor & (const at::Tensor &, at::ScalarType, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::quantize_per_tensor_dynamic";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "quantize_per_tensor_dynamic.out(Tensor self, ScalarType dtype, bool reduce_range, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::ScalarType dtype, bool reduce_range, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::ScalarType dtype, bool reduce_range, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_batch_norm_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_batch_norm_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..91d7258b63d464d1cdd247161ff8e678f1193ac0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_batch_norm_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API quantized_batch_norm {
+  using schema = at::Tensor (const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<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::quantized_batch_norm";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "quantized_batch_norm(Tensor input, Tensor? weight, Tensor? bias, Tensor mean, Tensor var, float eps, float output_scale, int output_zero_point) -> Tensor";
+  static at::Tensor call(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & mean, const at::Tensor & var, double eps, double output_scale, int64_t output_zero_point);
+  static 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 & mean, const at::Tensor & var, double eps, double output_scale, int64_t output_zero_point);
+};
+
+struct TORCH_API quantized_batch_norm_out {
+  using schema = at::Tensor & (const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const 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::quantized_batch_norm";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "quantized_batch_norm.out(Tensor input, Tensor? weight, Tensor? bias, Tensor mean, Tensor var, float eps, float output_scale, int output_zero_point, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & mean, const at::Tensor & var, double eps, double output_scale, int64_t output_zero_point, at::Tensor & out);
+  static 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 & mean, const at::Tensor & var, double eps, double output_scale, int64_t output_zero_point, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_lstm_cell_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_lstm_cell_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..28ea7b94e5a8c02c59053e32f0aeb69cc42d58cc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_lstm_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 ::std::tuple<at::Tensor,at::Tensor> quantized_lstm_cell(const at::Tensor & input, at::TensorList hx, const at::Tensor & w_ih, const at::Tensor & w_hh, const at::Tensor & b_ih, const at::Tensor & b_hh, const at::Tensor & packed_ih, const at::Tensor & packed_hh, const at::Tensor & col_offsets_ih, const at::Tensor & col_offsets_hh, const at::Scalar & scale_ih, const at::Scalar & scale_hh, const at::Scalar & zero_point_ih, const at::Scalar & zero_point_hh);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/randint_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/randint_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..d3b871e0a7fcb56ce76ec985f04a5be10e068311
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/randint_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 randint {
+  using schema = at::Tensor (c10::SymInt, 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::randint";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "randint(SymInt high, SymInt[] size, *, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(c10::SymInt high, 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::SymInt high, 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 randint_generator {
+  using schema = at::Tensor (c10::SymInt, 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::randint";
+  static constexpr const char* overload_name = "generator";
+  static constexpr const char* schema_str = "randint.generator(SymInt high, SymInt[] size, *, Generator? generator, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(c10::SymInt high, 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::SymInt high, 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 randint_low {
+  using schema = at::Tensor (c10::SymInt, c10::SymInt, 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::randint";
+  static constexpr const char* overload_name = "low";
+  static constexpr const char* schema_str = "randint.low(SymInt low, SymInt high, SymInt[] size, *, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(c10::SymInt low, c10::SymInt high, 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::SymInt low, c10::SymInt high, 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 randint_low_generator {
+  using schema = at::Tensor (c10::SymInt, c10::SymInt, 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::randint";
+  static constexpr const char* overload_name = "low_generator";
+  static constexpr const char* schema_str = "randint.low_generator(SymInt low, SymInt high, SymInt[] size, *, Generator? generator, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(c10::SymInt low, c10::SymInt high, 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::SymInt low, c10::SymInt high, 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 randint_out {
+  using schema = at::Tensor & (c10::SymInt, c10::SymIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::randint";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "randint.out(SymInt high, SymInt[] size, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymInt high, c10::SymIntArrayRef size, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymInt high, c10::SymIntArrayRef size, at::Tensor & out);
+};
+
+struct TORCH_API randint_generator_out {
+  using schema = at::Tensor & (c10::SymInt, 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::randint";
+  static constexpr const char* overload_name = "generator_out";
+  static constexpr const char* schema_str = "randint.generator_out(SymInt high, SymInt[] size, *, Generator? generator, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymInt high, c10::SymIntArrayRef size, ::std::optional<at::Generator> generator, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymInt high, c10::SymIntArrayRef size, ::std::optional<at::Generator> generator, at::Tensor & out);
+};
+
+struct TORCH_API randint_low_out {
+  using schema = at::Tensor & (c10::SymInt, c10::SymInt, c10::SymIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::randint";
+  static constexpr const char* overload_name = "low_out";
+  static constexpr const char* schema_str = "randint.low_out(SymInt low, SymInt high, SymInt[] size, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymInt low, c10::SymInt high, c10::SymIntArrayRef size, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymInt low, c10::SymInt high, c10::SymIntArrayRef size, at::Tensor & out);
+};
+
+struct TORCH_API randint_low_generator_out {
+  using schema = at::Tensor & (c10::SymInt, c10::SymInt, 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::randint";
+  static constexpr const char* overload_name = "low_generator_out";
+  static constexpr const char* schema_str = "randint.low_generator_out(SymInt low, SymInt high, SymInt[] size, *, Generator? generator, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(c10::SymInt low, c10::SymInt high, c10::SymIntArrayRef size, ::std::optional<at::Generator> generator, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, c10::SymInt low, c10::SymInt high, c10::SymIntArrayRef size, ::std::optional<at::Generator> generator, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/random_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/random_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..d456376e2017cbb3132fc232b7360adf60a06e38
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/random_ops.h
@@ -0,0 +1,116 @@
+#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 random__from {
+  using schema = at::Tensor & (at::Tensor &, int64_t, ::std::optional<int64_t>, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::random_";
+  static constexpr const char* overload_name = "from";
+  static constexpr const char* schema_str = "random_.from(Tensor(a!) self, int from, int? to, *, Generator? generator=None) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, int64_t from, ::std::optional<int64_t> to, ::std::optional<at::Generator> generator);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, int64_t from, ::std::optional<int64_t> to, ::std::optional<at::Generator> generator);
+};
+
+struct TORCH_API random__to {
+  using schema = at::Tensor & (at::Tensor &, int64_t, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::random_";
+  static constexpr const char* overload_name = "to";
+  static constexpr const char* schema_str = "random_.to(Tensor(a!) self, int to, *, Generator? generator=None) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, int64_t to, ::std::optional<at::Generator> generator);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, int64_t to, ::std::optional<at::Generator> generator);
+};
+
+struct TORCH_API random_ {
+  using schema = at::Tensor & (at::Tensor &, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::random_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "random_(Tensor(a!) self, *, Generator? generator=None) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, ::std::optional<at::Generator> generator);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, ::std::optional<at::Generator> generator);
+};
+
+struct TORCH_API random_from_out {
+  using schema = at::Tensor & (const at::Tensor &, int64_t, ::std::optional<int64_t>, ::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::random";
+  static constexpr const char* overload_name = "from_out";
+  static constexpr const char* schema_str = "random.from_out(Tensor self, int from, int? to, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, int64_t from, ::std::optional<int64_t> to, ::std::optional<at::Generator> generator, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t from, ::std::optional<int64_t> to, ::std::optional<at::Generator> generator, at::Tensor & out);
+};
+
+struct TORCH_API random_from {
+  using schema = at::Tensor (const at::Tensor &, int64_t, ::std::optional<int64_t>, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::random";
+  static constexpr const char* overload_name = "from";
+  static constexpr const char* schema_str = "random.from(Tensor self, int from, int? to, *, Generator? generator=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t from, ::std::optional<int64_t> to, ::std::optional<at::Generator> generator);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t from, ::std::optional<int64_t> to, ::std::optional<at::Generator> generator);
