diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1164daf58526c797f3b16f9646130512c1107659
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool2d_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & _adaptive_avg_pool2d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API at::Tensor & _adaptive_avg_pool2d_outf(const at::Tensor & self, at::IntArrayRef output_size, at::Tensor & out);
+TORCH_API at::Tensor & _adaptive_avg_pool2d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size);
+TORCH_API at::Tensor & _adaptive_avg_pool2d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool3d_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_avg_pool3d_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..38032ee70b4880a393283d1c1d6c6327b1a78c71
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_adaptive_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/_adaptive_avg_pool3d_backward_ops.h>
+
+namespace at {
+
+
+// aten::_adaptive_avg_pool3d_backward(Tensor grad_output, Tensor self) -> Tensor
+inline at::Tensor _adaptive_avg_pool3d_backward(const at::Tensor & grad_output, const at::Tensor & self) {
+    return at::_ops::_adaptive_avg_pool3d_backward::call(grad_output, self);
+}
+
+// aten::_adaptive_avg_pool3d_backward.out(Tensor grad_output, Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _adaptive_avg_pool3d_backward_out(at::Tensor & out, const at::Tensor & grad_output, const at::Tensor & self) {
+    return at::_ops::_adaptive_avg_pool3d_backward_out::call(grad_output, self, out);
+}
+// aten::_adaptive_avg_pool3d_backward.out(Tensor grad_output, Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _adaptive_avg_pool3d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::_adaptive_avg_pool3d_backward_out::call(grad_output, self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_add_relu_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_add_relu_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..589725f0c02111130b5ebd55d7827cbe06a84df8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_add_relu_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 _add_relu(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & _add_relu_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & _add_relu_outf(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha, at::Tensor & out);
+TORCH_API at::Tensor & _add_relu_(at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor _add_relu(const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & _add_relu_(at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha=1);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_amp_update_scale_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_amp_update_scale_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1b740f5d433d49e009215021779cfc8a15f8a1a3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_amp_update_scale_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<at::Tensor,at::Tensor> _amp_update_scale(const at::Tensor & self, const at::Tensor & growth_tracker, const at::Tensor & found_inf, double scale_growth_factor, double scale_backoff_factor, int64_t growth_interval);
+TORCH_API at::Tensor & _amp_update_scale_out(at::Tensor & out, const at::Tensor & self, at::Tensor & growth_tracker, const at::Tensor & found_inf, double scale_growth_factor, double scale_backoff_factor, int64_t growth_interval);
+TORCH_API at::Tensor & _amp_update_scale_outf(const at::Tensor & self, at::Tensor & growth_tracker, const at::Tensor & found_inf, double scale_growth_factor, double scale_backoff_factor, int64_t growth_interval, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_async_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_async_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..18426dd1ad79e3d8725c9b103dc19591a84f254f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_async_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 void _assert_async_cpu(const at::Tensor & self);
+TORCH_API void _assert_async_cuda(const at::Tensor & self);
+TORCH_API void _assert_async_msg_cpu(const at::Tensor & self, c10::string_view assert_msg);
+TORCH_API void _assert_async_msg_cuda(const at::Tensor & self, c10::string_view assert_msg);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_tensor_metadata_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_tensor_metadata_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9400149222882026f69f3d224c15fed0487815fb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_assert_tensor_metadata_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 _assert_tensor_metadata {
+  using schema = void (const at::Tensor &, at::OptionalSymIntArrayRef, at::OptionalSymIntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Device>, ::std::optional<at::Layout>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_assert_tensor_metadata";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_assert_tensor_metadata(Tensor a, SymInt[]? size=None, SymInt[]? stride=None, ScalarType? dtype=None, *, Device? device=None, Layout? layout=None) -> ()";
+  static void call(const at::Tensor & a, at::OptionalSymIntArrayRef size, at::OptionalSymIntArrayRef stride, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Device> device, ::std::optional<at::Layout> layout);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & a, at::OptionalSymIntArrayRef size, at::OptionalSymIntArrayRef stride, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Device> device, ::std::optional<at::Layout> layout);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_with_update_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_with_update_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..613fed3c16e921d07c08c974b21bc14fa6e1a668
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_with_update_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 ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _batch_norm_with_update_functional(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _batch_norm_with_update_cpu_out(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, double momentum, double eps, at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd, at::Tensor & reserve);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> _batch_norm_with_update_cpu(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _batch_norm_with_update_cuda_out(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, double momentum, double eps, at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd, at::Tensor & reserve);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> _batch_norm_with_update_cuda(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> _batch_norm_with_update_mkldnn(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, double momentum, double eps);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Short_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Short_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1f65e26e1f115b2a89c28baee37bee3bd859fe49
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Short_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 _cast_Short {
+  using schema = 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::_cast_Short";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_cast_Short(Tensor self, bool non_blocking=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, bool non_blocking);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, bool non_blocking);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cdist_forward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cdist_forward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..387a8e8cb37fb586967283c5c25abb1cce196369
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cdist_forward_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 _cdist_forward(const at::Tensor & x1, const at::Tensor & x2, double p, ::std::optional<int64_t> compute_mode);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_compute_linear_combination_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_compute_linear_combination_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2fa31ea5ac8f96aab8b02629c6dbf399e5f1fc94
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_compute_linear_combination_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 _compute_linear_combination(const at::Tensor & input, const at::Tensor & coefficients);
+TORCH_API at::Tensor & _compute_linear_combination_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & coefficients);
+TORCH_API at::Tensor & _compute_linear_combination_outf(const at::Tensor & input, const at::Tensor & coefficients, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_conj_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_conj_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..9a231ccbffa2ac33e6ac87d5b880869b99f7a40f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_conj_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/_conj_copy_ops.h>
+
+namespace at {
+
+
+// aten::_conj_copy(Tensor self) -> Tensor
+inline at::Tensor _conj_copy(const at::Tensor & self) {
+    return at::_ops::_conj_copy::call(self);
+}
+
+// aten::_conj_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _conj_copy_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::_conj_copy_out::call(self, out);
+}
+// aten::_conj_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _conj_copy_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::_conj_copy_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_weight_to_int4pack_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_weight_to_int4pack_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..14df73b867581d86492a36a9154747711d7d952f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_convert_weight_to_int4pack_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 _convert_weight_to_int4pack(const at::Tensor & self, int64_t innerKTiles);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_clear_plan_cache_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_clear_plan_cache_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..57b88023de0c6738a82b852dc52961b575fd755b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_clear_plan_cache_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_clear_plan_cache {
+  using schema = void (at::DeviceIndex);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_cufft_clear_plan_cache";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_cufft_clear_plan_cache(DeviceIndex device_index) -> ()";
+  static void call(at::DeviceIndex device_index);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::DeviceIndex device_index);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_get_plan_cache_max_size_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_get_plan_cache_max_size_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..272236e4e2ee46514b79444ef7ced1dd116fc593
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cufft_get_plan_cache_max_size_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API int64_t _cufft_get_plan_cache_max_size(at::DeviceIndex device_index);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_cummax_helper_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_cummax_helper_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..90088840724de01cd028d561b30429dd746f545e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_cummax_helper_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 cummax_helper_cpu(const at::Tensor & self, at::Tensor & values, at::Tensor & indices, int64_t dim);
+TORCH_API void cummax_helper_cuda(const at::Tensor & self, at::Tensor & values, at::Tensor & indices, int64_t dim);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_dirichlet_grad_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_dirichlet_grad_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f5bba09c28ab025abe774d899e68f19924a7f914
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_dirichlet_grad_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 & _dirichlet_grad_out(at::Tensor & out, const at::Tensor & x, const at::Tensor & alpha, const at::Tensor & total);
+TORCH_API at::Tensor & _dirichlet_grad_outf(const at::Tensor & x, const at::Tensor & alpha, const at::Tensor & total, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_dirichlet_grad_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_dirichlet_grad_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cad3753cceb79ccebd01473690f95c6dea50f0a9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_dirichlet_grad_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 _dirichlet_grad(const at::Tensor & x, const at::Tensor & alpha, const at::Tensor & total);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_efficient_attention_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_efficient_attention_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..f40ee7426940d67a3af2ccd3c14ae836018ee6d3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_efficient_attention_backward.h
@@ -0,0 +1,47 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_efficient_attention_backward_ops.h>
+
+namespace at {
+
+
+// aten::_efficient_attention_backward(Tensor grad_out_, Tensor query, Tensor key, Tensor value, Tensor? bias, Tensor out, Tensor? cu_seqlens_q, Tensor? cu_seqlens_k, SymInt max_seqlen_q, SymInt max_seqlen_k, Tensor logsumexp, float dropout_p, Tensor philox_seed, Tensor philox_offset, int custom_mask_type, bool bias_requires_grad, *, float? scale=None, int? num_splits_key=None, int? window_size=None, bool shared_storage_dqdkdv=False) -> (Tensor, Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> _efficient_attention_backward(const at::Tensor & grad_out_, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const at::Tensor & out, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, int64_t max_seqlen_q, int64_t max_seqlen_k, const at::Tensor & logsumexp, double dropout_p, const at::Tensor & philox_seed, const at::Tensor & philox_offset, int64_t custom_mask_type, bool bias_requires_grad, ::std::optional<double> scale=::std::nullopt, ::std::optional<int64_t> num_splits_key=::std::nullopt, ::std::optional<int64_t> window_size=::std::nullopt, bool shared_storage_dqdkdv=false) {
+    return at::_ops::_efficient_attention_backward::call(grad_out_, query, key, value, bias, out, cu_seqlens_q, cu_seqlens_k, max_seqlen_q, max_seqlen_k, logsumexp, dropout_p, philox_seed, philox_offset, custom_mask_type, bias_requires_grad, scale, num_splits_key, window_size, shared_storage_dqdkdv);
+}
+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> _efficient_attention_backward(const at::Tensor & grad_out_, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const at::Tensor & out, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, int64_t max_seqlen_q, int64_t max_seqlen_k, const at::Tensor & logsumexp, double dropout_p, const at::Tensor & philox_seed, const at::Tensor & philox_offset, int64_t custom_mask_type, bool bias_requires_grad, ::std::optional<double> scale=::std::nullopt, ::std::optional<int64_t> num_splits_key=::std::nullopt, ::std::optional<int64_t> window_size=::std::nullopt, bool shared_storage_dqdkdv=false) {
+    return at::_ops::_efficient_attention_backward::call(grad_out_, query, key, value, bias, out, cu_seqlens_q, cu_seqlens_k, max_seqlen_q, max_seqlen_k, logsumexp, dropout_p, philox_seed, philox_offset, custom_mask_type, bias_requires_grad, scale, num_splits_key, window_size, shared_storage_dqdkdv);
+  }
+}
+
+// aten::_efficient_attention_backward(Tensor grad_out_, Tensor query, Tensor key, Tensor value, Tensor? bias, Tensor out, Tensor? cu_seqlens_q, Tensor? cu_seqlens_k, SymInt max_seqlen_q, SymInt max_seqlen_k, Tensor logsumexp, float dropout_p, Tensor philox_seed, Tensor philox_offset, int custom_mask_type, bool bias_requires_grad, *, float? scale=None, int? num_splits_key=None, int? window_size=None, bool shared_storage_dqdkdv=False) -> (Tensor, Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> _efficient_attention_backward_symint(const at::Tensor & grad_out_, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const at::Tensor & out, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, c10::SymInt max_seqlen_q, c10::SymInt max_seqlen_k, const at::Tensor & logsumexp, double dropout_p, const at::Tensor & philox_seed, const at::Tensor & philox_offset, int64_t custom_mask_type, bool bias_requires_grad, ::std::optional<double> scale=::std::nullopt, ::std::optional<int64_t> num_splits_key=::std::nullopt, ::std::optional<int64_t> window_size=::std::nullopt, bool shared_storage_dqdkdv=false) {
+    return at::_ops::_efficient_attention_backward::call(grad_out_, query, key, value, bias, out, cu_seqlens_q, cu_seqlens_k, max_seqlen_q, max_seqlen_k, logsumexp, dropout_p, philox_seed, philox_offset, custom_mask_type, bias_requires_grad, scale, num_splits_key, window_size, shared_storage_dqdkdv);
+}
+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> _efficient_attention_backward(const at::Tensor & grad_out_, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const at::Tensor & out, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, c10::SymInt max_seqlen_q, c10::SymInt max_seqlen_k, const at::Tensor & logsumexp, double dropout_p, const at::Tensor & philox_seed, const at::Tensor & philox_offset, int64_t custom_mask_type, bool bias_requires_grad, ::std::optional<double> scale=::std::nullopt, ::std::optional<int64_t> num_splits_key=::std::nullopt, ::std::optional<int64_t> window_size=::std::nullopt, bool shared_storage_dqdkdv=false) {
+    return at::_ops::_efficient_attention_backward::call(grad_out_, query, key, value, bias, out, cu_seqlens_q, cu_seqlens_k, max_seqlen_q, max_seqlen_k, logsumexp, dropout_p, philox_seed, philox_offset, custom_mask_type, bias_requires_grad, scale, num_splits_key, window_size, shared_storage_dqdkdv);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_efficientzerotensor_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_efficientzerotensor_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d12d44e4be536991210588957c36ddd4bcc08178
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_efficientzerotensor_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 & _efficientzerotensor_out_symint(c10::SymIntArrayRef size, at::Tensor & out);
+TORCH_API at::Tensor _efficientzerotensor(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 _efficientzerotensor_cuda(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 _efficientzerotensor_meta_symint(c10::SymIntArrayRef size, ::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/_embedding_bag_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..e0d698948cac62fde19301f9c36f6ad01d395d1f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_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 _embedding_bag {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, bool, int64_t, bool, const ::std::optional<at::Tensor> &, bool, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_embedding_bag";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_embedding_bag(Tensor weight, Tensor indices, Tensor offsets, bool scale_grad_by_freq=False, int mode=0, bool sparse=False, Tensor? per_sample_weights=None, bool include_last_offset=False, int padding_idx=-1) -> (Tensor, Tensor, Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, bool include_last_offset, int64_t padding_idx);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, bool include_last_offset, int64_t padding_idx);
+};
+
+struct TORCH_API _embedding_bag_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> (const at::Tensor &, const at::Tensor &, const at::Tensor &, bool, int64_t, bool, const ::std::optional<at::Tensor> &, bool, int64_t, at::Tensor &, 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::_embedding_bag";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_embedding_bag.out(Tensor weight, Tensor indices, Tensor offsets, bool scale_grad_by_freq=False, int mode=0, bool sparse=False, Tensor? per_sample_weights=None, bool include_last_offset=False, int padding_idx=-1, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2, Tensor(d!) out3) -> (Tensor(a!), Tensor(b!), Tensor(c!), Tensor(d!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> call(const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, bool include_last_offset, int64_t padding_idx, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3);
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, bool include_last_offset, int64_t padding_idx, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_empty_affine_quantized.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_empty_affine_quantized.h
new file mode 100644
index 0000000000000000000000000000000000000000..75c2ef2590f5af7a01e04aaf9057d4a56b231d95
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_empty_affine_quantized.h
@@ -0,0 +1,113 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_empty_affine_quantized_ops.h>
+
+namespace at {
+
+
+// aten::_empty_affine_quantized(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, float scale=1, int zero_point=0, MemoryFormat? memory_format=contiguous_format) -> Tensor
+inline at::Tensor _empty_affine_quantized(at::IntArrayRef size, at::TensorOptions options={}, double scale=1, int64_t zero_point=0, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous) {
+    return at::_ops::_empty_affine_quantized::call(c10::fromIntArrayRefSlow(size), c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt(), scale, zero_point, c10::impl::check_tensor_options_and_extract_memory_format(options, memory_format));
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor _empty_affine_quantized(at::IntArrayRef size, at::TensorOptions options={}, double scale=1, int64_t zero_point=0, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous) {
+    return at::_ops::_empty_affine_quantized::call(c10::fromIntArrayRefSlow(size), c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt(), scale, zero_point, c10::impl::check_tensor_options_and_extract_memory_format(options, memory_format));
+  }
+}
+
+// aten::_empty_affine_quantized(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, float scale=1, int zero_point=0, MemoryFormat? memory_format=contiguous_format) -> Tensor
+inline at::Tensor _empty_affine_quantized(at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, double scale, int64_t zero_point, ::std::optional<at::MemoryFormat> memory_format) {
+    return at::_ops::_empty_affine_quantized::call(c10::fromIntArrayRefSlow(size), dtype, layout, device, pin_memory, scale, zero_point, memory_format);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor _empty_affine_quantized(at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, double scale, int64_t zero_point, ::std::optional<at::MemoryFormat> memory_format) {
+    return at::_ops::_empty_affine_quantized::call(c10::fromIntArrayRefSlow(size), dtype, layout, device, pin_memory, scale, zero_point, memory_format);
+  }
+}
+
+// aten::_empty_affine_quantized(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, float scale=1, int zero_point=0, MemoryFormat? memory_format=contiguous_format) -> Tensor
+inline at::Tensor _empty_affine_quantized_symint(c10::SymIntArrayRef size, at::TensorOptions options={}, double scale=1, int64_t zero_point=0, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous) {
+    return at::_ops::_empty_affine_quantized::call(size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt(), scale, zero_point, c10::impl::check_tensor_options_and_extract_memory_format(options, memory_format));
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor _empty_affine_quantized(c10::SymIntArrayRef size, at::TensorOptions options={}, double scale=1, int64_t zero_point=0, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous) {
+    return at::_ops::_empty_affine_quantized::call(size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt(), scale, zero_point, c10::impl::check_tensor_options_and_extract_memory_format(options, memory_format));
+  }
+}
+
+// aten::_empty_affine_quantized(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, float scale=1, int zero_point=0, MemoryFormat? memory_format=contiguous_format) -> Tensor
+inline at::Tensor _empty_affine_quantized_symint(c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, double scale, int64_t zero_point, ::std::optional<at::MemoryFormat> memory_format) {
+    return at::_ops::_empty_affine_quantized::call(size, dtype, layout, device, pin_memory, scale, zero_point, memory_format);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor _empty_affine_quantized(c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, double scale, int64_t zero_point, ::std::optional<at::MemoryFormat> memory_format) {
+    return at::_ops::_empty_affine_quantized::call(size, dtype, layout, device, pin_memory, scale, zero_point, memory_format);
+  }
+}
+
+// aten::_empty_affine_quantized.out(SymInt[] size, *, float scale=1, int zero_point=0, MemoryFormat? memory_format=contiguous_format, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _empty_affine_quantized_out(at::Tensor & out, at::IntArrayRef size, double scale=1, int64_t zero_point=0, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous) {