+};
+
+struct TORCH_API random_to_out {
+  using schema = at::Tensor & (const at::Tensor &, int64_t, ::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::random";
+  static constexpr const char* overload_name = "to_out";
+  static constexpr const char* schema_str = "random.to_out(Tensor self, int to, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, int64_t to, ::std::optional<at::Generator> generator, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t to, ::std::optional<at::Generator> generator, at::Tensor & out);
+};
+
+struct TORCH_API random_to {
+  using schema = at::Tensor (const at::Tensor &, int64_t, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::random";
+  static constexpr const char* overload_name = "to";
+  static constexpr const char* schema_str = "random.to(Tensor self, int to, *, Generator? generator=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t to, ::std::optional<at::Generator> generator);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t to, ::std::optional<at::Generator> generator);
+};
+
+struct TORCH_API random_out {
+  using schema = at::Tensor & (const at::Tensor &, ::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::random";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "random.out(Tensor self, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, ::std::optional<at::Generator> generator, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, ::std::optional<at::Generator> generator, at::Tensor & out);
+};
+
+struct TORCH_API random {
+  using schema = at::Tensor (const at::Tensor &, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::random";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "random(Tensor self, *, Generator? generator=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, ::std::optional<at::Generator> generator);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, ::std::optional<at::Generator> generator);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/real_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/real_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..1e2d5ca1f782ed7c3f306933b2dc0fbbe3ab1d7a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/real_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 real(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/record_stream_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/record_stream_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..585178ba39e928faa2c639e95c52f9f7f4e02f47
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/record_stream_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 record_stream {
+  using schema = void (at::Tensor &, at::Stream);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::record_stream";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "record_stream(Tensor(a!) self, Stream s) -> ()";
+  static void call(at::Tensor & self, at::Stream s);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, at::Stream s);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad3d_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad3d_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..847d8525881b68ee68795da89da763edfdbc8e34
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad3d_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_reflection_pad3d : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, at::ArrayRef<int64_t> padding);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/remainder_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/remainder_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d29a4bee8b4d293a1d8a6840694f3acb3368c4d1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/remainder_cuda_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 cuda {
+
+TORCH_API at::Tensor remainder(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & remainder_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & remainder_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & remainder_(at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor remainder(const at::Scalar & self, const at::Tensor & other);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rename_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rename_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f09504bc5656c6c39124299e0ee16585090d8e5a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rename_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 & rename_(at::Tensor & self, ::std::optional<at::DimnameList> names);
+TORCH_API at::Tensor rename(const at::Tensor & self, ::std::optional<at::DimnameList> names);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cf4034a032c3516b1685c1f987194d17f155b245
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/repeat_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 repeat(const at::Tensor & self, at::IntArrayRef repeats);
+TORCH_API at::Tensor repeat_symint(const at::Tensor & self, c10::SymIntArrayRef repeats);
+TORCH_API at::Tensor & repeat_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef repeats);
+TORCH_API at::Tensor & repeat_outf(const at::Tensor & self, at::IntArrayRef repeats, at::Tensor & out);
+TORCH_API at::Tensor & repeat_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef repeats);
+TORCH_API at::Tensor & repeat_symint_outf(const at::Tensor & self, c10::SymIntArrayRef repeats, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad3d_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad3d_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..de8337e6364af2fc66c1834def206fecc3da2690
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad3d_backward_cpu_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor replication_pad3d_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor replication_pad3d_backward_symint(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & replication_pad3d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor & replication_pad3d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding, at::Tensor & grad_input);
+TORCH_API at::Tensor & replication_pad3d_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & replication_pad3d_backward_symint_outf(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & grad_input);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad3d_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad3d_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..7d73fb499eb600170ca5d0b1ae4b7f89945cf7c1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad3d_backward_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor replication_pad3d_backward_cpu(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor & replication_pad3d_backward_out_cpu(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding, at::Tensor & grad_input);
+TORCH_API at::Tensor replication_pad3d_backward_cuda(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor & replication_pad3d_backward_out_cuda(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding, at::Tensor & grad_input);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/resize_as_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/resize_as_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d13e4150e3df70a84b78cd0a3cb438060bfac307
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/resize_as_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 resize_as(const at::Tensor & self, const at::Tensor & the_template, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API const at::Tensor & resize_as_out(const at::Tensor & out, const at::Tensor & self, const at::Tensor & the_template, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API const at::Tensor & resize_as_outf(const at::Tensor & self, const at::Tensor & the_template, ::std::optional<at::MemoryFormat> memory_format, const at::Tensor & out);
+TORCH_API const at::Tensor & resize_as_(const at::Tensor & self, const at::Tensor & the_template, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/retain_grad.h b/lib/python3.12/site-packages/torch/include/ATen/ops/retain_grad.h
new file mode 100644
index 0000000000000000000000000000000000000000..7bf1335900dc202359bacb6075d4b24136d8325b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/retain_grad.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/retain_grad_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/retain_grad_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/retain_grad_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f00e3a63196bddfcab48ef837355e2f937951a41
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/retain_grad_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 retain_grad(at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rms_norm.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rms_norm.h
new file mode 100644
index 0000000000000000000000000000000000000000..89f2ab87923d27a485c138f22fd14bbe1e9cb18d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rms_norm.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/rms_norm_ops.h>
+
+namespace at {
+
+
+// aten::rms_norm(Tensor input, SymInt[] normalized_shape, Tensor? weight=None, float? eps=None) -> Tensor
+inline at::Tensor rms_norm(const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight={}, ::std::optional<double> eps=::std::nullopt) {
+    return at::_ops::rms_norm::call(input, c10::fromIntArrayRefSlow(normalized_shape), weight, eps);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor rms_norm(const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight={}, ::std::optional<double> eps=::std::nullopt) {
+    return at::_ops::rms_norm::call(input, c10::fromIntArrayRefSlow(normalized_shape), weight, eps);
+  }
+}
+
+// aten::rms_norm(Tensor input, SymInt[] normalized_shape, Tensor? weight=None, float? eps=None) -> Tensor
+inline at::Tensor rms_norm_symint(const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight={}, ::std::optional<double> eps=::std::nullopt) {
+    return at::_ops::rms_norm::call(input, normalized_shape, weight, eps);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor rms_norm(const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight={}, ::std::optional<double> eps=::std::nullopt) {
+    return at::_ops::rms_norm::call(input, normalized_shape, weight, eps);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/roll_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/roll_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..dee2d129a38b72de591ea99507136c0bc9c7bbb8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/roll_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 roll(const at::Tensor & self, at::IntArrayRef shifts, at::IntArrayRef dims={});
+TORCH_API at::Tensor roll_symint(const at::Tensor & self, c10::SymIntArrayRef shifts, at::IntArrayRef dims={});
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/round_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/round_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2c5ea7bc95cf3837176bdda92f107f0065dfb3cb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/round_meta_dispatch.h
@@ -0,0 +1,30 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor round(const at::Tensor & self);
+TORCH_API at::Tensor & round_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & round_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & round_(at::Tensor & self);
+TORCH_API at::Tensor round(const at::Tensor & self, int64_t decimals);
+TORCH_API at::Tensor & round_out(at::Tensor & out, const at::Tensor & self, int64_t decimals);
+TORCH_API at::Tensor & round_outf(const at::Tensor & self, int64_t decimals, at::Tensor & out);
+TORCH_API at::Tensor & round_(at::Tensor & self, int64_t decimals);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/round_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/round_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9b660dfd67ba03a6eb21679136b580e8e8a0b8f8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/round_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/round_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_round_out : public at::meta::structured_round {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+TORCH_API at::Tensor round_sparse(const at::Tensor & self);
+TORCH_API at::Tensor & round_sparse_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & round_sparse_(at::Tensor & self);
+TORCH_API at::Tensor round_sparse_csr(const at::Tensor & self);
+TORCH_API at::Tensor & round_sparse_csr_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & round_sparse_csr_(at::Tensor & self);
+struct TORCH_API structured_round_decimals_out : public at::meta::structured_round_decimals {
+void impl(const at::Tensor & self, int64_t decimals, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/row_indices.h b/lib/python3.12/site-packages/torch/include/ATen/ops/row_indices.h
new file mode 100644
index 0000000000000000000000000000000000000000..74407abcc23b1b9c538506e335b85c9f14919cbe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/row_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/row_indices_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/row_indices_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/row_indices_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..12388950c54be0b69759f83201d098e10628f67f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/row_indices_copy.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/row_indices_copy_ops.h>
+
+namespace at {
+
+
+// aten::row_indices_copy(Tensor self) -> Tensor
+inline at::Tensor row_indices_copy(const at::Tensor & self) {
+    return at::_ops::row_indices_copy::call(self);
+}
+
+// aten::row_indices_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & row_indices_copy_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::row_indices_copy_out::call(self, out);
+}
+// aten::row_indices_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & row_indices_copy_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::row_indices_copy_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..ce98e9aa24c704e868b72ac0ca8e952c7d4e1e54
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_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 rrelu {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &, const at::Scalar &, bool, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::rrelu";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "rrelu(Tensor self, Scalar lower=0.125, Scalar upper=0.3333333333333333, bool training=False, Generator? generator=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Scalar & lower, const at::Scalar & upper, bool training, ::std::optional<at::Generator> generator);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Scalar & lower, const at::Scalar & upper, bool training, ::std::optional<at::Generator> generator);
+};
+
+struct TORCH_API rrelu_ {
+  using schema = at::Tensor & (at::Tensor &, const at::Scalar &, const at::Scalar &, bool, ::std::optional<at::Generator>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::rrelu_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "rrelu_(Tensor(a!) self, Scalar lower=0.125, Scalar upper=0.3333333333333333, bool training=False, Generator? generator=None) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Scalar & lower, const at::Scalar & upper, bool training, ::std::optional<at::Generator> generator);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Scalar & lower, const at::Scalar & upper, bool training, ::std::optional<at::Generator> generator);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rsqrt.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rsqrt.h
new file mode 100644
index 0000000000000000000000000000000000000000..ecd944274e83bab2ca5d66881534088ebb49ff45
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rsqrt.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/rsqrt_ops.h>
+
+namespace at {
+
+
+// aten::rsqrt(Tensor self) -> Tensor
+inline at::Tensor rsqrt(const at::Tensor & self) {
+    return at::_ops::rsqrt::call(self);
+}
+
+// aten::rsqrt_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & rsqrt_(at::Tensor & self) {
+    return at::_ops::rsqrt_::call(self);
+}
+
+// aten::rsqrt.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & rsqrt_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::rsqrt_out::call(self, out);
+}
+// aten::rsqrt.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & rsqrt_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::rsqrt_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_add_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_add_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3ddab01670d0799bb3b858b056f1d39cfe855759
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_add_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 scatter_add(const at::Tensor & self, at::Dimname dim, const at::Tensor & index, const at::Tensor & src);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_reduce_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_reduce_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2add648340654ab66e33c3e5775ee3e6a080966b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_reduce_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 scatter_reduce(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, c10::string_view reduce, bool include_self=true);
+TORCH_API at::Tensor & scatter_reduce_out(at::Tensor & out, const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, c10::string_view reduce, bool include_self=true);
+TORCH_API at::Tensor & scatter_reduce_outf(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, c10::string_view reduce, bool include_self, at::Tensor & out);
+TORCH_API at::Tensor & scatter_reduce_(at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, c10::string_view reduce, bool include_self=true);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_reduce_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_reduce_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..33df43154d0a25aa81e55416d2deabf4f68dffc5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_reduce_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_scatter_reduce_two : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, c10::string_view reduce, bool include_self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/select_copy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/select_copy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e30135d61eb1aa859570a8dd9a6af2d3ba0a79f2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/select_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 select_copy(const at::Tensor & self, int64_t dim, int64_t index);
+TORCH_API at::Tensor select_copy_symint(const at::Tensor & self, int64_t dim, c10::SymInt index);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/select_copy_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/select_copy_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..e7b1a4be7e2b74e07ebe570e65214e8606761533
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/select_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 select_copy_int {
+  using schema = at::Tensor (const at::Tensor &, int64_t, c10::SymInt);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::select_copy";
+  static constexpr const char* overload_name = "int";
+  static constexpr const char* schema_str = "select_copy.int(Tensor self, int dim, SymInt index) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t dim, c10::SymInt index);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, c10::SymInt index);
+};
+
+struct TORCH_API select_copy_int_out {
+  using schema = at::Tensor & (const at::Tensor &, int64_t, c10::SymInt, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::select_copy";
+  static constexpr const char* overload_name = "int_out";
+  static constexpr const char* schema_str = "select_copy.int_out(Tensor self, int dim, SymInt index, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, int64_t dim, c10::SymInt index, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, c10::SymInt index, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/selu.h b/lib/python3.12/site-packages/torch/include/ATen/ops/selu.h
new file mode 100644
index 0000000000000000000000000000000000000000..1130831c138cb6a3dc42cec44b259b4acf54f1a8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/selu.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/selu_ops.h>
+
+namespace at {
+
+
+// aten::selu(Tensor self) -> Tensor
+inline at::Tensor selu(const at::Tensor & self) {
+    return at::_ops::selu::call(self);
+}
+
+// aten::selu_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & selu_(at::Tensor & self) {
+    return at::_ops::selu_::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/selu_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/selu_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8d300bd1f6cd40adb8a2db709271b5d7d12577d0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/selu_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 selu(const at::Tensor & self);
+TORCH_API at::Tensor & selu_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/set_data.h b/lib/python3.12/site-packages/torch/include/ATen/ops/set_data.h
new file mode 100644