+    return at::_ops::_empty_affine_quantized_out::call(c10::fromIntArrayRefSlow(size), scale, zero_point, memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & _empty_affine_quantized_out(at::Tensor & out, at::IntArrayRef size, double scale=1, int64_t zero_point=0, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous) {
+    return at::_ops::_empty_affine_quantized_out::call(c10::fromIntArrayRefSlow(size), scale, zero_point, memory_format, out);
+  }
+}
+
+// aten::_empty_affine_quantized.out(SymInt[] size, *, float scale=1, int zero_point=0, MemoryFormat? memory_format=contiguous_format, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _empty_affine_quantized_outf(at::IntArrayRef size, double scale, int64_t zero_point, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out) {
+    return at::_ops::_empty_affine_quantized_out::call(c10::fromIntArrayRefSlow(size), scale, zero_point, memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & _empty_affine_quantized_outf(at::IntArrayRef size, double scale, int64_t zero_point, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out) {
+    return at::_ops::_empty_affine_quantized_out::call(c10::fromIntArrayRefSlow(size), scale, zero_point, memory_format, out);
+  }
+}
+
+// aten::_empty_affine_quantized.out(SymInt[] size, *, float scale=1, int zero_point=0, MemoryFormat? memory_format=contiguous_format, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _empty_affine_quantized_symint_out(at::Tensor & out, c10::SymIntArrayRef size, double scale=1, int64_t zero_point=0, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous) {
+    return at::_ops::_empty_affine_quantized_out::call(size, scale, zero_point, memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & _empty_affine_quantized_out(at::Tensor & out, c10::SymIntArrayRef size, double scale=1, int64_t zero_point=0, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous) {
+    return at::_ops::_empty_affine_quantized_out::call(size, scale, zero_point, memory_format, out);
+  }
+}
+
+// aten::_empty_affine_quantized.out(SymInt[] size, *, float scale=1, int zero_point=0, MemoryFormat? memory_format=contiguous_format, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _empty_affine_quantized_symint_outf(c10::SymIntArrayRef size, double scale, int64_t zero_point, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out) {
+    return at::_ops::_empty_affine_quantized_out::call(size, scale, zero_point, memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & _empty_affine_quantized_outf(c10::SymIntArrayRef size, double scale, int64_t zero_point, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out) {
+    return at::_ops::_empty_affine_quantized_out::call(size, scale, zero_point, memory_format, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_empty_affine_quantized_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_empty_affine_quantized_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..959c10f73a1ffb41301b2fb996dd5bd90ccd4854
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_empty_affine_quantized_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_affine_quantized_out(at::Tensor & out, at::IntArrayRef size, double scale=1, int64_t zero_point=0, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous);
+TORCH_API at::Tensor & _empty_affine_quantized_outf(at::IntArrayRef size, double scale, int64_t zero_point, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+TORCH_API at::Tensor & _empty_affine_quantized_symint_out(at::Tensor & out, c10::SymIntArrayRef size, double scale=1, int64_t zero_point=0, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous);
+TORCH_API at::Tensor & _empty_affine_quantized_symint_outf(c10::SymIntArrayRef size, double scale, int64_t zero_point, ::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/_fake_quantize_learnable_per_channel_affine.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_channel_affine.h
new file mode 100644
index 0000000000000000000000000000000000000000..ed0f6edacd986f18c4b3042d046bc401eb7bf587
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_channel_affine.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_fake_quantize_learnable_per_channel_affine_ops.h>
+
+namespace at {
+
+
+// aten::_fake_quantize_learnable_per_channel_affine(Tensor self, Tensor scale, Tensor zero_point, int axis, int quant_min, int quant_max, float grad_factor=1.0) -> Tensor
+inline at::Tensor _fake_quantize_learnable_per_channel_affine(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max, double grad_factor=1.0) {
+    return at::_ops::_fake_quantize_learnable_per_channel_affine::call(self, scale, zero_point, axis, quant_min, quant_max, grad_factor);
+}
+
+// aten::_fake_quantize_learnable_per_channel_affine.out(Tensor self, Tensor scale, Tensor zero_point, int axis, int quant_min, int quant_max, float grad_factor=1.0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _fake_quantize_learnable_per_channel_affine_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max, double grad_factor=1.0) {
+    return at::_ops::_fake_quantize_learnable_per_channel_affine_out::call(self, scale, zero_point, axis, quant_min, quant_max, grad_factor, out);
+}
+// aten::_fake_quantize_learnable_per_channel_affine.out(Tensor self, Tensor scale, Tensor zero_point, int axis, int quant_min, int quant_max, float grad_factor=1.0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _fake_quantize_learnable_per_channel_affine_outf(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max, double grad_factor, at::Tensor & out) {
+    return at::_ops::_fake_quantize_learnable_per_channel_affine_out::call(self, scale, zero_point, axis, quant_min, quant_max, grad_factor, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_tensor_affine_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_tensor_affine_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..044d98cbec3c38c42e7b0f6399f8e21d1bc7eba9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_tensor_affine_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 _fake_quantize_learnable_per_tensor_affine(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t quant_min, int64_t quant_max, double grad_factor=1.0);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fft_c2c_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fft_c2c_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..dac223687c607b761ff09e7671d3cc937dee3690
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fft_c2c_cuda_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor _fft_c2c(const at::Tensor & self, at::IntArrayRef dim, int64_t normalization, bool forward);
+TORCH_API at::Tensor _fft_c2c_symint(const at::Tensor & self, c10::SymIntArrayRef dim, int64_t normalization, bool forward);
+TORCH_API at::Tensor & _fft_c2c_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim, int64_t normalization, bool forward);
+TORCH_API at::Tensor & _fft_c2c_outf(const at::Tensor & self, at::IntArrayRef dim, int64_t normalization, bool forward, at::Tensor & out);
+TORCH_API at::Tensor & _fft_c2c_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef dim, int64_t normalization, bool forward);
+TORCH_API at::Tensor & _fft_c2c_symint_outf(const at::Tensor & self, c10::SymIntArrayRef dim, int64_t normalization, bool forward, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_ceil_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_ceil_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..c018d40a4c6dfdd6a139508e377d7f09d1b3d37e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_ceil_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_ceil {
+  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_ceil";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_ceil(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_ceil_ {
+  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_ceil_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_ceil_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_ceil_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_ceil";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_ceil.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_erf_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_erf_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d9f105c9250fcf9d68fdb236ff05cf24ccf3dd7e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_erf_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_erf(at::TensorList self);
+TORCH_API void _foreach_erf_out(at::TensorList out, at::TensorList self);
+TORCH_API void _foreach_erf_outf(at::TensorList self, at::TensorList out);
+TORCH_API void _foreach_erf_(at::TensorList self);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_erf_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_erf_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..73e3fddfd0ec1c7581acc9bece5de294819a03c6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_erf_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_erf_slow(at::TensorList self);
+TORCH_API void _foreach_erf_out(at::TensorList self, at::TensorList out);
+TORCH_API void foreach_tensor_erf_slow_(at::TensorList self);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_erf_cuda(at::TensorList self);
+TORCH_API void foreach_tensor_erf_cuda_(at::TensorList self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_exp_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_exp_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..082771bc5c96ecae2f933bef259f32a6a21dfb08
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_exp_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_exp {
+  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_exp";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_exp(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_exp_ {
+  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_exp_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_exp_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_exp_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_exp";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_exp.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_expm1.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_expm1.h
new file mode 100644
index 0000000000000000000000000000000000000000..6e7ad09ecdf59b82f1762a0de4f5d2b9fb3e92a0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_expm1.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_expm1_ops.h>
+
+namespace at {
+
+
+// aten::_foreach_expm1(Tensor[] self) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_expm1(at::TensorList self) {
+    return at::_ops::_foreach_expm1::call(self);
+}
+
+// aten::_foreach_expm1_(Tensor(a!)[] self) -> ()
+inline void _foreach_expm1_(at::TensorList self) {
+    return at::_ops::_foreach_expm1_::call(self);
+}
+
+// aten::_foreach_expm1.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_expm1_out(at::TensorList out, at::TensorList self) {
+    return at::_ops::_foreach_expm1_out::call(self, out);
+}
+// aten::_foreach_expm1.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_expm1_outf(at::TensorList self, at::TensorList out) {
+    return at::_ops::_foreach_expm1_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_maximum_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_maximum_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d1d6ae1015a90658fe95bc22492e1bae146b44e7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_maximum_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_min_scalar_kernel_slow(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_maximum_Scalar_out(at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+TORCH_API void foreach_tensor_clamp_min_scalar_kernel_slow_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_min_scalar_kernel_cuda(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void foreach_tensor_clamp_min_scalar_kernel_cuda_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_min_list_kernel_slow(at::TensorList self, at::TensorList other);
+TORCH_API void _foreach_maximum_List_out(at::TensorList self, at::TensorList other, at::TensorList out);
+TORCH_API void foreach_tensor_clamp_min_list_kernel_slow_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_min_list_kernel_cuda(at::TensorList self, at::TensorList other);
+TORCH_API void foreach_tensor_clamp_min_list_kernel_cuda_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_min_scalarlist_kernel_slow(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_maximum_ScalarList_out(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+TORCH_API void foreach_tensor_clamp_min_scalarlist_kernel_slow_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_min_scalarlist_kernel_cuda(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void foreach_tensor_clamp_min_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_maximum_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_maximum_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1f00c4e4594dcc1948d154e4b469c669022bb825
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_maximum_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 _foreach_maximum_Scalar {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_maximum";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "_foreach_maximum.Scalar(Tensor[] self, Scalar scalar) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(at::TensorList self, const at::Scalar & scalar);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Scalar & scalar);
+};
+
+struct TORCH_API _foreach_maximum__Scalar {
+  using schema = void (at::TensorList, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_maximum_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "_foreach_maximum_.Scalar(Tensor(a!)[] self, Scalar scalar) -> ()";
+  static void call(at::TensorList self, const at::Scalar & scalar);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Scalar & scalar);
+};
+
+struct TORCH_API _foreach_maximum_List {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_maximum";
+  static constexpr const char* overload_name = "List";
+  static constexpr const char* schema_str = "_foreach_maximum.List(Tensor[] self, Tensor[] other) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(at::TensorList self, at::TensorList other);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::TensorList other);
+};
+
+struct TORCH_API _foreach_maximum__List {
+  using schema = void (at::TensorList, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_maximum_";
+  static constexpr const char* overload_name = "List";
+  static constexpr const char* schema_str = "_foreach_maximum_.List(Tensor(a!)[] self, Tensor[] other) -> ()";
+  static void call(at::TensorList self, at::TensorList other);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::TensorList other);
+};
+
+struct TORCH_API _foreach_maximum_ScalarList {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, at::ArrayRef<at::Scalar>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_maximum";
+  static constexpr const char* overload_name = "ScalarList";
+  static constexpr const char* schema_str = "_foreach_maximum.ScalarList(Tensor[] self, Scalar[] scalars) -> Tensor[]";
+  static ::std::vector<at::Tensor> call(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+};
+
+struct TORCH_API _foreach_maximum__ScalarList {
+  using schema = void (at::TensorList, at::ArrayRef<at::Scalar>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_maximum_";
+  static constexpr const char* overload_name = "ScalarList";
+  static constexpr const char* schema_str = "_foreach_maximum_.ScalarList(Tensor(a!)[] self, Scalar[] scalars) -> ()";
+  static void call(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+};
+
+struct TORCH_API _foreach_maximum_Scalar_out {
+  using schema = void (at::TensorList, const at::Scalar &, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_maximum";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "_foreach_maximum.Scalar_out(Tensor[] self, Scalar scalar, *, Tensor(a!)[] out) -> ()";
+  static void call(at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+};
+
+struct TORCH_API _foreach_maximum_List_out {
+  using schema = void (at::TensorList, at::TensorList, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_maximum";
+  static constexpr const char* overload_name = "List_out";
+  static constexpr const char* schema_str = "_foreach_maximum.List_out(Tensor[] self, Tensor[] other, *, Tensor(a!)[] out) -> ()";
+  static void call(at::TensorList self, at::TensorList other, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::TensorList other, at::TensorList out);
+};
+
+struct TORCH_API _foreach_maximum_ScalarList_out {
+  using schema = void (at::TensorList, at::ArrayRef<at::Scalar>, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_foreach_maximum";
+  static constexpr const char* overload_name = "ScalarList_out";
+  static constexpr const char* schema_str = "_foreach_maximum.ScalarList_out(Tensor[] self, Scalar[] scalars, *, Tensor(a!)[] out) -> ()";
+  static void call(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum.h
new file mode 100644
index 0000000000000000000000000000000000000000..115cbfcb2d6a8e81c7d6a2f38ec6ba73e86e3ea5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum.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_minimum_ops.h>
+
+namespace at {
+
+
+// aten::_foreach_minimum.Scalar(Tensor[] self, Scalar scalar) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_minimum(at::TensorList self, const at::Scalar & scalar) {
+    return at::_ops::_foreach_minimum_Scalar::call(self, scalar);
+}
+
+// aten::_foreach_minimum_.Scalar(Tensor(a!)[] self, Scalar scalar) -> ()
+inline void _foreach_minimum_(at::TensorList self, const at::Scalar & scalar) {
+    return at::_ops::_foreach_minimum__Scalar::call(self, scalar);
+}
+
+// aten::_foreach_minimum.List(Tensor[] self, Tensor[] other) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_minimum(at::TensorList self, at::TensorList other) {
+    return at::_ops::_foreach_minimum_List::call(self, other);
+}
+
+// aten::_foreach_minimum_.List(Tensor(a!)[] self, Tensor[] other) -> ()
+inline void _foreach_minimum_(at::TensorList self, at::TensorList other) {
+    return at::_ops::_foreach_minimum__List::call(self, other);
+}
+
+// aten::_foreach_minimum.ScalarList(Tensor[] self, Scalar[] scalars) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_minimum(at::TensorList self, at::ArrayRef<at::Scalar> scalars) {
+    return at::_ops::_foreach_minimum_ScalarList::call(self, scalars);
+}
+
+// aten::_foreach_minimum_.ScalarList(Tensor(a!)[] self, Scalar[] scalars) -> ()
+inline void _foreach_minimum_(at::TensorList self, at::ArrayRef<at::Scalar> scalars) {
+    return at::_ops::_foreach_minimum__ScalarList::call(self, scalars);
+}
+
+// aten::_foreach_minimum.Scalar_out(Tensor[] self, Scalar scalar, *, Tensor(a!)[] out) -> ()
+inline void _foreach_minimum_out(at::TensorList out, at::TensorList self, const at::Scalar & scalar) {
+    return at::_ops::_foreach_minimum_Scalar_out::call(self, scalar, out);
+}
+// aten::_foreach_minimum.Scalar_out(Tensor[] self, Scalar scalar, *, Tensor(a!)[] out) -> ()
+inline void _foreach_minimum_outf(at::TensorList self, const at::Scalar & scalar, at::TensorList out) {
+    return at::_ops::_foreach_minimum_Scalar_out::call(self, scalar, out);
+}
+
+// aten::_foreach_minimum.List_out(Tensor[] self, Tensor[] other, *, Tensor(a!)[] out) -> ()
+inline void _foreach_minimum_out(at::TensorList out, at::TensorList self, at::TensorList other) {
+    return at::_ops::_foreach_minimum_List_out::call(self, other, out);
+}
+// aten::_foreach_minimum.List_out(Tensor[] self, Tensor[] other, *, Tensor(a!)[] out) -> ()
+inline void _foreach_minimum_outf(at::TensorList self, at::TensorList other, at::TensorList out) {
+    return at::_ops::_foreach_minimum_List_out::call(self, other, out);
+}
+
+// aten::_foreach_minimum.ScalarList_out(Tensor[] self, Scalar[] scalars, *, Tensor(a!)[] out) -> ()
+inline void _foreach_minimum_out(at::TensorList out, at::TensorList self, at::ArrayRef<at::Scalar> scalars) {
+    return at::_ops::_foreach_minimum_ScalarList_out::call(self, scalars, out);
+}
+// aten::_foreach_minimum.ScalarList_out(Tensor[] self, Scalar[] scalars, *, Tensor(a!)[] out) -> ()
+inline void _foreach_minimum_outf(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out) {
+    return at::_ops::_foreach_minimum_ScalarList_out::call(self, scalars, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..902e58633e39732f3165f1b65e887bc2e3f6857e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_minimum_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_minimum(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_minimum_out(at::TensorList out, at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_minimum_outf(at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+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_out(at::TensorList out, at::TensorList self, at::TensorList other);
+TORCH_API void _foreach_minimum_outf(at::TensorList self, at::TensorList other, at::TensorList out);
+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_out(at::TensorList out, at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_minimum_outf(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+TORCH_API void _foreach_minimum_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_tanh_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_tanh_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6ec3acc32cadfb06cd2ee1498e1f70ff361feb35
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_tanh_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_tanh(at::TensorList self);
+TORCH_API void _foreach_tanh_out(at::TensorList out, at::TensorList self);
+TORCH_API void _foreach_tanh_outf(at::TensorList self, at::TensorList out);
+TORCH_API void _foreach_tanh_(at::TensorList self);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_assert_scalar_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_assert_scalar_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4b400fb1d091342135f9e1f5a2d97a3d4abfa5c8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_assert_scalar_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 _functional_assert_scalar(const at::Scalar & self, c10::string_view assert_msg, const at::Tensor & dep_token);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_sym_constrain_range.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_sym_constrain_range.h
new file mode 100644
index 0000000000000000000000000000000000000000..877faba919d31cfbe05629127554e7d62716b84e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_sym_constrain_range.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_sym_constrain_range_ops.h>
+
+namespace at {
+
+
+// aten::_functional_sym_constrain_range(Scalar size, int? min, int? max, Tensor dep_token) -> Tensor
+inline at::Tensor _functional_sym_constrain_range(const at::Scalar & size, ::std::optional<int64_t> min, ::std::optional<int64_t> max, const at::Tensor & dep_token) {
+    return at::_ops::_functional_sym_constrain_range::call(size, min, max, dep_token);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_sym_constrain_range_for_size_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_sym_constrain_range_for_size_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5641355f8211169772662a83e9c118bc489163a2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_functional_sym_constrain_range_for_size_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor _functional_sym_constrain_range_for_size(const at::Scalar & size, ::std::optional<int64_t> min, ::std::optional<int64_t> max, const at::Tensor & dep_token);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adamw_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adamw_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..85bf650a835d73ca7161b3c0a563e856b964808d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adamw_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> _fused_adamw(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, double lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+TORCH_API void _fused_adamw_out(at::TensorList out, at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, double lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+TORCH_API void _fused_adamw_outf(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, double lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale, const ::std::optional<at::Tensor> & found_inf, at::TensorList out);
+TORCH_API ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> _fused_adamw(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, const at::Tensor & lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+TORCH_API void _fused_adamw_out(at::TensorList out, at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, const at::Tensor & lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+TORCH_API void _fused_adamw_outf(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, const at::Tensor & lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale, const ::std::optional<at::Tensor> & found_inf, at::TensorList out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adamw_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adamw_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b08f69cfc0d318039d63c279165d90fb4161119f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adamw_cuda_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API void _fused_adamw_(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, double lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+TORCH_API void _fused_adamw_(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, const at::Tensor & lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper.h