index 0000000000000000000000000000000000000000..cbc357b2fe05363b4ed81cb56cd382cb27a7148d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/set_data.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/set_data_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sgn_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sgn_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4cc2b59e8a40894bdda1aa9b2c94700c1eaccdd4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sgn_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/sgn_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_sgn_out : public at::meta::structured_sgn {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+TORCH_API at::Tensor NestedTensor_sgn(const at::Tensor & self);
+TORCH_API at::Tensor & NestedTensor_sgn_(at::Tensor & self);
+TORCH_API at::Tensor sgn_sparse(const at::Tensor & self);
+TORCH_API at::Tensor & sgn_sparse_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & sgn_sparse_(at::Tensor & self);
+TORCH_API at::Tensor sgn_sparse_csr(const at::Tensor & self);
+TORCH_API at::Tensor & sgn_sparse_csr_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & sgn_sparse_csr_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a502e09e46e0fc2d96df0d7b9b94b66866d92ba2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_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 sigmoid {
+  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::sigmoid";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "sigmoid(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 sigmoid_ {
+  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::sigmoid_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "sigmoid_(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 sigmoid_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::sigmoid";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "sigmoid.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/silu_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/silu_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..ff11df5103e4caa4566105c4a5eb27656427fdd7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/silu_native.h
@@ -0,0 +1,25 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/silu_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_silu_out : public at::meta::structured_silu {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+TORCH_API at::Tensor NestedTensor_silu(const at::Tensor & self);
+TORCH_API at::Tensor & NestedTensor_silu_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sin.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sin.h
new file mode 100644
index 0000000000000000000000000000000000000000..cca820a29fdd847286932af680ec45512bcc3a64
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/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/sin_ops.h>
+
+namespace at {
+
+
+// aten::sin(Tensor self) -> Tensor
+inline at::Tensor sin(const at::Tensor & self) {
+    return at::_ops::sin::call(self);
+}
+
+// aten::sin_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & sin_(at::Tensor & self) {
+    return at::_ops::sin_::call(self);
+}
+
+// aten::sin.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & sin_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::sin_out::call(self, out);
+}
+// aten::sin.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & sin_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::sin_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sinc_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sinc_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..5de51db4f4c52ee41381f16b96f8b53f9d28e1ca
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sinc_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_sinc : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slice_copy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_copy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..182ddcf27a0ed0f50bb95ce9100bc9011cc16606
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_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 slice_copy(const at::Tensor & self, int64_t dim=0, ::std::optional<int64_t> start=::std::nullopt, ::std::optional<int64_t> end=::std::nullopt, int64_t step=1);
+TORCH_API at::Tensor slice_copy_symint(const at::Tensor & self, int64_t dim=0, ::std::optional<c10::SymInt> start=::std::nullopt, ::std::optional<c10::SymInt> end=::std::nullopt, c10::SymInt step=1);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slice_scatter.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_scatter.h
new file mode 100644
index 0000000000000000000000000000000000000000..3adcc60de217e8b9d6d757fe8b0ac588fd4b9f24
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_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/slice_scatter_ops.h>
+
+namespace at {
+
+
+// aten::slice_scatter(Tensor self, Tensor src, int dim=0, SymInt? start=None, SymInt? end=None, SymInt step=1) -> Tensor
+inline at::Tensor slice_scatter(const at::Tensor & self, const at::Tensor & src, int64_t dim=0, ::std::optional<int64_t> start=::std::nullopt, ::std::optional<int64_t> end=::std::nullopt, int64_t step=1) {
+    return at::_ops::slice_scatter::call(self, src, dim, start.has_value() ? ::std::make_optional(c10::SymInt(*start)) : ::std::nullopt, end.has_value() ? ::std::make_optional(c10::SymInt(*end)) : ::std::nullopt, step);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor slice_scatter(const at::Tensor & self, const at::Tensor & src, int64_t dim=0, ::std::optional<int64_t> start=::std::nullopt, ::std::optional<int64_t> end=::std::nullopt, int64_t step=1) {
+    return at::_ops::slice_scatter::call(self, src, dim, start.has_value() ? ::std::make_optional(c10::SymInt(*start)) : ::std::nullopt, end.has_value() ? ::std::make_optional(c10::SymInt(*end)) : ::std::nullopt, step);
+  }
+}
+
+// aten::slice_scatter(Tensor self, Tensor src, int dim=0, SymInt? start=None, SymInt? end=None, SymInt step=1) -> Tensor
+inline at::Tensor slice_scatter_symint(const at::Tensor & self, const at::Tensor & src, int64_t dim=0, ::std::optional<c10::SymInt> start=::std::nullopt, ::std::optional<c10::SymInt> end=::std::nullopt, c10::SymInt step=1) {
+    return at::_ops::slice_scatter::call(self, src, dim, start, end, step);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor slice_scatter(const at::Tensor & self, const at::Tensor & src, int64_t dim=0, ::std::optional<c10::SymInt> start=::std::nullopt, ::std::optional<c10::SymInt> end=::std::nullopt, c10::SymInt step=1) {
+    return at::_ops::slice_scatter::call(self, src, dim, start, end, step);
+  }
+}
+
+// aten::slice_scatter.out(Tensor self, Tensor src, int dim=0, SymInt? start=None, SymInt? end=None, SymInt step=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & slice_scatter_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & src, int64_t dim=0, ::std::optional<int64_t> start=::std::nullopt, ::std::optional<int64_t> end=::std::nullopt, int64_t step=1) {
+    return at::_ops::slice_scatter_out::call(self, src, dim, start.has_value() ? ::std::make_optional(c10::SymInt(*start)) : ::std::nullopt, end.has_value() ? ::std::make_optional(c10::SymInt(*end)) : ::std::nullopt, step, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & slice_scatter_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & src, int64_t dim=0, ::std::optional<int64_t> start=::std::nullopt, ::std::optional<int64_t> end=::std::nullopt, int64_t step=1) {
+    return at::_ops::slice_scatter_out::call(self, src, dim, start.has_value() ? ::std::make_optional(c10::SymInt(*start)) : ::std::nullopt, end.has_value() ? ::std::make_optional(c10::SymInt(*end)) : ::std::nullopt, step, out);
+  }
+}
+
+// aten::slice_scatter.out(Tensor self, Tensor src, int dim=0, SymInt? start=None, SymInt? end=None, SymInt step=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & slice_scatter_outf(const at::Tensor & self, const at::Tensor & src, int64_t dim, ::std::optional<int64_t> start, ::std::optional<int64_t> end, int64_t step, at::Tensor & out) {
+    return at::_ops::slice_scatter_out::call(self, src, dim, start.has_value() ? ::std::make_optional(c10::SymInt(*start)) : ::std::nullopt, end.has_value() ? ::std::make_optional(c10::SymInt(*end)) : ::std::nullopt, step, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & slice_scatter_outf(const at::Tensor & self, const at::Tensor & src, int64_t dim, ::std::optional<int64_t> start, ::std::optional<int64_t> end, int64_t step, at::Tensor & out) {
+    return at::_ops::slice_scatter_out::call(self, src, dim, start.has_value() ? ::std::make_optional(c10::SymInt(*start)) : ::std::nullopt, end.has_value() ? ::std::make_optional(c10::SymInt(*end)) : ::std::nullopt, step, out);
+  }
+}
+
+// aten::slice_scatter.out(Tensor self, Tensor src, int dim=0, SymInt? start=None, SymInt? end=None, SymInt step=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & slice_scatter_symint_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & src, int64_t dim=0, ::std::optional<c10::SymInt> start=::std::nullopt, ::std::optional<c10::SymInt> end=::std::nullopt, c10::SymInt step=1) {
+    return at::_ops::slice_scatter_out::call(self, src, dim, start, end, step, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & slice_scatter_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & src, int64_t dim=0, ::std::optional<c10::SymInt> start=::std::nullopt, ::std::optional<c10::SymInt> end=::std::nullopt, c10::SymInt step=1) {
+    return at::_ops::slice_scatter_out::call(self, src, dim, start, end, step, out);
+  }
+}
+
+// aten::slice_scatter.out(Tensor self, Tensor src, int dim=0, SymInt? start=None, SymInt? end=None, SymInt step=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & slice_scatter_symint_outf(const at::Tensor & self, const at::Tensor & src, int64_t dim, ::std::optional<c10::SymInt> start, ::std::optional<c10::SymInt> end, c10::SymInt step, at::Tensor & out) {
+    return at::_ops::slice_scatter_out::call(self, src, dim, start, end, step, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & slice_scatter_outf(const at::Tensor & self, const at::Tensor & src, int64_t dim, ::std::optional<c10::SymInt> start, ::std::optional<c10::SymInt> end, c10::SymInt step, at::Tensor & out) {
+    return at::_ops::slice_scatter_out::call(self, src, dim, start, end, step, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv3d_forward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv3d_forward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..51a6e04e8de70701100d2f47cf33926d5e8e85cc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv3d_forward_cpu_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor slow_conv3d_forward(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding);
+TORCH_API at::Tensor slow_conv3d_forward_symint(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & slow_conv3d_forward_out(at::Tensor & output, const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding);
+TORCH_API at::Tensor & slow_conv3d_forward_outf(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::Tensor & output);
+TORCH_API at::Tensor & slow_conv3d_forward_symint_out(at::Tensor & output, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & slow_conv3d_forward_symint_outf(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, at::Tensor & output);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/smooth_l1_loss_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/smooth_l1_loss_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4dbcf467b269036c6b2ed122506a6e27f6d4f3e5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/smooth_l1_loss_backward_cuda_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor & smooth_l1_loss_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction, double beta);
+TORCH_API at::Tensor & smooth_l1_loss_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction, double beta, at::Tensor & grad_input);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/softplus_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/softplus_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..235553d184921f0fc2b6a761e6c803b0194c2c7d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/softplus_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 softplus(const at::Tensor & self, const at::Scalar & beta=1, const at::Scalar & threshold=20);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sort_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sort_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6ac7fdb526fffeafaf69c2fb19c48b422bafac4c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sort_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> sort(const at::Tensor & self, ::std::optional<bool> stable, int64_t dim=-1, bool descending=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> sort_out(at::Tensor & values, at::Tensor & indices, const at::Tensor & self, ::std::optional<bool> stable, int64_t dim=-1, bool descending=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> sort_outf(const at::Tensor & self, ::std::optional<bool> stable, int64_t dim, bool descending, at::Tensor & values, at::Tensor & indices);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_compressed_tensor_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_compressed_tensor_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9cf682b9f5cb5c16e43c850ec093badd5ae2a09b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_compressed_tensor_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_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(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 native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csr_tensor_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csr_tensor_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f3972d9fae14032193f67bbc6f3ed8491d13d309
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_csr_tensor_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor sparse_csr_tensor(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, at::IntArrayRef size, at::TensorOptions options);
+TORCH_API at::Tensor sparse_csr_tensor(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor sparse_csr_tensor(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, at::TensorOptions options);
+TORCH_API at::Tensor sparse_csr_tensor(const at::Tensor & crow_indices, const at::Tensor & col_indices, const at::Tensor & values, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_dim_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_dim_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..0be0f7aaf9b3b22a8376e792a6a62679676c6b51
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_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 sparse_dim {
+  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::sparse_dim";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "sparse_dim(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/sparse_resize_and_clear_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_resize_and_clear_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6b6ea186c66e42435ffef926eef280ebf72b279e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_resize_and_clear_meta_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API const at::Tensor & sparse_resize_and_clear_(const at::Tensor & self, at::IntArrayRef size, int64_t sparse_dim, int64_t dense_dim);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_t.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_t.h
new file mode 100644
index 0000000000000000000000000000000000000000..15cd0d8325e512a7787881687889e017e6aa1e2a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_t.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_chebyshev_polynomial_t_ops.h>
+
+namespace at {
+
+
+// aten::special_chebyshev_polynomial_t(Tensor x, Tensor n) -> Tensor
+inline at::Tensor special_chebyshev_polynomial_t(const at::Tensor & x, const at::Tensor & n) {
+    return at::_ops::special_chebyshev_polynomial_t::call(x, n);
+}
+
+// aten::special_chebyshev_polynomial_t.x_scalar(Scalar x, Tensor n) -> Tensor
+inline at::Tensor special_chebyshev_polynomial_t(const at::Scalar & x, const at::Tensor & n) {
+    return at::_ops::special_chebyshev_polynomial_t_x_scalar::call(x, n);
+}
+
+// aten::special_chebyshev_polynomial_t.n_scalar(Tensor x, Scalar n) -> Tensor
+inline at::Tensor special_chebyshev_polynomial_t(const at::Tensor & x, const at::Scalar & n) {
+    return at::_ops::special_chebyshev_polynomial_t_n_scalar::call(x, n);
+}
+
+// aten::special_chebyshev_polynomial_t.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_chebyshev_polynomial_t_out(at::Tensor & out, const at::Tensor & x, const at::Tensor & n) {
+    return at::_ops::special_chebyshev_polynomial_t_out::call(x, n, out);
+}
+// aten::special_chebyshev_polynomial_t.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_chebyshev_polynomial_t_outf(const at::Tensor & x, const at::Tensor & n, at::Tensor & out) {
+    return at::_ops::special_chebyshev_polynomial_t_out::call(x, n, out);
+}
+
+// aten::special_chebyshev_polynomial_t.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_chebyshev_polynomial_t_out(at::Tensor & out, const at::Scalar & x, const at::Tensor & n) {
+    return at::_ops::special_chebyshev_polynomial_t_x_scalar_out::call(x, n, out);
+}
+// aten::special_chebyshev_polynomial_t.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_chebyshev_polynomial_t_outf(const at::Scalar & x, const at::Tensor & n, at::Tensor & out) {
+    return at::_ops::special_chebyshev_polynomial_t_x_scalar_out::call(x, n, out);
+}
+
+// aten::special_chebyshev_polynomial_t.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_chebyshev_polynomial_t_out(at::Tensor & out, const at::Tensor & x, const at::Scalar & n) {
+    return at::_ops::special_chebyshev_polynomial_t_n_scalar_out::call(x, n, out);
+}
+// aten::special_chebyshev_polynomial_t.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_chebyshev_polynomial_t_outf(const at::Tensor & x, const at::Scalar & n, at::Tensor & out) {
+    return at::_ops::special_chebyshev_polynomial_t_n_scalar_out::call(x, n, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_erf_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_erf_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..fe6f6e83c78f707add83299c3b24b7a172eb93ab
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_erf_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_erf {
+  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_erf";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_erf(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_erf_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_erf";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_erf.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_erfc_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_erfc_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..03d6c64703d968c74d67cbbdbf3f9d935c450286
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_erfc_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_erfc(const at::Tensor & self);
+TORCH_API at::Tensor & special_erfc_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_he_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_hermite_polynomial_he_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6dac4c151575a2b71a112b665fac6196d6a60236
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_hermite_polynomial_he_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_hermite_polynomial_he(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_i1_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i1_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..88a69f8343e3c91d8d26bc3f28cfeda84c9ca983
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i1_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_i1(const at::Tensor & self);
+TORCH_API at::Tensor & special_i1_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & special_i1_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_laguerre_polynomial_l_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_laguerre_polynomial_l_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2493a1c1ba3b9743552f059e43855e6d56a10fbd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_laguerre_polynomial_l_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 special_laguerre_polynomial_l(const at::Scalar & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_laguerre_polynomial_l_out(at::Tensor & out, const at::Scalar & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_laguerre_polynomial_l_outf(const at::Scalar & x, const at::Tensor & n, at::Tensor & out);
+TORCH_API at::Tensor special_laguerre_polynomial_l(const at::Tensor & x, const at::Scalar & n);
+TORCH_API at::Tensor & special_laguerre_polynomial_l_out(at::Tensor & out, const at::Tensor & x, const at::Scalar & n);
+TORCH_API at::Tensor & special_laguerre_polynomial_l_outf(const at::Tensor & x, const at::Scalar & n, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_laguerre_polynomial_l_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_laguerre_polynomial_l_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..6a274db6916adc22092f4aa037130278cba9779e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_laguerre_polynomial_l_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_laguerre_polynomial_l : 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_legendre_polynomial_p_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_legendre_polynomial_p_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0dbd2b384fcf5cac1193e2e87e1d7df0944eed44