new file mode 100644
index 0000000000000000000000000000000000000000..f1cefe1c730a3f338f6bfe547e9ee7ef0cee19f3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper.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/_fused_moving_avg_obs_fq_helper_ops.h>
+
+namespace at {
+
+
+// aten::_fused_moving_avg_obs_fq_helper(Tensor self, Tensor observer_on, Tensor fake_quant_on, Tensor(a!) running_min, Tensor(b!) running_max, Tensor(c!) scale, Tensor(d!) zero_point, float averaging_const, int quant_min, int quant_max, int ch_axis, bool per_row_fake_quant=False, bool symmetric_quant=False) -> (Tensor output, Tensor mask)
+inline ::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) {
+    return at::_ops::_fused_moving_avg_obs_fq_helper::call(self, observer_on, fake_quant_on, running_min, running_max, scale, zero_point, averaging_const, quant_min, quant_max, ch_axis, per_row_fake_quant, symmetric_quant);
+}
+
+// aten::_fused_moving_avg_obs_fq_helper.out(Tensor self, Tensor observer_on, Tensor fake_quant_on, Tensor(a!) running_min, Tensor(b!) running_max, Tensor(c!) scale, Tensor(d!) zero_point, float averaging_const, int quant_min, int quant_max, int ch_axis, bool per_row_fake_quant=False, bool symmetric_quant=False, *, Tensor(e!) out0, Tensor(f!) out1) -> (Tensor(e!), Tensor(f!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _fused_moving_avg_obs_fq_helper_out(at::Tensor & out0, at::Tensor & out1, 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) {
+    return at::_ops::_fused_moving_avg_obs_fq_helper_out::call(self, observer_on, fake_quant_on, running_min, running_max, scale, zero_point, averaging_const, quant_min, quant_max, ch_axis, per_row_fake_quant, symmetric_quant, out0, out1);
+}
+// aten::_fused_moving_avg_obs_fq_helper.out(Tensor self, Tensor observer_on, Tensor fake_quant_on, Tensor(a!) running_min, Tensor(b!) running_max, Tensor(c!) scale, Tensor(d!) zero_point, float averaging_const, int quant_min, int quant_max, int ch_axis, bool per_row_fake_quant=False, bool symmetric_quant=False, *, Tensor(e!) out0, Tensor(f!) out1) -> (Tensor(e!), Tensor(f!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _fused_moving_avg_obs_fq_helper_outf(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, bool symmetric_quant, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::_fused_moving_avg_obs_fq_helper_out::call(self, observer_on, fake_quant_on, running_min, running_max, scale, zero_point, averaging_const, quant_min, quant_max, ch_axis, per_row_fake_quant, symmetric_quant, out0, out1);
+}
+
+// aten::_fused_moving_avg_obs_fq_helper_functional(Tensor self, Tensor observer_on, Tensor fake_quant_on, Tensor running_min, Tensor running_max, Tensor scale, Tensor zero_point, float averaging_const, int quant_min, int quant_max, int ch_axis, bool per_row_fake_quant=False, bool symmetric_quant=False) -> (Tensor output, Tensor mask, Tensor running_min_out, Tensor running_max_out, Tensor scale_out, Tensor zero_point_out)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _fused_moving_avg_obs_fq_helper_functional(const at::Tensor & self, const at::Tensor & observer_on, const at::Tensor & fake_quant_on, const at::Tensor & running_min, const at::Tensor & running_max, const at::Tensor & scale, const 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) {
+    return at::_ops::_fused_moving_avg_obs_fq_helper_functional::call(self, observer_on, fake_quant_on, running_min, running_max, scale, zero_point, averaging_const, quant_min, quant_max, ch_axis, per_row_fake_quant, symmetric_quant);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..80929c51e87de8f7a844b362c70faaad37a4fce0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_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<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _fused_moving_avg_obs_fq_helper_functional(const at::Tensor & self, const at::Tensor & observer_on, const at::Tensor & fake_quant_on, const at::Tensor & running_min, const at::Tensor & running_max, const at::Tensor & scale, const 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);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> _fused_moving_avg_obs_fq_helper_out(at::Tensor & out0, at::Tensor & out1, 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);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> _fused_moving_avg_obs_fq_helper_outf(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, bool symmetric_quant, at::Tensor & out0, at::Tensor & out1);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..7580e748c5ad144196704600196c39e6f9c30c8f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_moving_avg_obs_fq_helper_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 _fused_moving_avg_obs_fq_helper {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, at::Tensor &, at::Tensor &, at::Tensor &, at::Tensor &, double, int64_t, int64_t, int64_t, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fused_moving_avg_obs_fq_helper";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_fused_moving_avg_obs_fq_helper(Tensor self, Tensor observer_on, Tensor fake_quant_on, Tensor(a!) running_min, Tensor(b!) running_max, Tensor(c!) scale, Tensor(d!) zero_point, float averaging_const, int quant_min, int quant_max, int ch_axis, bool per_row_fake_quant=False, bool symmetric_quant=False) -> (Tensor output, Tensor mask)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(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, bool symmetric_quant);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, 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, bool symmetric_quant);
+};
+
+struct TORCH_API _fused_moving_avg_obs_fq_helper_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const at::Tensor &, const at::Tensor &, const at::Tensor &, at::Tensor &, at::Tensor &, at::Tensor &, at::Tensor &, double, int64_t, int64_t, int64_t, bool, bool, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fused_moving_avg_obs_fq_helper";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_fused_moving_avg_obs_fq_helper.out(Tensor self, Tensor observer_on, Tensor fake_quant_on, Tensor(a!) running_min, Tensor(b!) running_max, Tensor(c!) scale, Tensor(d!) zero_point, float averaging_const, int quant_min, int quant_max, int ch_axis, bool per_row_fake_quant=False, bool symmetric_quant=False, *, Tensor(e!) out0, Tensor(f!) out1) -> (Tensor(e!), Tensor(f!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(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, bool symmetric_quant, at::Tensor & out0, at::Tensor & out1);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, 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, bool symmetric_quant, at::Tensor & out0, at::Tensor & out1);
+};
+
+struct TORCH_API _fused_moving_avg_obs_fq_helper_functional {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, double, int64_t, int64_t, int64_t, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fused_moving_avg_obs_fq_helper_functional";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_fused_moving_avg_obs_fq_helper_functional(Tensor self, Tensor observer_on, Tensor fake_quant_on, Tensor running_min, Tensor running_max, Tensor scale, Tensor zero_point, float averaging_const, int quant_min, int quant_max, int ch_axis, bool per_row_fake_quant=False, bool symmetric_quant=False) -> (Tensor output, Tensor mask, Tensor running_min_out, Tensor running_max_out, Tensor scale_out, Tensor zero_point_out)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & self, const at::Tensor & observer_on, const at::Tensor & fake_quant_on, const at::Tensor & running_min, const at::Tensor & running_max, const at::Tensor & scale, const at::Tensor & zero_point, double averaging_const, int64_t quant_min, int64_t quant_max, int64_t ch_axis, bool per_row_fake_quant, bool symmetric_quant);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & observer_on, const at::Tensor & fake_quant_on, const at::Tensor & running_min, const at::Tensor & running_max, const at::Tensor & scale, const at::Tensor & zero_point, double averaging_const, int64_t quant_min, int64_t quant_max, int64_t ch_axis, bool per_row_fake_quant, bool symmetric_quant);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_sgd.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_sgd.h
new file mode 100644
index 0000000000000000000000000000000000000000..f73cd2ac84a050b539a6ec32bd512b3cddfba03b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_fused_sgd.h
@@ -0,0 +1,63 @@
+#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/_fused_sgd_ops.h>
+
+namespace at {
+
+
+// aten::_fused_sgd_(Tensor(a!)[] self, Tensor(b!)[] grads, Tensor(c!)[] momentum_buffer_list, *, float weight_decay, float momentum, float lr, float dampening, bool nesterov, bool maximize, bool is_first_step, Tensor? grad_scale=None, Tensor? found_inf=None) -> ()
+inline void _fused_sgd_(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, double lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={}) {
+    return at::_ops::_fused_sgd_::call(self, grads, momentum_buffer_list, weight_decay, momentum, lr, dampening, nesterov, maximize, is_first_step, grad_scale, found_inf);
+}
+
+// aten::_fused_sgd_.tensor_lr(Tensor(a!)[] self, Tensor(b!)[] grads, Tensor(c!)[] momentum_buffer_list, *, float weight_decay, float momentum, Tensor lr, float dampening, bool nesterov, bool maximize, bool is_first_step, Tensor? grad_scale=None, Tensor? found_inf=None) -> ()
+inline void _fused_sgd_(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, const at::Tensor & lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={}) {
+    return at::_ops::_fused_sgd__tensor_lr::call(self, grads, momentum_buffer_list, weight_decay, momentum, lr, dampening, nesterov, maximize, is_first_step, grad_scale, found_inf);
+}
+
+// aten::_fused_sgd.out(Tensor[] self, Tensor(b!)[] grads, Tensor(c!)[] momentum_buffer_list, *, float weight_decay, float momentum, float lr, float dampening, bool nesterov, bool maximize, bool is_first_step, Tensor? grad_scale=None, Tensor? found_inf=None, Tensor(a!)[] out) -> ()
+inline void _fused_sgd_out(at::TensorList out, at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, double lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={}) {
+    return at::_ops::_fused_sgd_out::call(self, grads, momentum_buffer_list, weight_decay, momentum, lr, dampening, nesterov, maximize, is_first_step, grad_scale, found_inf, out);
+}
+// aten::_fused_sgd.out(Tensor[] self, Tensor(b!)[] grads, Tensor(c!)[] momentum_buffer_list, *, float weight_decay, float momentum, float lr, float dampening, bool nesterov, bool maximize, bool is_first_step, Tensor? grad_scale=None, Tensor? found_inf=None, Tensor(a!)[] out) -> ()
+inline void _fused_sgd_outf(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, double lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale, const ::std::optional<at::Tensor> & found_inf, at::TensorList out) {
+    return at::_ops::_fused_sgd_out::call(self, grads, momentum_buffer_list, weight_decay, momentum, lr, dampening, nesterov, maximize, is_first_step, grad_scale, found_inf, out);
+}
+
+// aten::_fused_sgd(Tensor[] self, Tensor[] grads, Tensor[] momentum_buffer_list, *, float weight_decay, float momentum, float lr, float dampening, bool nesterov, bool maximize, bool is_first_step, Tensor? grad_scale=None, Tensor? found_inf=None) -> (Tensor[] self_out, Tensor[] grads_out, Tensor[] momentum_buffer_list_out)
+inline ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> _fused_sgd(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, double lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={}) {
+    return at::_ops::_fused_sgd::call(self, grads, momentum_buffer_list, weight_decay, momentum, lr, dampening, nesterov, maximize, is_first_step, grad_scale, found_inf);
+}
+
+// aten::_fused_sgd.tensor_lr_out(Tensor[] self, Tensor(b!)[] grads, Tensor(c!)[] momentum_buffer_list, *, float weight_decay, float momentum, Tensor lr, float dampening, bool nesterov, bool maximize, bool is_first_step, Tensor? grad_scale=None, Tensor? found_inf=None, Tensor(a!)[] out) -> ()
+inline void _fused_sgd_out(at::TensorList out, at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, const at::Tensor & lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={}) {
+    return at::_ops::_fused_sgd_tensor_lr_out::call(self, grads, momentum_buffer_list, weight_decay, momentum, lr, dampening, nesterov, maximize, is_first_step, grad_scale, found_inf, out);
+}
+// aten::_fused_sgd.tensor_lr_out(Tensor[] self, Tensor(b!)[] grads, Tensor(c!)[] momentum_buffer_list, *, float weight_decay, float momentum, Tensor lr, float dampening, bool nesterov, bool maximize, bool is_first_step, Tensor? grad_scale=None, Tensor? found_inf=None, Tensor(a!)[] out) -> ()
+inline void _fused_sgd_outf(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, const at::Tensor & lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale, const ::std::optional<at::Tensor> & found_inf, at::TensorList out) {
+    return at::_ops::_fused_sgd_tensor_lr_out::call(self, grads, momentum_buffer_list, weight_decay, momentum, lr, dampening, nesterov, maximize, is_first_step, grad_scale, found_inf, out);
+}
+
+// aten::_fused_sgd.tensor_lr(Tensor[] self, Tensor[] grads, Tensor[] momentum_buffer_list, *, float weight_decay, float momentum, Tensor lr, float dampening, bool nesterov, bool maximize, bool is_first_step, Tensor? grad_scale=None, Tensor? found_inf=None) -> (Tensor[] self_out, Tensor[] grads_out, Tensor[] momentum_buffer_list_out)
+inline ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> _fused_sgd(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, const at::Tensor & lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={}) {
+    return at::_ops::_fused_sgd_tensor_lr::call(self, grads, momentum_buffer_list, weight_decay, momentum, lr, dampening, nesterov, maximize, is_first_step, grad_scale, found_inf);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_grid_sampler_2d_cpu_fallback_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_grid_sampler_2d_cpu_fallback_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..0a6034ab6c78984b82917ff897f1a9d60652d626
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_grid_sampler_2d_cpu_fallback_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 _grid_sampler_2d_cpu_fallback(const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners);
+TORCH_API at::Tensor & _grid_sampler_2d_cpu_fallback_out(const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_has_compatible_shallow_copy_type.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_has_compatible_shallow_copy_type.h
new file mode 100644
index 0000000000000000000000000000000000000000..428f61e986e7306e5642b98d91a4e8d2179b9e5b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_has_compatible_shallow_copy_type.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/_has_compatible_shallow_copy_type_ops.h>
+
+namespace at {
+
+
+// aten::_has_compatible_shallow_copy_type(Tensor self, Tensor from) -> bool
+inline bool _has_compatible_shallow_copy_type(const at::Tensor & self, const at::Tensor & from) {
+    return at::_ops::_has_compatible_shallow_copy_type::call(self, from);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_bin_edges_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_bin_edges_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5f04a754f12861f348fc9e5cc793b84c279b2308
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_bin_edges_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API void _histogramdd_bin_edges_out(at::TensorList out, const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range=::std::nullopt, const ::std::optional<at::Tensor> & weight={}, bool density=false);
+TORCH_API void _histogramdd_bin_edges_outf(const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range, const ::std::optional<at::Tensor> & weight, bool density, at::TensorList out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_bin_edges_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_bin_edges_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b726388d5a2988b01a52fa29ca92d268c1cce086
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_bin_edges_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 _histogramdd_bin_edges_out(const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range, const ::std::optional<at::Tensor> & weight, bool density, at::TensorList out);
+TORCH_API ::std::vector<at::Tensor> histogramdd_bin_edges(const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range=::std::nullopt, const ::std::optional<at::Tensor> & weight={}, bool density=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_cts_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_cts_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..cf1ae63486ca148c67e232e9ddd8677db21fbf92
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_histogramdd_from_bin_cts_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 & _histogramdd_from_bin_cts_out(const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range, const ::std::optional<at::Tensor> & weight, bool density, at::Tensor & out);
+TORCH_API at::Tensor _histogramdd(const at::Tensor & self, at::IntArrayRef bins, ::std::optional<at::ArrayRef<double>> range=::std::nullopt, const ::std::optional<at::Tensor> & weight={}, bool density=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_index_put_impl_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_index_put_impl_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..0dd752d305148133de22da3d2bfa59e604476189
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_index_put_impl_native.h
@@ -0,0 +1,25 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor _index_put_impl(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false, bool unsafe=false);
+TORCH_API at::Tensor & _index_put_impl_out(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate, bool unsafe, at::Tensor & out);
+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);
+TORCH_API at::Tensor & _index_put_impl_quantized_cpu_(at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false, bool unsafe=false);
+TORCH_API at::Tensor & _index_put_impl_quantized_cuda_(at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false, bool unsafe=false);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_is_all_true_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_is_all_true_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..0a83b0adbea2f91b9b19019d07906401f07186a8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_is_all_true_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 _is_all_true(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_lazy_clone.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_lazy_clone.h
new file mode 100644
index 0000000000000000000000000000000000000000..8cdef261034419c06cb35c24a3890e2f34051fc3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_lazy_clone.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/_lazy_clone_ops.h>
+
+namespace at {
+
+
+// aten::_lazy_clone(Tensor self) -> Tensor
+inline at::Tensor _lazy_clone(const at::Tensor & self) {
+    return at::_ops::_lazy_clone::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_det_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_det_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..e4e27432fae033147ad24486861053566fb5ff18
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_det_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured__linalg_det : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & A);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..58eecfdce740d244719542a75cef18a4feec4827
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _log_softmax {
+  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::_log_softmax";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_log_softmax(Tensor self, int dim, bool half_to_float) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t dim, bool half_to_float);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, bool half_to_float);
+};
+
+struct TORCH_API _log_softmax_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::_log_softmax";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_log_softmax.out(Tensor self, int dim, bool half_to_float, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, int64_t dim, bool half_to_float, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, bool half_to_float, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_logcumsumexp_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_logcumsumexp_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0885d11999d5d3f2c5a627f6e8d82cc2ee9c2ae3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_logcumsumexp_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 _logcumsumexp(const at::Tensor & self, int64_t dim);
+TORCH_API at::Tensor & _logcumsumexp_out(at::Tensor & out, const at::Tensor & self, int64_t dim);
+TORCH_API at::Tensor & _logcumsumexp_outf(const at::Tensor & self, int64_t dim, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_lu_with_info_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_lu_with_info_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..fbae1e8da573dd2c0cca4fec597d5b8fe17a72db
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_lu_with_info_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _lu_with_info(const at::Tensor & self, bool pivot=true, bool check_errors=true);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_mixed_dtypes_linear.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_mixed_dtypes_linear.h
new file mode 100644
index 0000000000000000000000000000000000000000..326d2d9117f3f7a6fa46569028dfccb37df514be
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_mixed_dtypes_linear.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_mixed_dtypes_linear_ops.h>
+
+namespace at {
+
+
+// aten::_mixed_dtypes_linear(Tensor input, Tensor weight, Tensor scale, *, Tensor? bias=None, str? activation=None) -> Tensor
+inline at::Tensor _mixed_dtypes_linear(const at::Tensor & input, const at::Tensor & weight, const at::Tensor & scale, const ::std::optional<at::Tensor> & bias={}, ::std::optional<c10::string_view> activation=::std::nullopt) {
+    return at::_ops::_mixed_dtypes_linear::call(input, weight, scale, bias, activation);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..56d138ca96a6c97f5f19fd3cd391c87f1d23c26e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_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>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _native_batch_norm_legit_functional(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);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _batch_norm_legit_cpu_out(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, bool training, double momentum, double eps, at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _batch_norm_legit_cpu(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, bool training, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _batch_norm_legit_cuda_out(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, bool training, double momentum, double eps, at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _batch_norm_legit_cuda(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, bool training, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _mkldnn_batch_norm_legit(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, bool training, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _batch_norm_legit_no_stats_cpu(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, bool training, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _batch_norm_legit_no_stats_cpu_out(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);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _batch_norm_legit_no_stats_cuda(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, bool training, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _batch_norm_legit_no_stats_cuda_out(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);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _mkldnn_batch_norm_legit_no_stats(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, bool training, double momentum, double eps);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_no_training_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_no_training_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..702a2e19db668dab12494beb9120091f32e20856
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_no_training_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> _batch_norm_legit_no_training(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, double momentum, double eps);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _native_batch_norm_legit_no_training_out(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, double momentum, double eps, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_native_multi_head_attention.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_multi_head_attention.h
new file mode 100644