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_legendre_polynomial_p_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 special_legendre_polynomial_p(const at::Scalar & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_legendre_polynomial_p_out(at::Tensor & out, const at::Scalar & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_legendre_polynomial_p_outf(const at::Scalar & x, const at::Tensor & n, at::Tensor & out);
+TORCH_API at::Tensor special_legendre_polynomial_p(const at::Tensor & x, const at::Scalar & n);
+TORCH_API at::Tensor & special_legendre_polynomial_p_out(at::Tensor & out, const at::Tensor & x, const at::Scalar & n);
+TORCH_API at::Tensor & special_legendre_polynomial_p_outf(const at::Tensor & x, const at::Scalar & n, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_k0_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_k0_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0207b570537da2f4761aa29146312be579cd3cc6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_k0_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_modified_bessel_k0(const at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_k0_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_k0_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9d2af42765c6e392d21674be88394e04fcf24503
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_k0_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/special_modified_bessel_k0_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_modified_bessel_k0_out : public at::meta::structured_special_modified_bessel_k0 {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_k0_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_k0_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..5b9724cc0b1967e563f7e3057f723ed6ca6dafc3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_modified_bessel_k0_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_modified_bessel_k0 {
+  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_modified_bessel_k0";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_modified_bessel_k0(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_modified_bessel_k0_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_modified_bessel_k0";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_modified_bessel_k0.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_multigammaln_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_multigammaln_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..38bb71b842e207c5c2dc1da0495641afc5b38e57
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_multigammaln_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_multigammaln(const at::Tensor & self, int64_t p);
+TORCH_API at::Tensor & special_multigammaln_out(const at::Tensor & self, int64_t p, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtri_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtri_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f11d4555575af466f07130925b361ea250529f23
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtri_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_ndtri {
+  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_ndtri";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_ndtri(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_ndtri_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_ndtri";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_ndtri.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_shifted_chebyshev_polynomial_u_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_u_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..3dea27053111069ec6eb363c433f03367462ad4a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_u_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_u_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_shifted_chebyshev_polynomial_u_out : public at::meta::structured_special_shifted_chebyshev_polynomial_u {
+void impl(const at::Tensor & x, const at::Tensor & n, const at::Tensor & out);
+};
+TORCH_API at::Tensor special_shifted_chebyshev_polynomial_u(const at::Scalar & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_shifted_chebyshev_polynomial_u_out(const at::Scalar & x, const at::Tensor & n, at::Tensor & out);
+TORCH_API at::Tensor special_shifted_chebyshev_polynomial_u(const at::Tensor & x, const at::Scalar & n);
+TORCH_API at::Tensor & special_shifted_chebyshev_polynomial_u_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_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_v_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..b1f6bf5d44b5e411869fbfb311e33807008f78e9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_v_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_special_shifted_chebyshev_polynomial_v : 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_xlog1py_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlog1py_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..fa7ec4cccdb037ded79d45a4d0b77e08f9fa9926
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlog1py_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_xlog1py_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_xlog1py_out : public at::meta::structured_special_xlog1py {
+void impl(const at::Tensor & self, const at::Tensor & other, const at::Tensor & out);
+};
+TORCH_API at::Tensor special_xlog1py(const at::Scalar & self, const at::Tensor & other);
+TORCH_API at::Tensor & special_xlog1py_out(const at::Scalar & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor special_xlog1py(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & special_xlog1py_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_with_sizes_copy_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/split_with_sizes_copy_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..fe19c6124ccc6edef073aa8cc29875a50082103d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/split_with_sizes_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 split_with_sizes_copy {
+  using schema = ::std::vector<at::Tensor> (const at::Tensor &, c10::SymIntArrayRef, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::split_with_sizes_copy";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "split_with_sizes_copy(Tensor self, SymInt[] split_sizes, int dim=0) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim);
+};
+
+struct TORCH_API split_with_sizes_copy_out {
+  using schema = void (const at::Tensor &, c10::SymIntArrayRef, int64_t, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::split_with_sizes_copy";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "split_with_sizes_copy.out(Tensor self, SymInt[] split_sizes, int dim=0, *, Tensor(a!)[] out) -> ()";
+  static void call(const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim, at::TensorList out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/square.h b/lib/python3.12/site-packages/torch/include/ATen/ops/square.h
new file mode 100644
index 0000000000000000000000000000000000000000..851c233ae88aa74495f0586c6f361b91b4387790
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/square.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/square_ops.h>
+
+namespace at {
+
+
+// aten::square(Tensor self) -> Tensor
+inline at::Tensor square(const at::Tensor & self) {
+    return at::_ops::square::call(self);
+}
+
+// aten::square_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & square_(at::Tensor & self) {
+    return at::_ops::square_::call(self);
+}
+
+// aten::square.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & square_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::square_out::call(self, out);
+}
+// aten::square.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & square_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::square_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze.h b/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze.h
new file mode 100644
index 0000000000000000000000000000000000000000..88c55f7b69f7bc9e7963a33edecf0eb9e8dddc41
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze.h
@@ -0,0 +1,45 @@
+#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/squeeze_ops.h>
+
+namespace at {
+
+
+// aten::squeeze(Tensor(a) self) -> Tensor(a)
+inline at::Tensor squeeze(const at::Tensor & self) {
+    return at::_ops::squeeze::call(self);
+}
+
+// aten::squeeze.dim(Tensor(a) self, int dim) -> Tensor(a)
+inline at::Tensor squeeze(const at::Tensor & self, int64_t dim) {
+    return at::_ops::squeeze_dim::call(self, dim);
+}
+
+// aten::squeeze.dimname(Tensor(a) self, Dimname dim) -> Tensor(a)
+inline at::Tensor squeeze(const at::Tensor & self, at::Dimname dim) {
+    return at::_ops::squeeze_dimname::call(self, dim);
+}
+
+// aten::squeeze.dims(Tensor(a) self, int[] dim) -> Tensor(a)
+inline at::Tensor squeeze(const at::Tensor & self, at::IntArrayRef dim) {
+    return at::_ops::squeeze_dims::call(self, dim);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..00ca6dfc42be8d1c47723a96f188a37f2a0d3d41
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze_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 squeeze(const at::Tensor & self, at::Dimname dim);
+TORCH_API at::Tensor & squeeze_(at::Tensor & self, at::Dimname dim);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze_copy_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze_copy_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b1f1cc35e69e9ce77635c63edb91700aea56f270
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/squeeze_copy_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 & squeeze_copy_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & squeeze_copy_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & squeeze_copy_out(at::Tensor & out, const at::Tensor & self, int64_t dim);
+TORCH_API at::Tensor & squeeze_copy_outf(const at::Tensor & self, int64_t dim, at::Tensor & out);
+TORCH_API at::Tensor & squeeze_copy_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim);
+TORCH_API at::Tensor & squeeze_copy_outf(const at::Tensor & self, at::IntArrayRef dim, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/stack.h b/lib/python3.12/site-packages/torch/include/ATen/ops/stack.h
new file mode 100644
index 0000000000000000000000000000000000000000..575b938c54e78678d90ef7e905a11a0a549f8c74
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/stack.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/stack_ops.h>
+
+namespace at {
+
+
+// aten::stack(Tensor[] tensors, int dim=0) -> Tensor
+inline at::Tensor stack(at::TensorList tensors, int64_t dim=0) {
+    return at::_ops::stack::call(tensors, dim);
+}
+
+// aten::stack.out(Tensor[] tensors, int dim=0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & stack_out(at::Tensor & out, at::TensorList tensors, int64_t dim=0) {
+    return at::_ops::stack_out::call(tensors, dim, out);
+}
+// aten::stack.out(Tensor[] tensors, int dim=0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & stack_outf(at::TensorList tensors, int64_t dim, at::Tensor & out) {
+    return at::_ops::stack_out::call(tensors, dim, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/std_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/std_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ead92c4e782fa43da8357e777b7fea6dda547a24
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/std_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 std(const at::Tensor & self, at::OptionalIntArrayRef dim=::std::nullopt, const ::std::optional<at::Scalar> & correction=::std::nullopt, bool keepdim=false);
+TORCH_API at::Tensor & std_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef dim=::std::nullopt, const ::std::optional<at::Scalar> & correction=::std::nullopt, bool keepdim=false);
+TORCH_API at::Tensor & std_outf(const at::Tensor & self, at::OptionalIntArrayRef dim, const ::std::optional<at::Scalar> & correction, bool keepdim, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sum_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sum_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..dfbac17bd7943b1023ffc400c3c11a29fcd59b02
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sum_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 sum(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & sum_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & sum_outf(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/swapaxes.h b/lib/python3.12/site-packages/torch/include/ATen/ops/swapaxes.h
new file mode 100644
index 0000000000000000000000000000000000000000..7927ef54846bd06550082f1e557550bc41b37148
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/swapaxes.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/swapaxes_ops.h>
+
+namespace at {
+
+
+// aten::swapaxes(Tensor(a) self, int axis0, int axis1) -> Tensor(a)
+inline at::Tensor swapaxes(const at::Tensor & self, int64_t axis0, int64_t axis1) {
+    return at::_ops::swapaxes::call(self, axis0, axis1);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/t_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/t_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..94c9acbe526ef0c48e83f69bd62a44600d9c8f0b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/t_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 t(const at::Tensor & self);
+TORCH_API at::Tensor & t_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/take_along_dim_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/take_along_dim_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b7b4dcbd18c74ee71abc570b39c68411e29d7202
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/take_along_dim_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor take_along_dim(const at::Tensor & self, const at::Tensor & indices, ::std::optional<int64_t> dim=::std::nullopt);
+TORCH_API at::Tensor & take_along_dim_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & indices, ::std::optional<int64_t> dim=::std::nullopt);
+TORCH_API at::Tensor & take_along_dim_outf(const at::Tensor & self, const at::Tensor & indices, ::std::optional<int64_t> dim, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/take_along_dim_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/take_along_dim_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..622a9474fa4da394944a61600bcb0b17a9f67350
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/take_along_dim_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 take_along_dim_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, ::std::optional<int64_t>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::take_along_dim";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "take_along_dim.out(Tensor self, Tensor indices, int? dim=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & indices, ::std::optional<int64_t> dim, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & indices, ::std::optional<int64_t> dim, at::Tensor & out);
+};
+
+struct TORCH_API take_along_dim {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, ::std::optional<int64_t>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::take_along_dim";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "take_along_dim(Tensor self, Tensor indices, int? dim=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & indices, ::std::optional<int64_t> dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & indices, ::std::optional<int64_t> dim);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/tensordot_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/tensordot_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d62319161eaf12e4f7958fd3ef0f220c5ce13d8a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/tensordot_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor tensordot(const at::Tensor & self, const at::Tensor & other, at::IntArrayRef dims_self, at::IntArrayRef dims_other);
+TORCH_API at::Tensor & tensordot_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other, at::IntArrayRef dims_self, at::IntArrayRef dims_other);
+TORCH_API at::Tensor & tensordot_outf(const at::Tensor & self, const at::Tensor & other, at::IntArrayRef dims_self, at::IntArrayRef dims_other, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/thnn_conv2d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/thnn_conv2d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..b769b6843e5217d5375c988113d2b1fb3fb90b1b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/thnn_conv2d_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 thnn_conv2d_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, c10::SymIntArrayRef, const ::std::optional<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::thnn_conv2d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "thnn_conv2d.out(Tensor self, Tensor weight, SymInt[2] kernel_size, Tensor? bias=None, SymInt[2] stride=1, SymInt[2] padding=0, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, at::Tensor & out);
+};
+
+struct TORCH_API thnn_conv2d {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, c10::SymIntArrayRef, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::thnn_conv2d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "thnn_conv2d(Tensor self, Tensor weight, SymInt[2] kernel_size, Tensor? bias=None, SymInt[2] stride=1, SymInt[2] padding=0) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..72e37ccd2b5ca05e96d857585f13b3789e4be55f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_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 threshold(const at::Tensor & self, const at::Scalar & threshold, const at::Scalar & value);
+TORCH_API at::Tensor & threshold_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & threshold, const at::Scalar & value);
+TORCH_API at::Tensor & threshold_outf(const at::Tensor & self, const at::Scalar & threshold, const at::Scalar & value, at::Tensor & out);
+TORCH_API at::Tensor & threshold_(at::Tensor & self, const at::Scalar & threshold, const at::Scalar & value);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/to_mkldnn_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/to_mkldnn_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0bc1ed76067d98651cac409fad1575824cfd8e7a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/to_mkldnn_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_mkldnn_out(at::Tensor & out, const at::Tensor & self, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & to_mkldnn_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/to_sparse_csc_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_csc_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fb276b6c73c907fcb1cacddeddb9a8f712eab17e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/to_sparse_csc_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_csc(const at::Tensor & self, ::std::optional<int64_t> dense_dim=::std::nullopt);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/trace_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/trace_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..108adcdabf633c78fa8191faadf2edea6a4fd1b2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/trace_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 & trace_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & trace_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/triangular_solve_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/triangular_solve_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6e2669da969287c79c9ecca69399b692a3d2ca1b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/triangular_solve_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> triangular_solve(const at::Tensor & self, const at::Tensor & A, bool upper=true, bool transpose=false, bool unitriangular=false);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unbind_copy_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unbind_copy_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..66b4d71aad935576f9965abd138a11e27de07f29
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unbind_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 unbind_copy_int {
+  using schema = ::std::vector<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::unbind_copy";
+  static constexpr const char* overload_name = "int";
+  static constexpr const char* schema_str = "unbind_copy.int(Tensor self, int dim=0) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(const at::Tensor & self, int64_t dim);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim);
+};
+
+struct TORCH_API unbind_copy_int_out {
+  using schema = void (const at::Tensor &, int64_t, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::unbind_copy";
+  static constexpr const char* overload_name = "int_out";
+  static constexpr const char* schema_str = "unbind_copy.int_out(Tensor self, int dim=0, *, Tensor(a!)[] out) -> ()";
+  static void call(const at::Tensor & self, int64_t dim, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, at::TensorList out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unfold.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unfold.h
new file mode 100644
index 0000000000000000000000000000000000000000..f139e8d0ef08640c1ab2e3c9a475b7c3c2731eec