index 0000000000000000000000000000000000000000..c63c504d3678ce1163a8394f6ce8ba7bfa148c9c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_native_multi_head_attention.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/_native_multi_head_attention_ops.h>
+
+namespace at {
+
+
+// aten::_native_multi_head_attention(Tensor query, Tensor key, Tensor value, int embed_dim, int num_head, Tensor qkv_weight, Tensor qkv_bias, Tensor proj_weight, Tensor proj_bias, Tensor? mask=None, bool need_weights=True, bool average_attn_weights=True, int? mask_type=None) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _native_multi_head_attention(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, int64_t embed_dim, int64_t num_head, const at::Tensor & qkv_weight, const at::Tensor & qkv_bias, const at::Tensor & proj_weight, const at::Tensor & proj_bias, const ::std::optional<at::Tensor> & mask={}, bool need_weights=true, bool average_attn_weights=true, ::std::optional<int64_t> mask_type=::std::nullopt) {
+    return at::_ops::_native_multi_head_attention::call(query, key, value, embed_dim, num_head, qkv_weight, qkv_bias, proj_weight, proj_bias, mask, need_weights, average_attn_weights, mask_type);
+}
+
+// aten::_native_multi_head_attention.out(Tensor query, Tensor key, Tensor value, int embed_dim, int num_head, Tensor qkv_weight, Tensor qkv_bias, Tensor proj_weight, Tensor proj_bias, Tensor? mask=None, bool need_weights=True, bool average_attn_weights=True, int? mask_type=None, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _native_multi_head_attention_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, int64_t embed_dim, int64_t num_head, const at::Tensor & qkv_weight, const at::Tensor & qkv_bias, const at::Tensor & proj_weight, const at::Tensor & proj_bias, const ::std::optional<at::Tensor> & mask={}, bool need_weights=true, bool average_attn_weights=true, ::std::optional<int64_t> mask_type=::std::nullopt) {
+    return at::_ops::_native_multi_head_attention_out::call(query, key, value, embed_dim, num_head, qkv_weight, qkv_bias, proj_weight, proj_bias, mask, need_weights, average_attn_weights, mask_type, out0, out1);
+}
+// aten::_native_multi_head_attention.out(Tensor query, Tensor key, Tensor value, int embed_dim, int num_head, Tensor qkv_weight, Tensor qkv_bias, Tensor proj_weight, Tensor proj_bias, Tensor? mask=None, bool need_weights=True, bool average_attn_weights=True, int? mask_type=None, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _native_multi_head_attention_outf(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, int64_t embed_dim, int64_t num_head, const at::Tensor & qkv_weight, const at::Tensor & qkv_bias, const at::Tensor & proj_weight, const at::Tensor & proj_bias, const ::std::optional<at::Tensor> & mask, bool need_weights, bool average_attn_weights, ::std::optional<int64_t> mask_type, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::_native_multi_head_attention_out::call(query, key, value, embed_dim, num_head, qkv_weight, qkv_bias, proj_weight, proj_bias, mask, need_weights, average_attn_weights, mask_type, out0, out1);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_neg_view_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_neg_view_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..fb022d7baefc91a6aae086a4006f23b0eb7da934
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_neg_view_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/_neg_view_copy_ops.h>
+
+namespace at {
+
+
+// aten::_neg_view_copy(Tensor self) -> Tensor
+inline at::Tensor _neg_view_copy(const at::Tensor & self) {
+    return at::_ops::_neg_view_copy::call(self);
+}
+
+// aten::_neg_view_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _neg_view_copy_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::_neg_view_copy_out::call(self, out);
+}
+// aten::_neg_view_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _neg_view_copy_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::_neg_view_copy_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_from_padded_tensor_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_from_padded_tensor_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b9d2e5fd5be542077158871f57ae9c937e4f7895
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_from_padded_tensor_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_get_offsets_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_offsets_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b9d2e5fd5be542077158871f57ae9c937e4f7895
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_get_offsets_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_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_from_mask_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cc262cfd9ce0e0cfefcfb63388fe2e6b2dbeedc6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_from_mask_cpu_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor _nested_tensor_from_mask(const at::Tensor & t, const at::Tensor & mask, bool mask_check=true);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_storage_offsets_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_storage_offsets_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..072240d934037140834e96f56c596784950f4a6f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_storage_offsets_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & _nested_tensor_storage_offsets_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & _nested_tensor_storage_offsets_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_storage_offsets_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_storage_offsets_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d702cc8299be5f37e1d50dafb3232ea5b82170bc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_tensor_storage_offsets_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _nested_tensor_storage_offsets_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor _nested_tensor_storage_offsets(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_buffer.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_buffer.h
new file mode 100644
index 0000000000000000000000000000000000000000..0a8f3798d37b82cc8f27b9f403dd8c0c35b9b426
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_buffer.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_view_from_buffer_ops.h>
+
+namespace at {
+
+
+// aten::_nested_view_from_buffer(Tensor(a) self, Tensor nested_size, Tensor nested_strides, Tensor offsets) -> Tensor(a)
+inline at::Tensor _nested_view_from_buffer(const at::Tensor & self, const at::Tensor & nested_size, const at::Tensor & nested_strides, const at::Tensor & offsets) {
+    return at::_ops::_nested_view_from_buffer::call(self, nested_size, nested_strides, offsets);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_buffer_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_buffer_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..4b4b8d557b59e91c801950a17bf173dc17dae521
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_nested_view_from_buffer_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/_nested_view_from_buffer_copy_ops.h>
+
+namespace at {
+
+
+// aten::_nested_view_from_buffer_copy(Tensor self, Tensor nested_size, Tensor nested_strides, Tensor offsets) -> Tensor
+inline at::Tensor _nested_view_from_buffer_copy(const at::Tensor & self, const at::Tensor & nested_size, const at::Tensor & nested_strides, const at::Tensor & offsets) {
+    return at::_ops::_nested_view_from_buffer_copy::call(self, nested_size, nested_strides, offsets);
+}
+
+// aten::_nested_view_from_buffer_copy.out(Tensor self, Tensor nested_size, Tensor nested_strides, Tensor offsets, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _nested_view_from_buffer_copy_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & nested_size, const at::Tensor & nested_strides, const at::Tensor & offsets) {
+    return at::_ops::_nested_view_from_buffer_copy_out::call(self, nested_size, nested_strides, offsets, out);
+}
+// aten::_nested_view_from_buffer_copy.out(Tensor self, Tensor nested_size, Tensor nested_strides, Tensor offsets, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _nested_view_from_buffer_copy_outf(const at::Tensor & self, const at::Tensor & nested_size, const at::Tensor & nested_strides, const at::Tensor & offsets, at::Tensor & out) {
+    return at::_ops::_nested_view_from_buffer_copy_out::call(self, nested_size, nested_strides, offsets, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_pack_padded_sequence_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_pack_padded_sequence_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..da48665e2876fdfd1cf8f02874ea155767dbdd77
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_pack_padded_sequence_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> _pack_padded_sequence(const at::Tensor & input, const at::Tensor & lengths, bool batch_first);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> _pack_padded_sequence_out(const at::Tensor & input, const at::Tensor & lengths, bool batch_first, at::Tensor & out0, at::Tensor & out1);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_pad_enum_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_pad_enum_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1ee4ae98ab4320a490695ff631de18922926b984
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_pad_enum_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 _pad_enum(const at::Tensor & self, at::IntArrayRef pad, int64_t mode, ::std::optional<double> value=::std::nullopt);
+TORCH_API at::Tensor _pad_enum_symint(const at::Tensor & self, c10::SymIntArrayRef pad, int64_t mode, ::std::optional<double> value=::std::nullopt);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_pin_memory_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_pin_memory_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..6c93a650705b8d41fdf4de5eec349653d561e226
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_pin_memory_native.h
@@ -0,0 +1,25 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor _pin_memory(const at::Tensor & self, ::std::optional<at::Device> device=::std::nullopt);
+TORCH_API at::Tensor & _pin_memory_out(const at::Tensor & self, ::std::optional<at::Device> device, at::Tensor & out);
+TORCH_API at::Tensor _pin_memory_nested(const at::Tensor & self, ::std::optional<at::Device> device=::std::nullopt);
+TORCH_API at::Tensor _pin_memory_sparse_coo(const at::Tensor & self, ::std::optional<at::Device> device=::std::nullopt);
+TORCH_API at::Tensor _pin_memory_sparse_compressed(const at::Tensor & self, ::std::optional<at::Device> device=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_prelu_kernel_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_prelu_kernel_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..42fc04ecf6748ed42cf69d8dd4af39f6e1d8d8f6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_prelu_kernel_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 _prelu_kernel(const at::Tensor & self, const at::Tensor & weight);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_saturate_weight_to_fp16_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_saturate_weight_to_fp16_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..bb73d221b9bc1e1df83d5185070f68c3612e552c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_saturate_weight_to_fp16_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 _saturate_weight_to_fp16(const at::Tensor & weight);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_for_mps.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_for_mps.h
new file mode 100644
index 0000000000000000000000000000000000000000..496ba989396723865741c8150e6e8edcf926da36
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_for_mps.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_for_mps_ops.h>
+
+namespace at {
+
+
+// aten::_scaled_dot_product_attention_math_for_mps(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) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _scaled_dot_product_attention_math_for_mps(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) {
+    return at::_ops::_scaled_dot_product_attention_math_for_mps::call(query, key, value, attn_mask, dropout_p, is_causal, dropout_mask, scale);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..10c899858b407c7e62a280b8d43fbe6631e26dcc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_attention_math_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_attention_math {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, double, bool, const ::std::optional<at::Tensor> &, ::std::optional<double>, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_scaled_dot_product_attention_math";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_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)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_mask, double dropout_p, bool is_causal, const ::std::optional<at::Tensor> & dropout_mask, ::std::optional<double> scale, bool enable_gqa);
+  static ::std::tuple<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_mask, double dropout_p, bool is_causal, const ::std::optional<at::Tensor> & dropout_mask, ::std::optional<double> scale, bool enable_gqa);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_cudnn_attention.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_cudnn_attention.h
new file mode 100644
index 0000000000000000000000000000000000000000..86b4521f024ec1c68e56be2a09a03269c2a703e1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_scaled_dot_product_cudnn_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_cudnn_attention_ops.h>
+
+namespace at {
+
+
+// aten::_scaled_dot_product_cudnn_attention(Tensor query, Tensor key, Tensor value, Tensor? attn_bias, bool compute_log_sumexp, 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_cudnn_attention(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=0.0, bool is_causal=false, bool return_debug_mask=false, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_cudnn_attention::call(query, key, value, attn_bias, compute_log_sumexp, dropout_p, is_causal, return_debug_mask, scale);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_softmax_backward_data_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_softmax_backward_data_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..47e97625d5bdc0c18a25a7721e517f2839f5aca5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_softmax_backward_data_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 _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 meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_softmax_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_softmax_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cc8b4cc24ed9da03fd11fd708a7bf14625044649
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_softmax_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 _softmax(const at::Tensor & self, int64_t dim, bool half_to_float);
+TORCH_API at::Tensor & _softmax_out(at::Tensor & out, const at::Tensor & self, int64_t dim, bool half_to_float);
+TORCH_API at::Tensor & _softmax_outf(const at::Tensor & self, int64_t dim, bool half_to_float, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_broadcast_to_copy.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_broadcast_to_copy.h
new file mode 100644
index 0000000000000000000000000000000000000000..01f6e748faacf9b7aeb2801198b98b00adc684ac
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_broadcast_to_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/_sparse_broadcast_to_copy_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_broadcast_to_copy(Tensor self, int[] size) -> Tensor
+inline at::Tensor _sparse_broadcast_to_copy(const at::Tensor & self, at::IntArrayRef size) {
+    return at::_ops::_sparse_broadcast_to_copy::call(self, size);
+}
+
+// aten::_sparse_broadcast_to_copy.out(Tensor self, int[] size, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_broadcast_to_copy_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef size) {
+    return at::_ops::_sparse_broadcast_to_copy_out::call(self, size, out);
+}
+// aten::_sparse_broadcast_to_copy.out(Tensor self, int[] size, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_broadcast_to_copy_outf(const at::Tensor & self, at::IntArrayRef size, at::Tensor & out) {
+    return at::_ops::_sparse_broadcast_to_copy_out::call(self, size, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_bsc_tensor_unsafe_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_bsc_tensor_unsafe_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..55dd0212c831702597871ab736fd95919f42fadc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_bsc_tensor_unsafe_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_bsc_tensor_unsafe(const at::Tensor & ccol_indices, const at::Tensor & row_indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={});
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_compressed_tensor_unsafe_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_compressed_tensor_unsafe_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..3df1f425d667fc8d8540f8c43db6558b1c58cc3e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_compressed_tensor_unsafe_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_compressed_tensor_unsafe {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, c10::SymIntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sparse_compressed_tensor_unsafe";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_sparse_compressed_tensor_unsafe(Tensor compressed_indices, Tensor plain_indices, Tensor values, SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_csr_prod_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_csr_prod_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9377c895a352ca1de40a2c627eb09f91c01fff80
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_csr_prod_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & _sparse_csr_prod_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & _sparse_csr_prod_outf(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mask_projection.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mask_projection.h
new file mode 100644
index 0000000000000000000000000000000000000000..1adfcea9f67281e76f4d259bcbf6e187314a17f6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mask_projection.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_mask_projection_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_mask_projection.out(Tensor self, Tensor mask, bool accumulate_matches=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_mask_projection_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & mask, bool accumulate_matches=false) {
+    return at::_ops::_sparse_mask_projection_out::call(self, mask, accumulate_matches, out);
+}
+// aten::_sparse_mask_projection.out(Tensor self, Tensor mask, bool accumulate_matches=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_mask_projection_outf(const at::Tensor & self, const at::Tensor & mask, bool accumulate_matches, at::Tensor & out) {
+    return at::_ops::_sparse_mask_projection_out::call(self, mask, accumulate_matches, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm_reduce_impl.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm_reduce_impl.h
new file mode 100644
index 0000000000000000000000000000000000000000..419efebaa3c0fd3e3c7822eb899f348c898937ed
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_mm_reduce_impl.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_sparse_mm_reduce_impl_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_mm_reduce_impl(Tensor self, Tensor other, str reduce) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _sparse_mm_reduce_impl(const at::Tensor & self, const at::Tensor & other, c10::string_view reduce) {
+    return at::_ops::_sparse_mm_reduce_impl::call(self, other, reduce);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_apply_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_apply_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..79125bc916f855ed865bfc395f5dcbcac50cd338
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_semi_structured_apply_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> _sparse_semi_structured_apply(const at::Tensor & input, const at::Tensor & thread_masks);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_softmax_backward_data_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_softmax_backward_data_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8bae65384b773b1a6e5024971ceabbe6c762067b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_sparse_softmax_backward_data_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_softmax_backward_data_out(at::Tensor & out, const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, const at::Tensor & self);
+TORCH_API at::Tensor & _sparse_softmax_backward_data_outf(const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, const at::Tensor & self, at::Tensor & out);
+
+} // 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..01ff5270214ac53b2eb8a9aa5c690c9240198144
--- /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/_stack_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_stack_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..98e3e4f963772389c2d0837c953540994d6c0f03
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_stack_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 _stack(at::TensorList tensors, int64_t dim=0);
+TORCH_API at::Tensor & _stack_out(at::Tensor & out, at::TensorList tensors, int64_t dim=0);
+TORCH_API at::Tensor & _stack_outf(at::TensorList tensors, int64_t dim, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..53fcd8583001c66b7a62685bdb9bab763f80d021
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_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/_thnn_fused_lstm_cell_backward_ops.h>
+
+namespace at {
+
+
+// aten::_thnn_fused_lstm_cell_backward(Tensor? grad_hy, Tensor? grad_cy, Tensor cx, Tensor cy, Tensor workspace, bool has_bias) -> (Tensor, Tensor, Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _thnn_fused_lstm_cell_backward(const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, const at::Tensor & cx, const at::Tensor & cy, const at::Tensor & workspace, bool has_bias) {
+    return at::_ops::_thnn_fused_lstm_cell_backward::call(grad_hy, grad_cy, cx, cy, workspace, has_bias);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_impl_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_impl_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..acf1f4f4a85f3d2fe0743407032698649234cfb1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_impl_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _thnn_fused_lstm_cell_backward_impl(const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, const at::Tensor & cx, const at::Tensor & cy, const at::Tensor & workspace, bool has_bias);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..af710c4a5f0cebb2e4d93358970d7f737f2d277e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_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_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & input_gates, const at::Tensor & hidden_gates, const at::Tensor & cx, const ::std::optional<at::Tensor> & input_bias={}, const ::std::optional<at::Tensor> & hidden_bias={});
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _thnn_fused_lstm_cell_outf(const at::Tensor & input_gates, const at::Tensor & hidden_gates, const at::Tensor & cx, const ::std::optional<at::Tensor> & input_bias, const ::std::optional<at::Tensor> & hidden_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_cpu.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_cpu.h
new file mode 100644
index 0000000000000000000000000000000000000000..9ad567d77f11f86fdf307c33e9ad0e7487cacc31
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_cpu.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_to_cpu_ops.h>
+
+namespace at {
+
+
+// aten::_to_cpu(Tensor[] tensors) -> Tensor[]
+inline ::std::vector<at::Tensor> _to_cpu(at::TensorList tensors) {
+    return at::_ops::_to_cpu::call(tensors);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse.h
new file mode 100644
index 0000000000000000000000000000000000000000..13b6b25f5636fa4e964a1190135c4895e8d9fcaf
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse.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/_to_sparse_ops.h>
+
+namespace at {
+
+
+// aten::_to_sparse.sparse_dim_out(Tensor self, int sparse_dim, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _to_sparse_out(at::Tensor & out, const at::Tensor & self, int64_t sparse_dim) {
+    return at::_ops::_to_sparse_sparse_dim_out::call(self, sparse_dim, out);
+}
+// aten::_to_sparse.sparse_dim_out(Tensor self, int sparse_dim, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _to_sparse_outf(const at::Tensor & self, int64_t sparse_dim, at::Tensor & out) {
+    return at::_ops::_to_sparse_sparse_dim_out::call(self, sparse_dim, out);
+}
+
+// aten::_to_sparse.out(Tensor self, *, Layout? layout=None, int[2]? blocksize=None, int? dense_dim=None, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::_to_sparse_out::call(self, layout, blocksize, dense_dim, out);
+}
+// aten::_to_sparse.out(Tensor self, *, Layout? layout=None, int[2]? blocksize=None, int? dense_dim=None, Tensor(a!) out) -> Tensor(a!)
+inline 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) {
+    return at::_ops::_to_sparse_out::call(self, layout, blocksize, dense_dim, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_csr_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_csr_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..11d25a7a1364cb3045282e3a6bf3564d93a2b760
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_to_sparse_csr_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 _to_sparse_csr(const at::Tensor & self, ::std::optional<int64_t> dense_dim=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_trilinear_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_trilinear_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2ea991c7452b1511f08f0c2d44cac3c286050108
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_trilinear_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 _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);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_unique2_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_unique2_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..29dae2c5c84013fcf980c702a8d2b9d8bf96bae6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_unique2_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> _unique2(const at::Tensor & self, 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/_upsample_bicubic2d_aa_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bicubic2d_aa_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..91c83aa910403338f942699402e0cf62979f6cea
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_bicubic2d_aa_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_bicubic2d_aa_backward_ops.h>
+
+namespace at {
+
+
+// aten::_upsample_bicubic2d_aa_backward.grad_input(Tensor grad_output, SymInt[2] output_size, SymInt[4] input_size, bool align_corners, float? scales_h=None, float? scales_w=None, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & _upsample_bicubic2d_aa_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_bicubic2d_aa_backward_grad_input::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), align_corners, 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_bicubic2d_aa_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_bicubic2d_aa_backward_grad_input::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), align_corners, scales_h, scales_w, grad_input);
+  }
+}
+
+// aten::_upsample_bicubic2d_aa_backward.grad_input(Tensor grad_output, SymInt[2] output_size, SymInt[4] input_size, bool align_corners, float? scales_h=None, float? scales_w=None, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & _upsample_bicubic2d_aa_backward_outf(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input) {
+    return at::_ops::_upsample_bicubic2d_aa_backward_grad_input::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), align_corners, 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_bicubic2d_aa_backward_outf(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input) {
+    return at::_ops::_upsample_bicubic2d_aa_backward_grad_input::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), align_corners, scales_h, scales_w, grad_input);
+  }
+}
+
+// aten::_upsample_bicubic2d_aa_backward.grad_input(Tensor grad_output, SymInt[2] output_size, SymInt[4] input_size, bool align_corners, float? scales_h=None, float? scales_w=None, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & _upsample_bicubic2d_aa_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_bicubic2d_aa_backward_grad_input::call(grad_output, output_size, input_size, align_corners, 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_bicubic2d_aa_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_bicubic2d_aa_backward_grad_input::call(grad_output, output_size, input_size, align_corners, scales_h, scales_w, grad_input);
+  }
+}
+
+// aten::_upsample_bicubic2d_aa_backward.grad_input(Tensor grad_output, SymInt[2] output_size, SymInt[4] input_size, bool align_corners, float? scales_h=None, float? scales_w=None, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & _upsample_bicubic2d_aa_backward_symint_outf(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input) {
+    return at::_ops::_upsample_bicubic2d_aa_backward_grad_input::call(grad_output, output_size, input_size, align_corners, 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_bicubic2d_aa_backward_outf(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input) {
+    return at::_ops::_upsample_bicubic2d_aa_backward_grad_input::call(grad_output, output_size, input_size, align_corners, scales_h, scales_w, grad_input);
+  }
+}
+
+// aten::_upsample_bicubic2d_aa_backward(Tensor grad_output, SymInt[2] output_size, SymInt[4] input_size, bool align_corners, float? scales_h=None, float? scales_w=None) -> Tensor
+inline at::Tensor _upsample_bicubic2d_aa_backward(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_bicubic2d_aa_backward::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), align_corners, scales_h, scales_w);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor _upsample_bicubic2d_aa_backward(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_bicubic2d_aa_backward::call(grad_output, c10::fromIntArrayRefSlow(output_size), c10::fromIntArrayRefSlow(input_size), align_corners, scales_h, scales_w);
+  }
+}
+
+// aten::_upsample_bicubic2d_aa_backward(Tensor grad_output, SymInt[2] output_size, SymInt[4] input_size, bool align_corners, float? scales_h=None, float? scales_w=None) -> Tensor
+inline at::Tensor _upsample_bicubic2d_aa_backward_symint(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_bicubic2d_aa_backward::call(grad_output, output_size, input_size, align_corners, scales_h, scales_w);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor _upsample_bicubic2d_aa_backward(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::_upsample_bicubic2d_aa_backward::call(grad_output, output_size, input_size, align_corners, scales_h, scales_w);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..433d069ede44ebd4964c7379c08cf5e4d27955b3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor _upsample_nearest_exact1d(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales=::std::nullopt);
+TORCH_API at::Tensor _upsample_nearest_exact1d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales=::std::nullopt);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_use_cudnn_rnn_flatten_weight.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_use_cudnn_rnn_flatten_weight.h
new file mode 100644
index 0000000000000000000000000000000000000000..a51b2bfa2c4003b2629f6b4fe73ebcd3205e7068
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_use_cudnn_rnn_flatten_weight.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/_use_cudnn_rnn_flatten_weight_ops.h>
+
+namespace at {
+
+
+// aten::_use_cudnn_rnn_flatten_weight() -> bool
+inline bool _use_cudnn_rnn_flatten_weight() {
+    return at::_ops::_use_cudnn_rnn_flatten_weight::call();
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_use_cudnn_rnn_flatten_weight_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_use_cudnn_rnn_flatten_weight_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..417426f595f74a4ca28d0bf553f8cdc10501b8b9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_use_cudnn_rnn_flatten_weight_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 _use_cudnn_rnn_flatten_weight();
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_int4pack_mm_for_cpu.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_int4pack_mm_for_cpu.h
new file mode 100644
index 0000000000000000000000000000000000000000..bec3617ede17df604a18b94caf162b5a00f5a412
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_int4pack_mm_for_cpu.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/_weight_int4pack_mm_for_cpu_ops.h>
+
+namespace at {
+
+
+// aten::_weight_int4pack_mm_for_cpu(Tensor self, Tensor mat2, int qGroupSize, Tensor qScaleAndZeros) -> Tensor
+inline at::Tensor _weight_int4pack_mm_for_cpu(const at::Tensor & self, const at::Tensor & mat2, int64_t qGroupSize, const at::Tensor & qScaleAndZeros) {
+    return at::_ops::_weight_int4pack_mm_for_cpu::call(self, mat2, qGroupSize, qScaleAndZeros);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..e48cf1c7cf19af376626a2a2f8fc145be520446d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/_weight_norm_interface_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/_weight_norm_interface_backward_ops.h>
+
+namespace at {
+
+
+// aten::_weight_norm_interface_backward(Tensor grad_w, Tensor saved_v, Tensor saved_g, Tensor saved_norms, int dim) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _weight_norm_interface_backward(const at::Tensor & grad_w, const at::Tensor & saved_v, const at::Tensor & saved_g, const at::Tensor & saved_norms, int64_t dim) {
+    return at::_ops::_weight_norm_interface_backward::call(grad_w, saved_v, saved_g, saved_norms, dim);
+}
+
+// aten::_weight_norm_interface_backward.out(Tensor grad_w, Tensor saved_v, Tensor saved_g, Tensor saved_norms, int dim, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::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) {
+    return at::_ops::_weight_norm_interface_backward_out::call(grad_w, saved_v, saved_g, saved_norms, dim, out0, out1);
+}
+// aten::_weight_norm_interface_backward.out(Tensor grad_w, Tensor saved_v, Tensor saved_g, Tensor saved_norms, int dim, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::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) {
+    return at::_ops::_weight_norm_interface_backward_out::call(grad_w, saved_v, saved_g, saved_norms, dim, out0, out1);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/absolute_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/absolute_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5f249d8e78eac10bcae2404504d16a6c4ddd6fc1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/absolute_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 absolute(const at::Tensor & self);
+TORCH_API at::Tensor & absolute_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & absolute_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/acos_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/acos_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..3908106f92e1d43247a2fcfa33d639f0cda63667
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/acos_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_acos : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool2d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool2d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1c5c8fc6993a13c4a1f10f5f3a73dffcfacec87d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_avg_pool2d_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API adaptive_avg_pool2d_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::adaptive_avg_pool2d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "adaptive_avg_pool2d.out(Tensor self, SymInt[2] output_size, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, c10::SymIntArrayRef output_size, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef output_size, at::Tensor & out);
+};
+
+struct TORCH_API adaptive_avg_pool2d {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::adaptive_avg_pool2d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "adaptive_avg_pool2d(Tensor self, SymInt[2] output_size) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef output_size);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef output_size);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8bc988555d4df387f3c54e96f577e88c87f38ac0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor adaptive_max_pool2d_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & indices);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d2464ca09e1416b25cd3beec54c4533c7d3451d5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_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 adaptive_max_pool2d_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & indices);
+TORCH_API at::Tensor & adaptive_max_pool2d_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_pool2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & indices, at::Tensor & grad_input);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..f8b35d396f0e38c90dae0866b6db137c7676834f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_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_pool2d_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/adaptive_max_pool3d_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4ca942ce0a005a46c4c49f7e3b793faaacc7b660
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_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> adaptive_max_pool3d(const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> adaptive_max_pool3d_out(at::Tensor & out, at::Tensor & indices, const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> adaptive_max_pool3d_outf(const at::Tensor & self, at::IntArrayRef output_size, at::Tensor & out, at::Tensor & indices);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..39c6f984d3eba1778f46b1366f84741963cd6a0d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/adaptive_max_pool3d_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> adaptive_max_pool3d(const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> adaptive_max_pool3d_out(at::Tensor & out, at::Tensor & indices, const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> adaptive_max_pool3d_outf(const at::Tensor & self, at::IntArrayRef output_size, at::Tensor & out, at::Tensor & indices);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..29d73942a462dc5fff9c7793e87231f0b81742e2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul_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 addcmul(const at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value=1);
+TORCH_API at::Tensor & addcmul_(at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value=1);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d1b3c8e5b72945c9ed2e4f4fd10155f81dec7e82
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addcmul_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 addcmul(const at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value=1);
+TORCH_API at::Tensor & addcmul_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value=1);
+TORCH_API at::Tensor & addcmul_outf(const at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value, at::Tensor & out);
+TORCH_API at::Tensor & addcmul_(at::Tensor & self, const at::Tensor & tensor1, const at::Tensor & tensor2, const at::Scalar & value=1);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addmm_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addmm_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..278a318677bf49f8a7a2a20ff58a1ab9b49d234b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addmm_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 addmm(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & addmm_(at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/addmm_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/addmm_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..79d1b5ba561d0c05aaacf174095fa8c62e188429
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/addmm_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 addmm(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & addmm_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & addmm_outf(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta, const at::Scalar & alpha, at::Tensor & out);
+TORCH_API at::Tensor & addmm_(at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/affine_grid_generator_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/affine_grid_generator_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..948c216d9d17de08e3e15660ae9357ef09e52410
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/affine_grid_generator_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 affine_grid_generator {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::affine_grid_generator";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "affine_grid_generator(Tensor theta, SymInt[] size, bool align_corners) -> Tensor";
+  static at::Tensor call(const at::Tensor & theta, c10::SymIntArrayRef size, bool align_corners);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & theta, c10::SymIntArrayRef size, bool align_corners);
+};
+
+struct TORCH_API affine_grid_generator_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::affine_grid_generator";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "affine_grid_generator.out(Tensor theta, SymInt[] size, bool align_corners, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & theta, c10::SymIntArrayRef size, bool align_corners, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & theta, c10::SymIntArrayRef size, bool align_corners, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/alpha_dropout_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/alpha_dropout_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..30375324fb7ec34d50b48d43b607261927a584cd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/alpha_dropout_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 alpha_dropout(const at::Tensor & input, double p, bool train);
+TORCH_API at::Tensor & alpha_dropout_(at::Tensor & self, double p, bool train);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/amax_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/amax_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d53b7d4c0702cc98054fa2b994c887571a2d9e44
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/amax_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/amax_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_amax_out : public at::meta::structured_amax {
+void impl(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/angle_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/angle_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..675ed3503cb8b85d87939a4cee7e448f766ab08e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/angle_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 angle(const at::Tensor & self);
+TORCH_API at::Tensor & angle_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & angle_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arccos_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arccos_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4782be8fd671ecf48e9409bf7865f67ad1110837
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arccos_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 arccos(const at::Tensor & self);
+TORCH_API at::Tensor & arccos_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & arccos_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/arccosh.h b/lib/python3.12/site-packages/torch/include/ATen/ops/arccosh.h
new file mode 100644
index 0000000000000000000000000000000000000000..4b93a251b0ac711f4cdfb4fe013d400029d1d1fb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/arccosh.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/arccosh_ops.h>
+
+namespace at {
+
+
+// aten::arccosh(Tensor self) -> Tensor
+inline at::Tensor arccosh(const at::Tensor & self) {
+    return at::_ops::arccosh::call(self);
+}
+
+// aten::arccosh_(Tensor(a!) self) -> Tensor(a!)
+inline at::Tensor & arccosh_(at::Tensor & self) {
+    return at::_ops::arccosh_::call(self);
+}
+
+// aten::arccosh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arccosh_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::arccosh_out::call(self, out);
+}
+// aten::arccosh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & arccosh_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::arccosh_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/asin_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/asin_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..1a20c148477977a5f2f9d3bf12eba07535094063
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/asin_native.h
@@ -0,0 +1,29 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/asin_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_asin_out : public at::meta::structured_asin {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+TORCH_API at::Tensor asin_sparse(const at::Tensor & self);
+TORCH_API at::Tensor & asin_sparse_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & asin_sparse_(at::Tensor & self);
+TORCH_API at::Tensor asin_sparse_csr(const at::Tensor & self);
+TORCH_API at::Tensor & asin_sparse_csr_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & asin_sparse_csr_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/asin_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/asin_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..abff9ced9e08e04694944473e16fa5653929c193
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/asin_ops.h
@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API asin {
+  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::asin";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "asin(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 asin_ {
+  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::asin_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "asin_(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 asin_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::asin";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "asin.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/atanh_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/atanh_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f5285c5980609026c5fd6f9dc861b418c4f12069
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/atanh_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 atanh(const at::Tensor & self);
+TORCH_API at::Tensor & atanh_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & atanh_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & atanh_(at::Tensor & self);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/atanh_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/atanh_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5d2e101a30d901bd6772ccf8069f7b1513b49294
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/atanh_native.h
@@ -0,0 +1,29 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/atanh_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_atanh_out : public at::meta::structured_atanh {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+TORCH_API at::Tensor atanh_sparse(const at::Tensor & self);
+TORCH_API at::Tensor & atanh_sparse_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & atanh_sparse_(at::Tensor & self);
+TORCH_API at::Tensor atanh_sparse_csr(const at::Tensor & self);
+TORCH_API at::Tensor & atanh_sparse_csr_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & atanh_sparse_csr_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool2d_backward_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool2d_backward_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5b3b7a0446858d8d0c3fb1bcb9469c459960742b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool2d_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 avg_pool2d_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override);
+TORCH_API at::Tensor & avg_pool2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override);
+TORCH_API at::Tensor & avg_pool2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override, at::Tensor & grad_input);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool2d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool2d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..e35e0061dbdf77a821368a0ea79f1f20511755f6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/avg_pool2d_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API avg_pool2d_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_pool2d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "avg_pool2d.out(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] 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_pool2d {
+  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_pool2d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "avg_pool2d(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] 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/baddbmm_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/baddbmm_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..29a6ed6633a4080a4a98de3db64d4291b37fcff0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/baddbmm_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 baddbmm(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 & baddbmm_(at::Tensor & self, const at::Tensor & batch1, const at::Tensor & batch2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_backward_reduce_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_backward_reduce_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..84a503e64f15de5966b848898c0f2b77f36f01eb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_backward_reduce_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 &> batch_norm_backward_reduce_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, const at::Tensor & grad_out, const at::Tensor & input, const at::Tensor & mean, const at::Tensor & invstd, const ::std::optional<at::Tensor> & weight, bool input_g, bool weight_g, bool bias_g);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> batch_norm_backward_reduce_outf(const at::Tensor & grad_out, const at::Tensor & input, const at::Tensor & mean, const at::Tensor & invstd, const ::std::optional<at::Tensor> & weight, bool input_g, bool weight_g, bool bias_g, 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/batch_norm_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_norm_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..d626de65daf7df984f7503cffeae3b7ccf258340
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/batch_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 batch_norm {
+  using schema = at::Tensor (const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, bool, double, double, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::batch_norm";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "batch_norm(Tensor input, Tensor? weight, Tensor? bias, Tensor? running_mean, Tensor? running_var, bool training, float momentum, float eps, bool cudnn_enabled) -> Tensor";
+  static at::Tensor call(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);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, bool training, double momentum, double eps, bool cudnn_enabled);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bernoulli_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bernoulli_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fefa46d4cde12302331acb659d0f4de6de7614fe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bernoulli_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 & bernoulli_(at::Tensor & self, const at::Tensor & p, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor & bernoulli_(at::Tensor & self, double p=0.5, ::std::optional<at::Generator> generator=::std::nullopt);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/binary_cross_entropy_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/binary_cross_entropy_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..873a36a9a47d92ed82ff8ab54abf132bdf7315a6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/binary_cross_entropy_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 binary_cross_entropy(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean);
+TORCH_API at::Tensor & binary_cross_entropy_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean);
+TORCH_API at::Tensor & binary_cross_entropy_outf(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_and_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_and_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1489472c1ada3d05bcb11c06ffb562551bd9b4f4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_and_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_and(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_and_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_and_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & bitwise_and_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e4f494350404ea095dc8db92fd8245357faebc56
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_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 bitwise_right_shift(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_right_shift_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_right_shift_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & bitwise_right_shift_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7b88ef6ab663c0b9525ce6db84f95609bbcb587f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bitwise_right_shift_meta_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor bitwise_right_shift(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_right_shift_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & bitwise_right_shift_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & bitwise_right_shift_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/blackman_window_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/blackman_window_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..3bd2650bfc5e1184d5342cf013c702cdfd8ee7bd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/blackman_window_ops.h
@@ -0,0 +1,61 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API blackman_window {
+  using schema = at::Tensor (int64_t, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::blackman_window";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "blackman_window(int window_length, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(int64_t window_length, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API blackman_window_periodic {
+  using schema = at::Tensor (int64_t, bool, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::blackman_window";
+  static constexpr const char* overload_name = "periodic";
+  static constexpr const char* schema_str = "blackman_window.periodic(int window_length, bool periodic, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(int64_t window_length, bool periodic, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, bool periodic, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API blackman_window_out {
+  using schema = at::Tensor & (int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::blackman_window";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "blackman_window.out(int window_length, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t window_length, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, at::Tensor & out);
+};
+
+struct TORCH_API blackman_window_periodic_out {
+  using schema = at::Tensor & (int64_t, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::blackman_window";
+  static constexpr const char* overload_name = "periodic_out";
+  static constexpr const char* schema_str = "blackman_window.periodic_out(int window_length, bool periodic, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t window_length, bool periodic, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, bool periodic, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/bmm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/bmm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..1d466a83338e608de6cff591701797c847084249
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/bmm_native.h
@@ -0,0 +1,33 @@
+#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/bmm_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_bmm_out_cpu : public at::meta::structured_bmm {
+void impl(const at::Tensor & self, const at::Tensor & mat2, const at::Tensor & out);
+};
+struct TORCH_API structured_bmm_out_cuda : public at::meta::structured_bmm {
+void impl(const at::Tensor & self, const at::Tensor & mat2, const at::Tensor & out);
+};
+TORCH_API at::Tensor bmm_nested(const at::Tensor & self, const at::Tensor & mat2);
+TORCH_API at::Tensor bmm_nested_cuda(const at::Tensor & self, const at::Tensor & mat2);
+TORCH_API at::Tensor bmm_sparse_cpu(const at::Tensor & self, const at::Tensor & mat2);
+TORCH_API at::Tensor & bmm_out_sparse_cpu(const at::Tensor & self, const at::Tensor & mat2, at::Tensor & out);
+TORCH_API at::Tensor bmm_sparse_cuda(const at::Tensor & self, const at::Tensor & mat2);
+TORCH_API at::Tensor & bmm_out_sparse_cuda(const at::Tensor & self, const at::Tensor & mat2, at::Tensor & out);
+TORCH_API at::Tensor & bmm_out_sparse_csr_cuda(const at::Tensor & self, const at::Tensor & mat2, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/can_cast.h b/lib/python3.12/site-packages/torch/include/ATen/ops/can_cast.h
new file mode 100644
index 0000000000000000000000000000000000000000..e387e4e0ceabacb131ca5134a2cbb645da7cc191
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/can_cast.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/can_cast_ops.h>
+
+namespace at {
+
+
+// aten::can_cast(ScalarType from_, ScalarType to) -> bool
+inline bool can_cast(at::ScalarType from_, at::ScalarType to) {
+    return at::_ops::can_cast::call(from_, to);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cartesian_prod.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cartesian_prod.h
new file mode 100644
index 0000000000000000000000000000000000000000..fed37078768b65c93b0089cfd18a2a251659f0d9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cartesian_prod.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/cartesian_prod_ops.h>
+
+namespace at {
+
+
+// aten::cartesian_prod(Tensor[] tensors) -> Tensor
+inline at::Tensor cartesian_prod(at::TensorList tensors) {
+    return at::_ops::cartesian_prod::call(tensors);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cauchy_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cauchy_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6391439f4b4568edcdc8164a2d5e02e303c4c70b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cauchy_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 cauchy(const at::Tensor & self, double median=0, double sigma=1, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor & cauchy_out(at::Tensor & out, const at::Tensor & self, double median=0, double sigma=1, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor & cauchy_outf(const at::Tensor & self, double median, double sigma, ::std::optional<at::Generator> generator, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/channel_shuffle_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/channel_shuffle_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..0b11e5ed7486c965b1f131222879aa42be764d0e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/channel_shuffle_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 & channel_shuffle_out_symint(const at::Tensor & self, c10::SymInt groups, at::Tensor & out);
+TORCH_API at::Tensor channel_shuffle(const at::Tensor & self, int64_t groups);
+TORCH_API at::Tensor channel_shuffle_quantized_cpu(const at::Tensor & self, int64_t groups);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/chunk.h b/lib/python3.12/site-packages/torch/include/ATen/ops/chunk.h
new file mode 100644
index 0000000000000000000000000000000000000000..f6ebb47904da0e0650378cca19ea01689bc3673f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/chunk.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/chunk_ops.h>
+
+namespace at {
+
+
+// aten::chunk(Tensor(a -> *) self, int chunks, int dim=0) -> Tensor(a)[]
+inline ::std::vector<at::Tensor> chunk(const at::Tensor & self, int64_t chunks, int64_t dim=0) {
+    return at::_ops::chunk::call(self, chunks, dim);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/clip.h b/lib/python3.12/site-packages/torch/include/ATen/ops/clip.h
new file mode 100644
index 0000000000000000000000000000000000000000..b143fbd554799f869f874b13bea40bafdff7da53
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/clip.h
@@ -0,0 +1,63 @@
+#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/clip_ops.h>
+
+namespace at {
+
+
+// aten::clip(Tensor self, Scalar? min=None, Scalar? max=None) -> Tensor
+inline at::Tensor clip(const at::Tensor & self, const ::std::optional<at::Scalar> & min, const ::std::optional<at::Scalar> & max=::std::nullopt) {
+    return at::_ops::clip::call(self, min, max);
+}
+
+// aten::clip.Tensor(Tensor self, Tensor? min=None, Tensor? max=None) -> Tensor
+inline at::Tensor clip(const at::Tensor & self, const ::std::optional<at::Tensor> & min={}, const ::std::optional<at::Tensor> & max={}) {
+    return at::_ops::clip_Tensor::call(self, min, max);
+}
+
+// aten::clip_(Tensor(a!) self, Scalar? min=None, Scalar? max=None) -> Tensor(a!)