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unfold.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/unfold_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/uniform_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/uniform_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7b003b202de45f0a2e30782cd1394df41b18bfb2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/uniform_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 & uniform_(at::Tensor & self, double from=0, double to=1, ::std::optional<at::Generator> generator=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unique_dim_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unique_dim_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e737debc10d916e66acc6a9e83aeb467bba118b2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unique_dim_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,at::Tensor> unique_dim(const at::Tensor & self, int64_t dim, bool sorted=true, bool return_inverse=false, bool return_counts=false);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_chunk_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_chunk_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..788911841b79beeb919586941d98ce0260789494
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_chunk_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::vector<at::Tensor> unsafe_chunk(const at::Tensor & self, int64_t chunks, int64_t dim=0);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_split_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_split_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..2aeae6dca18266a9cdcb8b50d6bb748ca3a1fb35
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_split_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::vector<at::Tensor> unsafe_split(const at::Tensor & self, int64_t split_size, int64_t dim=0);
+TORCH_API void unsafe_split_Tensor_out_symint(const at::Tensor & self, c10::SymInt split_size, int64_t dim, at::TensorList out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bicubic2d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bicubic2d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..bb4659ffaacf394f1542997a5f60d4b7ac87470c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bicubic2d_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_bicubic2d_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_bicubic2d";
+  static constexpr const char* overload_name = "vec";
+  static constexpr const char* schema_str = "upsample_bicubic2d.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_bicubic2d_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, bool, ::std::optional<double>, ::std::optional<double>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::upsample_bicubic2d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "upsample_bicubic2d.out(Tensor self, SymInt[2] output_size, bool align_corners, float? scales_h=None, float? scales_w=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+};
+
+struct TORCH_API upsample_bicubic2d {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, bool, ::std::optional<double>, ::std::optional<double>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::upsample_bicubic2d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "upsample_bicubic2d(Tensor self, SymInt[2] output_size, bool align_corners, float? scales_h=None, float? scales_w=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bilinear2d_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bilinear2d_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..8a98e713625b350f4e40e453f24795bd052eaed1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bilinear2d_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_bilinear2d : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, at::ArrayRef<int64_t> output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_backward_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_backward_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6e3ead5c0a8396806c54f852d8b6659f7307a129
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_backward_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_linear1d_backward(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor upsample_linear1d_backward_symint(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor & upsample_linear1d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor & upsample_linear1d_backward_outf(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales, at::Tensor & grad_input);
+TORCH_API at::Tensor & upsample_linear1d_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor & upsample_linear1d_backward_symint_outf(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales, at::Tensor & grad_input);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest1d_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest1d_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..3f9e935f7067ae16062a452a09b79fd90483ba99
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest1d_backward_native.h
@@ -0,0 +1,26 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/upsample_nearest1d_backward_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_upsample_nearest1d_backward_out_cpu : public at::meta::structured_upsample_nearest1d_backward {
+void impl(const at::Tensor & grad_output, at::ArrayRef<int64_t> output_size, at::ArrayRef<int64_t> input_size, ::std::optional<double> scales, const at::Tensor & grad_input);
+};
+struct TORCH_API structured_upsample_nearest1d_backward_out_cuda : public at::meta::structured_upsample_nearest1d_backward {
+void impl(const at::Tensor & grad_output, at::ArrayRef<int64_t> output_size, at::ArrayRef<int64_t> input_size, ::std::optional<double> scales, const at::Tensor & grad_input);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest2d_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest2d_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f0a9bae67d83e980a965f6bc38ff770017c35020
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest2d_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_nearest2d(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_nearest2d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, ::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_nearest3d_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..abd3727da5ca76405e00bc847c79efd6ecae3a11
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_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 upsample_nearest3d_backward_grad_input {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, c10::SymIntArrayRef, ::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_nearest3d_backward";
+  static constexpr const char* overload_name = "grad_input";
+  static constexpr const char* schema_str = "upsample_nearest3d_backward.grad_input(Tensor grad_output, SymInt[3] output_size, SymInt[5] input_size, float? scales_d=None, float? scales_h=None, float? scales_w=None, *, Tensor(a!) grad_input) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input);
+};
+
+struct TORCH_API upsample_nearest3d_backward {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, c10::SymIntArrayRef, ::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_nearest3d_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "upsample_nearest3d_backward(Tensor grad_output, SymInt[3] output_size, SymInt[5] input_size, float? scales_d=None, float? scales_h=None, float? scales_w=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::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 & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, ::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/upsample_nearest3d_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..99c9518bef1e414edfa0f417dfd711b81e480a49
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_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_nearest3d(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor upsample_nearest3d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_nearest3d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_nearest3d_outf(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+TORCH_API at::Tensor & upsample_nearest3d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d=::std::nullopt, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_nearest3d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d0686854fa67bc7cc96570e982e4f909e646099d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest3d_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/upsample_nearest3d_meta.h>
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor upsample_nearest3d(const at::Tensor & input, at::OptionalIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors);
+struct TORCH_API structured_upsample_nearest3d_out_cpu : public at::meta::structured_upsample_nearest3d {
+void impl(const at::Tensor & self, at::ArrayRef<int64_t> output_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, const at::Tensor & out);
+};
+struct TORCH_API structured_upsample_nearest3d_out_cuda : public at::meta::structured_upsample_nearest3d {
+void impl(const at::Tensor & self, at::ArrayRef<int64_t> output_size, ::std::optional<double> scales_d, ::std::optional<double> scales_h, ::std::optional<double> scales_w, const at::Tensor & out);
+};
+TORCH_API at::Tensor upsample_nearest3d_quantized_cpu(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);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/vander_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/vander_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..96255830849d21a5c14b1651cdf37d4ff44a52ca
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/vander_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 vander(const at::Tensor & x, ::std::optional<int64_t> N=::std::nullopt, bool increasing=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/vdot_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/vdot_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1049b7b4874e7f50311bef4b819d0f79a3bf15fe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/vdot_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 & vdot_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & vdot_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/vstack_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/vstack_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..59a1f20de5857ecfdc0340da07d712620aaa8556
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/vstack_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 vstack(at::TensorList tensors);
+TORCH_API at::Tensor & vstack_out(at::TensorList tensors, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..6fae59421c1f3596efe6e4d4cecb2a0c8226565f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/xlogy_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/xlogy_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_xlogy_out : public at::meta::structured_xlogy_Tensor {
+void impl(const at::Tensor & self, const at::Tensor & other, const at::Tensor & out);
+};
+TORCH_API at::Tensor xlogy(const at::Scalar & self, const at::Tensor & other);
+TORCH_API at::Tensor & xlogy_out(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(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & xlogy_(at::Tensor & self, const at::Scalar & other);
+} // namespace native
+} // namespace at