+inline at::Tensor & clip_(at::Tensor & self, const ::std::optional<at::Scalar> & min, const ::std::optional<at::Scalar> & max=::std::nullopt) {
+    return at::_ops::clip_::call(self, min, max);
+}
+
+// aten::clip_.Tensor(Tensor(a!) self, Tensor? min=None, Tensor? max=None) -> Tensor(a!)
+inline at::Tensor & clip_(at::Tensor & self, const ::std::optional<at::Tensor> & min={}, const ::std::optional<at::Tensor> & max={}) {
+    return at::_ops::clip__Tensor::call(self, min, max);
+}
+
+// aten::clip.out(Tensor self, Scalar? min=None, Scalar? max=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & clip_out(at::Tensor & out, const at::Tensor & self, const ::std::optional<at::Scalar> & min, const ::std::optional<at::Scalar> & max=::std::nullopt) {
+    return at::_ops::clip_out::call(self, min, max, out);
+}
+// aten::clip.out(Tensor self, Scalar? min=None, Scalar? max=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & clip_outf(const at::Tensor & self, const ::std::optional<at::Scalar> & min, const ::std::optional<at::Scalar> & max, at::Tensor & out) {
+    return at::_ops::clip_out::call(self, min, max, out);
+}
+
+// aten::clip.Tensor_out(Tensor self, Tensor? min=None, Tensor? max=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & clip_out(at::Tensor & out, const at::Tensor & self, const ::std::optional<at::Tensor> & min={}, const ::std::optional<at::Tensor> & max={}) {
+    return at::_ops::clip_Tensor_out::call(self, min, max, out);
+}
+// aten::clip.Tensor_out(Tensor self, Tensor? min=None, Tensor? max=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & clip_outf(const at::Tensor & self, const ::std::optional<at::Tensor> & min, const ::std::optional<at::Tensor> & max, at::Tensor & out) {
+    return at::_ops::clip_Tensor_out::call(self, min, max, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/coalesce_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/coalesce_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..229258526a4938da6830a1ddc260afe7007cb645
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/coalesce_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 coalesce {
+  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::coalesce";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "coalesce(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/col2im_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/col2im_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..bfc57361725f0270bcd6bcb1da8f79bc5de28280
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/col2im_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 col2im(const at::Tensor & self, at::IntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride);
+TORCH_API at::Tensor col2im_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride);
+TORCH_API at::Tensor & col2im_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride);
+TORCH_API at::Tensor & col2im_outf(const at::Tensor & self, at::IntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride, at::Tensor & out);
+TORCH_API at::Tensor & col2im_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride);
+TORCH_API at::Tensor & col2im_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d35d95514f4cc8e63e335dd84e1fed74d35e8d72
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/col_indices_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor col_indices_default(const at::Tensor & self);
+TORCH_API at::Tensor col_indices_sparse_csr(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/combinations.h b/lib/python3.12/site-packages/torch/include/ATen/ops/combinations.h
new file mode 100644
index 0000000000000000000000000000000000000000..b8f069a588d37f9107fac846ff36bbcb86f0cc0f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/combinations.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/combinations_ops.h>
+
+namespace at {
+
+
+// aten::combinations(Tensor self, int r=2, bool with_replacement=False) -> Tensor
+inline at::Tensor combinations(const at::Tensor & self, int64_t r=2, bool with_replacement=false) {
+    return at::_ops::combinations::call(self, r, with_replacement);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/copy_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/copy_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..942d1b8c262910c30e73d12ae1a7ba75f339fefa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/copy_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor copy(const at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/corrcoef_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/corrcoef_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..20b6127377beb61089a37539b23e1644903ff85f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/corrcoef_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 corrcoef(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/count_nonzero.h b/lib/python3.12/site-packages/torch/include/ATen/ops/count_nonzero.h
new file mode 100644
index 0000000000000000000000000000000000000000..1ea4e8c657591a8b9b473a68a91b7bc96f002d15
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/count_nonzero.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/count_nonzero_ops.h>
+
+namespace at {
+
+
+// aten::count_nonzero.dim_IntList(Tensor self, int[] dim) -> Tensor
+inline at::Tensor count_nonzero(const at::Tensor & self, at::IntArrayRef dim) {
+    return at::_ops::count_nonzero_dim_IntList::call(self, dim);
+}
+
+// aten::count_nonzero(Tensor self, int? dim=None) -> Tensor
+inline at::Tensor count_nonzero(const at::Tensor & self, ::std::optional<int64_t> dim=::std::nullopt) {
+    return at::_ops::count_nonzero::call(self, dim);
+}
+
+// aten::count_nonzero.dim_IntList_out(Tensor self, int[] dim, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & count_nonzero_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim) {
+    return at::_ops::count_nonzero_dim_IntList_out::call(self, dim, out);
+}
+// aten::count_nonzero.dim_IntList_out(Tensor self, int[] dim, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & count_nonzero_outf(const at::Tensor & self, at::IntArrayRef dim, at::Tensor & out) {
+    return at::_ops::count_nonzero_dim_IntList_out::call(self, dim, out);
+}
+
+// aten::count_nonzero.out(Tensor self, int? dim=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & count_nonzero_out(at::Tensor & out, const at::Tensor & self, ::std::optional<int64_t> dim=::std::nullopt) {
+    return at::_ops::count_nonzero_out::call(self, dim, out);
+}
+// aten::count_nonzero.out(Tensor self, int? dim=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & count_nonzero_outf(const at::Tensor & self, ::std::optional<int64_t> dim, at::Tensor & out) {
+    return at::_ops::count_nonzero_out::call(self, dim, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/count_nonzero_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/count_nonzero_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..14e74964e002c22eafa98c59e5dcc1612020d049
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/count_nonzero_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 count_nonzero_dim_IntList {
+  using schema = at::Tensor (const at::Tensor &, at::IntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::count_nonzero";
+  static constexpr const char* overload_name = "dim_IntList";
+  static constexpr const char* schema_str = "count_nonzero.dim_IntList(Tensor self, int[] dim) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::IntArrayRef dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef dim);
+};
+
+struct TORCH_API count_nonzero {
+  using schema = 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::count_nonzero";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "count_nonzero(Tensor self, int? dim=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, ::std::optional<int64_t> dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, ::std::optional<int64_t> dim);
+};
+
+struct TORCH_API count_nonzero_dim_IntList_out {
+  using schema = at::Tensor & (const at::Tensor &, at::IntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::count_nonzero";
+  static constexpr const char* overload_name = "dim_IntList_out";
+  static constexpr const char* schema_str = "count_nonzero.dim_IntList_out(Tensor self, int[] dim, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::IntArrayRef dim, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef dim, at::Tensor & out);
+};
+
+struct TORCH_API count_nonzero_out {
+  using schema = 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::count_nonzero";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "count_nonzero.out(Tensor self, int? dim=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, ::std::optional<int64_t> dim, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, ::std::optional<int64_t> dim, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cross_entropy_loss_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cross_entropy_loss_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3775f29b5ee450c9e7c05d40ff09564183fdf078
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cross_entropy_loss_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 cross_entropy_loss(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean, int64_t ignore_index=-100, double label_smoothing=0.0);
+TORCH_API at::Tensor cross_entropy_loss_symint(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean, c10::SymInt ignore_index=-100, double label_smoothing=0.0);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cross_entropy_loss_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cross_entropy_loss_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..184ed2ab986748c541318d01df2478844deb1540
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cross_entropy_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 cross_entropy_loss_symint(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean, c10::SymInt ignore_index=-100, double label_smoothing=0.0);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_add_relu_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_add_relu_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fbe6a675d17e830f85c1e9e38d8e71f9d3dd2a60
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_add_relu_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 cudnn_convolution_add_relu(const at::Tensor & self, const at::Tensor & weight, const at::Tensor & z, const ::std::optional<at::Scalar> & alpha, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, int64_t groups);
+TORCH_API at::Tensor cudnn_convolution_add_relu_symint(const at::Tensor & self, const at::Tensor & weight, const at::Tensor & z, const ::std::optional<at::Scalar> & alpha, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, c10::SymInt groups);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3351ab2d648e017c1ee90bf925f8f39d3d9f183b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cudnn_convolution_cuda_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor cudnn_convolution(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, bool allow_tf32);
+TORCH_API at::Tensor cudnn_convolution_symint(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32);
+TORCH_API at::Tensor & cudnn_convolution_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, bool allow_tf32);
+TORCH_API at::Tensor & cudnn_convolution_outf(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, bool allow_tf32, at::Tensor & out);
+TORCH_API at::Tensor & cudnn_convolution_symint_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32);
+TORCH_API at::Tensor & cudnn_convolution_symint_outf(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, bool allow_tf32, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e22996dfebc5d5e831f05521ea68fafa7b84fa29
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum_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 cumsum(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & cumsum_out(at::Tensor & out, const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & cumsum_outf(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+TORCH_API at::Tensor & cumsum_(at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..4756a1ad5ae360c62f663cc9f0025d0781223fe3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/cumsum_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 cumsum {
+  using schema = at::Tensor (const at::Tensor &, 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::cumsum";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "cumsum(Tensor self, int dim, *, ScalarType? dtype=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype);
+};
+
+struct TORCH_API cumsum_ {
+  using schema = at::Tensor & (at::Tensor &, 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::cumsum_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "cumsum_(Tensor(a!) self, int dim, *, ScalarType? dtype=None) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype);
+};
+
+struct TORCH_API cumsum_out {
+  using schema = at::Tensor & (const at::Tensor &, int64_t, ::std::optional<at::ScalarType>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::cumsum";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "cumsum.out(Tensor self, int dim, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+};
+
+struct TORCH_API cumsum_dimname {
+  using schema = at::Tensor (const at::Tensor &, at::Dimname, ::std::optional<at::ScalarType>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::cumsum";
+  static constexpr const char* overload_name = "dimname";
+  static constexpr const char* schema_str = "cumsum.dimname(Tensor self, Dimname dim, *, ScalarType? dtype=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::Dimname dim, ::std::optional<at::ScalarType> dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname dim, ::std::optional<at::ScalarType> dtype);
+};
+
+struct TORCH_API cumsum__dimname {
+  using schema = at::Tensor & (at::Tensor &, at::Dimname, ::std::optional<at::ScalarType>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::cumsum_";
+  static constexpr const char* overload_name = "dimname";
+  static constexpr const char* schema_str = "cumsum_.dimname(Tensor(a!) self, Dimname dim, *, ScalarType? dtype=None) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, at::Dimname dim, ::std::optional<at::ScalarType> dtype);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, at::Dimname dim, ::std::optional<at::ScalarType> dtype);
+};
+
+struct TORCH_API cumsum_dimname_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Dimname, ::std::optional<at::ScalarType>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::cumsum";
+  static constexpr const char* overload_name = "dimname_out";
+  static constexpr const char* schema_str = "cumsum.dimname_out(Tensor self, Dimname dim, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Dimname dim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname dim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/diagonal_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/diagonal_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9fe57a6c3bfaf5d299dd013205d8f9ccc3ef7ba7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/diagonal_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 diagonal_backward {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, int64_t, int64_t, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::diagonal_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "diagonal_backward(Tensor grad_output, SymInt[] input_sizes, int offset, int dim1, int dim2) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad_output, c10::SymIntArrayRef input_sizes, int64_t offset, int64_t dim1, int64_t dim2);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, c10::SymIntArrayRef input_sizes, int64_t offset, int64_t dim1, int64_t dim2);
+};
+
+struct TORCH_API diagonal_backward_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, int64_t, int64_t, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::diagonal_backward";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "diagonal_backward.out(Tensor grad_output, SymInt[] input_sizes, int offset, int dim1, int dim2, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & grad_output, c10::SymIntArrayRef input_sizes, int64_t offset, int64_t dim1, int64_t dim2, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, c10::SymIntArrayRef input_sizes, int64_t offset, int64_t dim1, int64_t dim2, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/diagonal_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/diagonal_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..578a18450b03fba039e055c5df87a31596dd3131
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/diagonal_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 diagonal {
+  using schema = at::Tensor (const at::Tensor &, int64_t, int64_t, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::diagonal";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "diagonal(Tensor(a) self, int offset=0, int dim1=0, int dim2=1) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, int64_t offset, int64_t dim1, int64_t dim2);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t offset, int64_t dim1, int64_t dim2);
+};
+
+struct TORCH_API diagonal_Dimname {
+  using schema = at::Tensor (const at::Tensor &, at::Dimname, at::Dimname, at::Dimname, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::diagonal";
+  static constexpr const char* overload_name = "Dimname";
+  static constexpr const char* schema_str = "diagonal.Dimname(Tensor(a) self, *, Dimname outdim, Dimname dim1, Dimname dim2, int offset=0) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, at::Dimname outdim, at::Dimname dim1, at::Dimname dim2, int64_t offset);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname outdim, at::Dimname dim1, at::Dimname dim2, int64_t offset);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/diff_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/diff_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..f7ae14bef428d6fc24eb2f39de7e7b8d09caac74
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/diff_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 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={});
+TORCH_API at::Tensor & diff_out(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);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/elu_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/elu_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..8ffc9cdbf33545b4738efa07f688983d37f8ef82
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/elu_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 elu_backward_grad_input {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, const at::Scalar &, const at::Scalar &, bool, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::elu_backward";
+  static constexpr const char* overload_name = "grad_input";
+  static constexpr const char* schema_str = "elu_backward.grad_input(Tensor grad_output, Scalar alpha, Scalar scale, Scalar input_scale, bool is_result, Tensor self_or_result, *, Tensor(a!) grad_input) -> Tensor(a!)";
+  static at::Tensor & call(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);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+struct TORCH_API elu_backward {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &, const at::Scalar &, const at::Scalar &, bool, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::elu_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "elu_backward(Tensor grad_output, Scalar alpha, Scalar scale, Scalar input_scale, bool is_result, Tensor self_or_result) -> Tensor";
+  static at::Tensor call(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);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..cc685f616174b4fa0133f4f7812c6689d79ec3ee
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_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 embedding_backward_symint(const at::Tensor & grad, const at::Tensor & indices, c10::SymInt num_weights, c10::SymInt padding_idx, bool scale_grad_by_freq, bool sparse);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_bag_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_bag_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a9a3ef86cbe01ab141e4d6b00607ecd86e44110a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/embedding_bag_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 embedding_bag {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, bool, int64_t, bool, const ::std::optional<at::Tensor> &, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::embedding_bag";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "embedding_bag(Tensor weight, Tensor indices, Tensor offsets, bool scale_grad_by_freq=False, int mode=0, bool sparse=False, Tensor? per_sample_weights=None, bool include_last_offset=False) -> (Tensor, Tensor, Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, bool include_last_offset);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, bool include_last_offset);
+};
+
+struct TORCH_API embedding_bag_padding_idx {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, bool, int64_t, bool, const ::std::optional<at::Tensor> &, bool, ::std::optional<int64_t>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::embedding_bag";
+  static constexpr const char* overload_name = "padding_idx";
+  static constexpr const char* schema_str = "embedding_bag.padding_idx(Tensor weight, Tensor indices, Tensor offsets, bool scale_grad_by_freq, int mode, bool sparse, Tensor? per_sample_weights, bool include_last_offset, int? padding_idx) -> (Tensor, Tensor, Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, bool include_last_offset, ::std::optional<int64_t> padding_idx);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, bool include_last_offset, ::std::optional<int64_t> padding_idx);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/empty_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0eb615ff9c7b5480a076c04161a813c99b84eee4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/empty_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 & empty_out(at::Tensor & out, at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor & empty_outf(at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+TORCH_API at::Tensor & empty_symint_out(at::Tensor & out, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor & empty_symint_outf(c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/erf_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/erf_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..316bdc4ba4f80b52f6cff0f56b24d24200a210b6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/erf_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 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::erf";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "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 erf_ {
+  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::erf_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "erf_(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 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::erf";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "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/exp2_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/exp2_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..81d8db5a0be21b8f08b67dcc0745e65a79af6a64
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/exp2_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 exp2 {
+  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::exp2";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "exp2(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 exp2_ {
+  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::exp2_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "exp2_(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 exp2_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::exp2";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "exp2.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/expm1_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/expm1_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..7c2e4a836efa76a8151261e1a0c592e3ab6fb809
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/expm1_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_expm1 : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/exponential.h b/lib/python3.12/site-packages/torch/include/ATen/ops/exponential.h
new file mode 100644
index 0000000000000000000000000000000000000000..7c7ec9c91385209c6e04f8c3aa00df74e05f2fa5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/exponential.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/exponential_ops.h>
+
+namespace at {
+
+
+// aten::exponential.out(Tensor self, float lambd=1, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & exponential_out(at::Tensor & out, const at::Tensor & self, double lambd=1, ::std::optional<at::Generator> generator=::std::nullopt) {
+    return at::_ops::exponential_out::call(self, lambd, generator, out);
+}
+// aten::exponential.out(Tensor self, float lambd=1, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & exponential_outf(const at::Tensor & self, double lambd, ::std::optional<at::Generator> generator, at::Tensor & out) {
+    return at::_ops::exponential_out::call(self, lambd, generator, out);
+}
+
+// aten::exponential(Tensor self, float lambd=1, *, Generator? generator=None) -> Tensor
+inline at::Tensor exponential(const at::Tensor & self, double lambd=1, ::std::optional<at::Generator> generator=::std::nullopt) {
+    return at::_ops::exponential::call(self, lambd, generator);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/exponential_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/exponential_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..603d07f70ae0a0f996c35959d245fd6d360346e8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/exponential_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 & exponential_(at::Tensor & self, double lambd=1, ::std::optional<at::Generator> generator=::std::nullopt);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_cachemask_backward_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_cachemask_backward_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..28e4c5e78074c36cc1e317110732569edabebf69
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_cachemask_backward_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor fake_quantize_per_channel_affine_cachemask_backward(const at::Tensor & grad, const at::Tensor & mask);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3a918173b1fe0aadebe2d77c28e658e8c879ac7a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_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 fake_quantize_per_channel_affine(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_tensor_affine_cachemask_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_tensor_affine_cachemask_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8553a6158d02dc5d7c1ff9b09dd19ce8c1b6a9e9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fake_quantize_per_tensor_affine_cachemask_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> fake_quantize_per_tensor_affine_cachemask(const at::Tensor & self, double scale, int64_t zero_point, int64_t quant_min, int64_t quant_max);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_fp16_weight_fp32_activation_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_fp16_weight_fp32_activation_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..78cf2b93526e5b1d604996322738a4d029966800
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_fp16_weight_fp32_activation_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 fbgemm_linear_fp16_weight_fp32_activation(const at::Tensor & input, const at::Tensor & packed_weight, const at::Tensor & bias);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_quantize_weight.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_quantize_weight.h
new file mode 100644
index 0000000000000000000000000000000000000000..8b4502b428df0462747d517c94bb536183e598fd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fbgemm_linear_quantize_weight.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/fbgemm_linear_quantize_weight_ops.h>
+
+namespace at {
+
+
+// aten::fbgemm_linear_quantize_weight(Tensor input) -> (Tensor, Tensor, float, int)
+inline ::std::tuple<at::Tensor,at::Tensor,double,int64_t> fbgemm_linear_quantize_weight(const at::Tensor & input) {
+    return at::_ops::fbgemm_linear_quantize_weight::call(input);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fft2_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fft2_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e27e16e9472317ddcb815f429f12b7711655b7f7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fft2_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor fft_fft2(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor fft_fft2_symint(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_fft2_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_fft2_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+TORCH_API at::Tensor & fft_fft2_symint_out(at::Tensor & out, const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_fft2_symint_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fftfreq_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fftfreq_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5fc19cf4611b5159c4c7ebcd286084feb91529e7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fftfreq_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_fftfreq(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_fftfreq_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_fftn.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fftn.h
new file mode 100644
index 0000000000000000000000000000000000000000..64df4d2ac781466b8368887d378ef5fc7b340e42
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_fftn.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_fftn_ops.h>
+
+namespace at {
+
+
+// aten::fft_fftn(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None) -> Tensor
+inline at::Tensor fft_fftn(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fftn::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_fftn(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fftn::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm);
+  }
+}
+
+// aten::fft_fftn(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None) -> Tensor
+inline at::Tensor fft_fftn_symint(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fftn::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_fftn(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fftn::call(self, s, dim, norm);
+  }
+}
+
+// aten::fft_fftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_fftn_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fftn_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_fftn_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fftn_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_fftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_fftn_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_fftn_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_fftn_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_fftn_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_fftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_fftn_symint_out(at::Tensor & out, const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fftn_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_fftn_out(at::Tensor & out, const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_fftn_out::call(self, s, dim, norm, out);
+  }
+}
+
+// aten::fft_fftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_fftn_symint_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_fftn_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_fftn_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_fftn_out::call(self, s, dim, norm, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_hfft2_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_hfft2_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5cdafd3e1c597bd50bcf8ed4c5c3d5ddb5b4f36e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_hfft2_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_hfft2_symint(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API const at::Tensor & fft_hfft2_symint_out(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, const at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftn_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftn_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3875596f6b5650bf5837bf033a5a678e4e0339c2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ifftn_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor fft_ifftn(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor fft_ifftn_symint(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_ifftn_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_ifftn_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+TORCH_API at::Tensor & fft_ifftn_symint_out(at::Tensor & out, const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::OptionalIntArrayRef dim=::std::nullopt, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_ifftn_symint_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ihfftn_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ihfftn_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..18b4749cd89bc85d58b8b314612cee65c4f305a3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_ihfftn_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_ihfftn {
+  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_ihfftn";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "fft_ihfftn(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_ihfftn_out {
+  using schema = const at::Tensor & (const at::Tensor &, at::OptionalSymIntArrayRef, at::OptionalIntArrayRef, ::std::optional<c10::string_view>, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::fft_ihfftn";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "fft_ihfftn.out(Tensor self, SymInt[1]? s=None, int[1]? dim=None, str? norm=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static const at::Tensor & call(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, const at::Tensor & out);
+  static const at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::OptionalSymIntArrayRef s, at::OptionalIntArrayRef dim, ::std::optional<c10::string_view> norm, const at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftfreq.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftfreq.h
new file mode 100644
index 0000000000000000000000000000000000000000..96c293947a67ec053ae63225e39b026b78a23f1a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftfreq.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/fft_rfftfreq_ops.h>
+
+namespace at {
+
+
+// aten::fft_rfftfreq(int n, float d=1.0, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor fft_rfftfreq(int64_t n, double d=1.0, at::TensorOptions options={}) {
+    return at::_ops::fft_rfftfreq::call(n, d, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::fft_rfftfreq(int n, float d=1.0, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor fft_rfftfreq(int64_t n, double d, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::fft_rfftfreq::call(n, d, dtype, layout, device, pin_memory);
+}
+
+// aten::fft_rfftfreq.out(int n, float d=1.0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_rfftfreq_out(at::Tensor & out, int64_t n, double d=1.0) {
+    return at::_ops::fft_rfftfreq_out::call(n, d, out);
+}
+// aten::fft_rfftfreq.out(int n, float d=1.0, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_rfftfreq_outf(int64_t n, double d, at::Tensor & out) {
+    return at::_ops::fft_rfftfreq_out::call(n, d, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftfreq_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftfreq_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ff9ff35144cace1e5592551bb188968366fb11ec
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/fft_rfftfreq_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 fft_rfftfreq(int64_t n, double d=1.0, at::TensorOptions options={});
+TORCH_API at::Tensor fft_rfftfreq(int64_t n, double d, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor & fft_rfftfreq_out(at::Tensor & out, int64_t n, double d=1.0);
+TORCH_API at::Tensor & fft_rfftfreq_outf(int64_t n, double d, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/flip.h b/lib/python3.12/site-packages/torch/include/ATen/ops/flip.h
new file mode 100644
index 0000000000000000000000000000000000000000..54369801e3e49b0cea2e8094e29a95f1672fad73
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/flip.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/flip_ops.h>
+
+namespace at {
+
+
+// aten::flip(Tensor self, int[] dims) -> Tensor
+inline at::Tensor flip(const at::Tensor & self, at::IntArrayRef dims) {
+    return at::_ops::flip::call(self, dims);
+}
+
+// aten::flip.out(Tensor self, int[] dims, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & flip_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dims) {
+    return at::_ops::flip_out::call(self, dims, out);
+}
+// aten::flip.out(Tensor self, int[] dims, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & flip_outf(const at::Tensor & self, at::IntArrayRef dims, at::Tensor & out) {
+    return at::_ops::flip_out::call(self, dims, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/frac_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/frac_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..6256a29564b28d06fe7140156fa2ffa79ea5f79c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/frac_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 frac {
+  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::frac";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "frac(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 frac_ {
+  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::frac_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "frac_(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 frac_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::frac";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "frac.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/full_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/full_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4d85aa566cec158d40123473ea769e3c5563307e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/full_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,34 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor full(at::IntArrayRef size, const at::Scalar & fill_value, ::std::optional<at::DimnameList> names, at::TensorOptions options={});
+TORCH_API at::Tensor full(at::IntArrayRef size, const at::Scalar & fill_value, ::std::optional<at::DimnameList> names, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor & full_out(at::Tensor & out, at::IntArrayRef size, const at::Scalar & fill_value, ::std::optional<at::DimnameList> names);
+TORCH_API at::Tensor & full_outf(at::IntArrayRef size, const at::Scalar & fill_value, ::std::optional<at::DimnameList> names, at::Tensor & out);
+TORCH_API at::Tensor full(at::IntArrayRef size, const at::Scalar & fill_value, at::TensorOptions options={});
+TORCH_API at::Tensor full(at::IntArrayRef size, const at::Scalar & fill_value, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor full_symint(c10::SymIntArrayRef size, const at::Scalar & fill_value, at::TensorOptions options={});
+TORCH_API at::Tensor full_symint(c10::SymIntArrayRef size, const at::Scalar & fill_value, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor & full_out(at::Tensor & out, at::IntArrayRef size, const at::Scalar & fill_value);
+TORCH_API at::Tensor & full_outf(at::IntArrayRef size, const at::Scalar & fill_value, at::Tensor & out);
+TORCH_API at::Tensor & full_symint_out(at::Tensor & out, c10::SymIntArrayRef size, const at::Scalar & fill_value);
+TORCH_API at::Tensor & full_symint_outf(c10::SymIntArrayRef size, const at::Scalar & fill_value, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/geqrf.h b/lib/python3.12/site-packages/torch/include/ATen/ops/geqrf.h
new file mode 100644
index 0000000000000000000000000000000000000000..02f8de5df78dcaa3503bd36a7c1d7660f3722570
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/geqrf.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/geqrf_ops.h>
+
+namespace at {
+
+
+// aten::geqrf.a(Tensor self, *, Tensor(a!) a, Tensor(b!) tau) -> (Tensor(a!) a, Tensor(b!) tau)
+inline ::std::tuple<at::Tensor &,at::Tensor &> geqrf_out(at::Tensor & a, at::Tensor & tau, const at::Tensor & self) {
+    return at::_ops::geqrf_a::call(self, a, tau);
+}
+// aten::geqrf.a(Tensor self, *, Tensor(a!) a, Tensor(b!) tau) -> (Tensor(a!) a, Tensor(b!) tau)
+inline ::std::tuple<at::Tensor &,at::Tensor &> geqrf_outf(const at::Tensor & self, at::Tensor & a, at::Tensor & tau) {
+    return at::_ops::geqrf_a::call(self, a, tau);
+}
+
+// aten::geqrf(Tensor self) -> (Tensor a, Tensor tau)
+inline ::std::tuple<at::Tensor,at::Tensor> geqrf(const at::Tensor & self) {
+    return at::_ops::geqrf::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/ger_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/ger_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f2705ad7b0110ea048ec94eff55d784845407e7b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/ger_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 ger {
+  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::ger";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "ger(Tensor self, Tensor vec2) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & vec2);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & vec2);
+};
+
+struct TORCH_API ger_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::ger";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "ger.out(Tensor self, Tensor vec2, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & vec2, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & vec2, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/glu_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/glu_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..b7eabe805fe7d31d050e44023458812937847b82
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/glu_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/glu_backward_ops.h>
+
+namespace at {
+
+
+// aten::glu_backward.grad_input(Tensor grad_output, Tensor self, int dim, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & glu_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, int64_t dim) {
+    return at::_ops::glu_backward_grad_input::call(grad_output, self, dim, grad_input);
+}
+// aten::glu_backward.grad_input(Tensor grad_output, Tensor self, int dim, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & glu_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, int64_t dim, at::Tensor & grad_input) {
+    return at::_ops::glu_backward_grad_input::call(grad_output, self, dim, grad_input);
+}
+
+// aten::glu_backward(Tensor grad_output, Tensor self, int dim) -> Tensor
+inline at::Tensor glu_backward(const at::Tensor & grad_output, const at::Tensor & self, int64_t dim) {
+    return at::_ops::glu_backward::call(grad_output, self, dim);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/glu_backward_jvp_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/glu_backward_jvp_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..9479ffb55eb00d262e157d60462b5921a9384e58
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/glu_backward_jvp_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 glu_backward_jvp(const at::Tensor & grad_x, const at::Tensor & grad_glu, const at::Tensor & x, const at::Tensor & dgrad_glu, const at::Tensor & dx, int64_t dim);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/glu_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/glu_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a454a6da1ee244b5bedbaccdca79d05e7e36411d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/glu_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 glu(const at::Tensor & self, int64_t dim=-1);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/greater_equal_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/greater_equal_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..001198a67530a3506e4f6cc2295c03abc598cd62
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/greater_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 greater_equal(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & greater_equal_out(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & greater_equal_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor greater_equal(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & greater_equal_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & greater_equal_(at::Tensor & self, const at::Tensor & other);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3ff02f0756280681b89748b14df667fd501ff374
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/grid_sampler_3d_backward_cpu_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> grid_sampler_3d_backward(const at::Tensor & grad_output, const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners, ::std::array<bool,2> output_mask);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gru_cell_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gru_cell_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4384a49b362c3a525b8421471dac1bf8626512a0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gru_cell_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 gru_cell(const at::Tensor & input, const at::Tensor & hx, const at::Tensor & w_ih, const at::Tensor & w_hh, const ::std::optional<at::Tensor> & b_ih={}, const ::std::optional<at::Tensor> & b_hh={});
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/gt_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/gt_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..d9069b98c3d16a2d8944e9e512ed5c4d167de773
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/gt_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 gt_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::gt";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "gt.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 gt_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::gt";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "gt.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 gt_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::gt";
+  static constexpr const char* overload_name = "Tensor_out";
+  static constexpr const char* schema_str = "gt.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 gt_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::gt";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "gt.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 gt__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::gt_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "gt_.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 gt__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::gt_";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "gt_.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/hamming_window.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hamming_window.h
new file mode 100644
index 0000000000000000000000000000000000000000..c73cb398e12efe83ba3467f067e318dcb7d0e552
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hamming_window.h
@@ -0,0 +1,97 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/hamming_window_ops.h>
+
+namespace at {
+
+
+// aten::hamming_window(int window_length, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor hamming_window(int64_t window_length, at::TensorOptions options={}) {
+    return at::_ops::hamming_window::call(window_length, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::hamming_window(int window_length, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor hamming_window(int64_t window_length, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::hamming_window::call(window_length, dtype, layout, device, pin_memory);
+}
+
+// aten::hamming_window.periodic(int window_length, bool periodic, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor hamming_window(int64_t window_length, bool periodic, at::TensorOptions options={}) {
+    return at::_ops::hamming_window_periodic::call(window_length, periodic, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::hamming_window.periodic(int window_length, bool periodic, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor hamming_window(int64_t window_length, bool periodic, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::hamming_window_periodic::call(window_length, periodic, dtype, layout, device, pin_memory);
+}
+
+// aten::hamming_window.periodic_alpha(int window_length, bool periodic, float alpha, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor hamming_window(int64_t window_length, bool periodic, double alpha, at::TensorOptions options={}) {
+    return at::_ops::hamming_window_periodic_alpha::call(window_length, periodic, alpha, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::hamming_window.periodic_alpha(int window_length, bool periodic, float alpha, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor hamming_window(int64_t window_length, bool periodic, double alpha, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::hamming_window_periodic_alpha::call(window_length, periodic, alpha, dtype, layout, device, pin_memory);
+}
+
+// aten::hamming_window.periodic_alpha_beta(int window_length, bool periodic, float alpha, float beta, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor hamming_window(int64_t window_length, bool periodic, double alpha, double beta, at::TensorOptions options={}) {
+    return at::_ops::hamming_window_periodic_alpha_beta::call(window_length, periodic, alpha, beta, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::hamming_window.periodic_alpha_beta(int window_length, bool periodic, float alpha, float beta, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor hamming_window(int64_t window_length, bool periodic, double alpha, double beta, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::hamming_window_periodic_alpha_beta::call(window_length, periodic, alpha, beta, dtype, layout, device, pin_memory);
+}
+
+// aten::hamming_window.out(int window_length, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hamming_window_out(at::Tensor & out, int64_t window_length) {
+    return at::_ops::hamming_window_out::call(window_length, out);
+}
+// aten::hamming_window.out(int window_length, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hamming_window_outf(int64_t window_length, at::Tensor & out) {
+    return at::_ops::hamming_window_out::call(window_length, out);
+}
+
+// aten::hamming_window.periodic_out(int window_length, bool periodic, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hamming_window_out(at::Tensor & out, int64_t window_length, bool periodic) {
+    return at::_ops::hamming_window_periodic_out::call(window_length, periodic, out);
+}
+// aten::hamming_window.periodic_out(int window_length, bool periodic, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hamming_window_outf(int64_t window_length, bool periodic, at::Tensor & out) {
+    return at::_ops::hamming_window_periodic_out::call(window_length, periodic, out);
+}
+
+// aten::hamming_window.periodic_alpha_out(int window_length, bool periodic, float alpha, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hamming_window_out(at::Tensor & out, int64_t window_length, bool periodic, double alpha) {
+    return at::_ops::hamming_window_periodic_alpha_out::call(window_length, periodic, alpha, out);
+}
+// aten::hamming_window.periodic_alpha_out(int window_length, bool periodic, float alpha, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hamming_window_outf(int64_t window_length, bool periodic, double alpha, at::Tensor & out) {
+    return at::_ops::hamming_window_periodic_alpha_out::call(window_length, periodic, alpha, out);
+}
+
+// aten::hamming_window.periodic_alpha_beta_out(int window_length, bool periodic, float alpha, float beta, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hamming_window_out(at::Tensor & out, int64_t window_length, bool periodic, double alpha, double beta) {
+    return at::_ops::hamming_window_periodic_alpha_beta_out::call(window_length, periodic, alpha, beta, out);
+}
+// aten::hamming_window.periodic_alpha_beta_out(int window_length, bool periodic, float alpha, float beta, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hamming_window_outf(int64_t window_length, bool periodic, double alpha, double beta, at::Tensor & out) {
+    return at::_ops::hamming_window_periodic_alpha_beta_out::call(window_length, periodic, alpha, beta, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hamming_window_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hamming_window_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..297dc6a1ce3738c4c30e41220a0523490ca89022
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hamming_window_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 hamming_window {
+  using schema = at::Tensor (int64_t, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::hamming_window";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "hamming_window(int window_length, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(int64_t window_length, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API hamming_window_periodic {
+  using schema = at::Tensor (int64_t, bool, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::hamming_window";
+  static constexpr const char* overload_name = "periodic";
+  static constexpr const char* schema_str = "hamming_window.periodic(int window_length, bool periodic, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(int64_t window_length, bool periodic, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, bool periodic, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API hamming_window_periodic_alpha {
+  using schema = at::Tensor (int64_t, bool, double, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::hamming_window";
+  static constexpr const char* overload_name = "periodic_alpha";
+  static constexpr const char* schema_str = "hamming_window.periodic_alpha(int window_length, bool periodic, float alpha, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(int64_t window_length, bool periodic, double alpha, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, bool periodic, double alpha, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API hamming_window_periodic_alpha_beta {
+  using schema = at::Tensor (int64_t, bool, double, double, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::hamming_window";
+  static constexpr const char* overload_name = "periodic_alpha_beta";
+  static constexpr const char* schema_str = "hamming_window.periodic_alpha_beta(int window_length, bool periodic, float alpha, float beta, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(int64_t window_length, bool periodic, double alpha, double beta, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, bool periodic, double alpha, double beta, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API hamming_window_out {
+  using schema = at::Tensor & (int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::hamming_window";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "hamming_window.out(int window_length, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t window_length, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, at::Tensor & out);
+};
+
+struct TORCH_API hamming_window_periodic_out {
+  using schema = at::Tensor & (int64_t, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::hamming_window";
+  static constexpr const char* overload_name = "periodic_out";
+  static constexpr const char* schema_str = "hamming_window.periodic_out(int window_length, bool periodic, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t window_length, bool periodic, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, bool periodic, at::Tensor & out);
+};
+
+struct TORCH_API hamming_window_periodic_alpha_out {
+  using schema = at::Tensor & (int64_t, bool, double, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::hamming_window";
+  static constexpr const char* overload_name = "periodic_alpha_out";
+  static constexpr const char* schema_str = "hamming_window.periodic_alpha_out(int window_length, bool periodic, float alpha, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t window_length, bool periodic, double alpha, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, bool periodic, double alpha, at::Tensor & out);
+};
+
+struct TORCH_API hamming_window_periodic_alpha_beta_out {
+  using schema = at::Tensor & (int64_t, bool, double, double, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::hamming_window";
+  static constexpr const char* overload_name = "periodic_alpha_beta_out";
+  static constexpr const char* schema_str = "hamming_window.periodic_alpha_beta_out(int window_length, bool periodic, float alpha, float beta, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(int64_t window_length, bool periodic, double alpha, double beta, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, int64_t window_length, bool periodic, double alpha, double beta, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..62684b1ac058212095c6435e0c28ed1d72b32125
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_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 hardshrink_backward(const at::Tensor & grad_out, const at::Tensor & self, const at::Scalar & lambd);
+TORCH_API at::Tensor & hardshrink_backward_out(at::Tensor & grad_input, const at::Tensor & grad_out, const at::Tensor & self, const at::Scalar & lambd);
+TORCH_API at::Tensor & hardshrink_backward_outf(const at::Tensor & grad_out, const at::Tensor & self, const at::Scalar & lambd, at::Tensor & grad_input);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9753bd7884ef0dc507be78a630f6971f562bd54d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardshrink_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 hardshrink_backward_grad_input {
+  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::hardshrink_backward";
+  static constexpr const char* overload_name = "grad_input";
+  static constexpr const char* schema_str = "hardshrink_backward.grad_input(Tensor grad_out, Tensor self, Scalar lambd, *, Tensor(a!) grad_input) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & grad_out, const at::Tensor & self, const at::Scalar & lambd, at::Tensor & grad_input);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_out, const at::Tensor & self, const at::Scalar & lambd, at::Tensor & grad_input);
+};
+
+struct TORCH_API hardshrink_backward {
+  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::hardshrink_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "hardshrink_backward(Tensor grad_out, Tensor self, Scalar lambd) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad_out, const at::Tensor & self, const at::Scalar & lambd);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_out, const at::Tensor & self, const at::Scalar & lambd);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_backward_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_backward_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..38e950a9e74183da370fda5496cec06e932a26d0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hardsigmoid_backward_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor hardsigmoid_backward(const at::Tensor & grad_output, const at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/histc_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/histc_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d1328d6bab03f8e24f8256931f04997facbbde84
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/histc_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 histc(const at::Tensor & self, int64_t bins=100, const at::Scalar & min=0, const at::Scalar & max=0);
+TORCH_API at::Tensor & histc_out(at::Tensor & out, const at::Tensor & self, int64_t bins=100, const at::Scalar & min=0, const at::Scalar & max=0);
+TORCH_API at::Tensor & histc_outf(const at::Tensor & self, int64_t bins, const at::Scalar & min, const at::Scalar & max, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/huber_loss_backward_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/huber_loss_backward_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c040c728a022c82d6495c6102fe9424e176b9fc3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/huber_loss_backward_cpu_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor & huber_loss_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction, double delta);
+TORCH_API at::Tensor & huber_loss_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction, double delta, at::Tensor & grad_input);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/hypot_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hypot_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..4825a2a111d53ee864a820cfa7b6201a6639c107
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hypot_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_hypot : 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/hypot_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/hypot_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..cfb04ab6805e24ee754a0374795222775995d5f8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/hypot_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 hypot_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::hypot";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "hypot.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 hypot {
+  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::hypot";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "hypot(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 hypot_ {
+  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::hypot_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "hypot_(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/igamma_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/igamma_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..993156e810d1e7fb1576fda00cf1855602473f28
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/igamma_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 igamma(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & igamma_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/igamma_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/igamma_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..346cefa58c38137cd370a4e64eed6531e14961b1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/igamma_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 igamma(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & igamma_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & igamma_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & igamma_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/igamma_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/igamma_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1abc1ed59555bb1223fa9b4fa14beb50b2da6cc0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/igamma_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 igamma(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & igamma_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & igamma_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & igamma_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/im2col_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/im2col_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..77db0b79ae45859727929e1c65b36d2a56155830
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/im2col_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 im2col(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride);
+TORCH_API at::Tensor & im2col_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride);
+TORCH_API at::Tensor & im2col_outf(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef dilation, at::IntArrayRef padding, at::IntArrayRef stride, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_copy_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_copy_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..1bc767b18279b2c9f92c52281a51e3f43dea4d4e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_copy_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 index_copy_out {
+  using schema = at::Tensor & (const at::Tensor &, int64_t, 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::index_copy";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "index_copy.out(Tensor self, int dim, Tensor index, Tensor source, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source, at::Tensor & out);
+};
+
+struct TORCH_API index_copy_ {
+  using schema = at::Tensor & (at::Tensor &, int64_t, 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::index_copy_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "index_copy_(Tensor(a!) self, int dim, Tensor index, Tensor source) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source);
+};
+
+struct TORCH_API index_copy {
+  using schema = at::Tensor (const at::Tensor &, int64_t, 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::index_copy";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "index_copy(Tensor self, int dim, Tensor index, Tensor source) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & source);
+};
+
+struct TORCH_API index_copy__dimname {
+  using schema = at::Tensor & (at::Tensor &, at::Dimname, 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::index_copy_";
+  static constexpr const char* overload_name = "dimname";
+  static constexpr const char* schema_str = "index_copy_.dimname(Tensor(a!) self, Dimname dim, Tensor index, Tensor source) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, at::Dimname dim, const at::Tensor & index, const at::Tensor & source);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, at::Dimname dim, const at::Tensor & index, const at::Tensor & source);
+};
+
+struct TORCH_API index_copy_dimname {
+  using schema = at::Tensor (const at::Tensor &, at::Dimname, 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::index_copy";
+  static constexpr const char* overload_name = "dimname";
+  static constexpr const char* schema_str = "index_copy.dimname(Tensor self, Dimname dim, Tensor index, Tensor source) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::Dimname dim, const at::Tensor & index, const at::Tensor & source);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname dim, const at::Tensor & index, const at::Tensor & source);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..aaa014c73fe97f251f4d804791a36d262f0bae0e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_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 index_select_backward_symint(const at::Tensor & grad, c10::SymIntArrayRef self_sizes, int64_t dim, const at::Tensor & index);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..771318844ccbb86af2de4dbbf4226a3450314e72
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/index_select_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_select(const at::Tensor & self, int64_t dim, const at::Tensor & index);
+TORCH_API at::Tensor & index_select_out(at::Tensor & out, const at::Tensor & self, int64_t dim, const at::Tensor & index);
+TORCH_API at::Tensor & index_select_outf(const at::Tensor & self, int64_t dim, const at::Tensor & index, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/inner_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/inner_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..a4343decb14b38c7d3b44043df77318367a6c0ca
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/inner_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 inner(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & inner_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/inverse_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/inverse_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5cb926c8ce00859568771cc93d920d687bb5bf6f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/inverse_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 inverse(const at::Tensor & self);
+TORCH_API at::Tensor & inverse_out(const at::Tensor & self, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_leaf_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_leaf_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fed78e5b4eb6316eaecd5f05cd15ad0e70476c2f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_leaf_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 is_leaf(const at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_nonzero_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_nonzero_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..091fa5cdb037e375c8ca7a16ca9681ba9be5bd88
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_nonzero_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 is_nonzero(const at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_same_size.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_same_size.h
new file mode 100644
index 0000000000000000000000000000000000000000..95a1933dc6ecfe544fc1d99ae237fd549849eae3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_same_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/is_same_size_ops.h>
+
+namespace at {
+
+
+// aten::is_same_size(Tensor self, Tensor other) -> bool
+inline bool is_same_size(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::is_same_size::call(self, other);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_set_to_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_set_to_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8385c5190105b13d293b910f29f08a0dd5fd3502
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_set_to_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API bool is_set_to(const at::Tensor & self, const at::Tensor & tensor);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_signed_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_signed_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..453329a0b014107b7d361a6beb1bced97d0081d8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_signed_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_signed {
+  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_signed";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "is_signed(Tensor self) -> bool";
+  static bool call(const at::Tensor & self);
+  static bool redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/is_vulkan_available_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/is_vulkan_available_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..66c546844affec0fc71b273d3eacacc7f212febd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/is_vulkan_available_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_vulkan_available();
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/isin_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/isin_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..bd02933656f4ecfdb3a19f02cc78303158f93ef1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/isin_cpu_dispatch.h
@@ -0,0 +1,31 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor isin(const at::Tensor & elements, const at::Tensor & test_elements, bool assume_unique=false, bool invert=false);
+TORCH_API at::Tensor & isin_out(at::Tensor & out, const at::Tensor & elements, const at::Tensor & test_elements, bool assume_unique=false, bool invert=false);
+TORCH_API at::Tensor & isin_outf(const at::Tensor & elements, const at::Tensor & test_elements, bool assume_unique, bool invert, at::Tensor & out);
+TORCH_API at::Tensor isin(const at::Tensor & elements, const at::Scalar & test_element, bool assume_unique=false, bool invert=false);
+TORCH_API at::Tensor & isin_out(at::Tensor & out, const at::Tensor & elements, const at::Scalar & test_element, bool assume_unique=false, bool invert=false);
+TORCH_API at::Tensor & isin_outf(const at::Tensor & elements, const at::Scalar & test_element, bool assume_unique, bool invert, at::Tensor & out);
+TORCH_API at::Tensor isin(const at::Scalar & element, const at::Tensor & test_elements, bool assume_unique=false, bool invert=false);
+TORCH_API at::Tensor & isin_out(at::Tensor & out, const at::Scalar & element, const at::Tensor & test_elements, bool assume_unique=false, bool invert=false);
+TORCH_API at::Tensor & isin_outf(const at::Scalar & element, const at::Tensor & test_elements, bool assume_unique, bool invert, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/isneginf.h b/lib/python3.12/site-packages/torch/include/ATen/ops/isneginf.h
new file mode 100644
index 0000000000000000000000000000000000000000..37c65dfa52e9daa99a88fb796fdfa2b5737f005c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/isneginf.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/isneginf_ops.h>
+
+namespace at {
+
+
+// aten::isneginf(Tensor self) -> Tensor
+inline at::Tensor isneginf(const at::Tensor & self) {
+    return at::_ops::isneginf::call(self);
+}
+
+// aten::isneginf.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & isneginf_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::isneginf_out::call(self, out);
+}
+// aten::isneginf.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & isneginf_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::isneginf_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/kron_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/kron_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..100ce3c0ee001792719d7648113a9ceeeff86277
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/kron_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 kron(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & kron_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/kthvalue_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/kthvalue_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..45b01801bf0d1523a53e1b7dd92f8537f629d224
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/kthvalue_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> kthvalue(const at::Tensor & self, int64_t k, int64_t dim=-1, bool keepdim=false);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/kthvalue_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/kthvalue_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b8d133a1995d3b0f9cb4c8cc6f141983936023c2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/kthvalue_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::tuple<at::Tensor,at::Tensor> kthvalue(const at::Tensor & self, int64_t k, int64_t dim=-1, bool keepdim=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> kthvalue_out_cpu(const at::Tensor & self, int64_t k, int64_t dim, bool keepdim, at::Tensor & values, at::Tensor & indices);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> kthvalue_out_cuda(const at::Tensor & self, int64_t k, int64_t dim, bool keepdim, at::Tensor & values, at::Tensor & indices);
+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(const at::Tensor & self, int64_t k, at::Dimname dim, bool keepdim, at::Tensor & values, at::Tensor & indices);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..5e37440c0bfe1a4071c5086c0bad3293b2052a77
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_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/leaky_relu_backward_ops.h>
+
+namespace at {
+
+
+// aten::leaky_relu_backward.grad_input(Tensor grad_output, Tensor self, Scalar negative_slope, bool self_is_result, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & leaky_relu_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & negative_slope, bool self_is_result) {
+    return at::_ops::leaky_relu_backward_grad_input::call(grad_output, self, negative_slope, self_is_result, grad_input);
+}
+// aten::leaky_relu_backward.grad_input(Tensor grad_output, Tensor self, Scalar negative_slope, bool self_is_result, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & leaky_relu_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & negative_slope, bool self_is_result, at::Tensor & grad_input) {
+    return at::_ops::leaky_relu_backward_grad_input::call(grad_output, self, negative_slope, self_is_result, grad_input);
+}
+
+// aten::leaky_relu_backward(Tensor grad_output, Tensor self, Scalar negative_slope, bool self_is_result) -> Tensor
+inline at::Tensor leaky_relu_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & negative_slope, bool self_is_result) {
+    return at::_ops::leaky_relu_backward::call(grad_output, self, negative_slope, self_is_result);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_backward_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_backward_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..52aee492b82182dcae681340598b93b43729152d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_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_leaky_relu_backward : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & negative_slope, bool self_is_result);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3fbc282ab518dc9d8348d5042c79e3a6ad580cad
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/leaky_relu_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 leaky_relu(const at::Tensor & self, const at::Scalar & negative_slope=0.01);
+TORCH_API at::Tensor & leaky_relu_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & negative_slope=0.01);
+TORCH_API at::Tensor & leaky_relu_outf(const at::Tensor & self, const at::Scalar & negative_slope, at::Tensor & out);
+TORCH_API at::Tensor & leaky_relu_(at::Tensor & self, const at::Scalar & negative_slope=0.01);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/less_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/less_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..910b068c156c301ee0da36ddcc39cfb435cb59a2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/less_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 less(const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & less_out(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & less_(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor less(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & less_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & less_(at::Tensor & self, const at::Tensor & other);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lift_fresh_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lift_fresh_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b1b2d992dd4d2ba5e188e0ba13210bce05efdbe0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lift_fresh_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 lift_fresh(const at::Tensor & self);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lift_fresh_copy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lift_fresh_copy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c58192db2d2ef29a8200f38d5fb2cf5820474e1f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lift_fresh_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 & lift_fresh_copy_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor lift_fresh_copy(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lift_fresh_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lift_fresh_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..35d6590600571e0d9a9b1b21dc4a6fd701715a8c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lift_fresh_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 lift_fresh(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cholesky_ex_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cholesky_ex_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..66b1c819c18cb3a72fcbc66d8041c662e78a80eb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cholesky_ex_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> linalg_cholesky_ex(const at::Tensor & self, bool upper=false, bool check_errors=false);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cholesky_ex_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cholesky_ex_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..54ec0580ffcc8711ceb043492d70e24ea25f5506
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cholesky_ex_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/linalg_cholesky_ex_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_linalg_cholesky_ex_out : public at::meta::structured_linalg_cholesky_ex {
+void impl(const at::Tensor & self, bool upper, bool check_errors, const at::Tensor & L, const at::Tensor & info);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cond.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cond.h
new file mode 100644
index 0000000000000000000000000000000000000000..fa89ed9ac98ad53e8cff4b5415dd7f9060f3d3dc
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cond.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/linalg_cond_ops.h>
+
+namespace at {
+
+
+// aten::linalg_cond(Tensor self, Scalar? p=None) -> Tensor
+inline at::Tensor linalg_cond(const at::Tensor & self, const ::std::optional<at::Scalar> & p=::std::nullopt) {
+    return at::_ops::linalg_cond::call(self, p);
+}
+
+// aten::linalg_cond.out(Tensor self, Scalar? p=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_cond_out(at::Tensor & out, const at::Tensor & self, const ::std::optional<at::Scalar> & p=::std::nullopt) {
+    return at::_ops::linalg_cond_out::call(self, p, out);
+}
+// aten::linalg_cond.out(Tensor self, Scalar? p=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_cond_outf(const at::Tensor & self, const ::std::optional<at::Scalar> & p, at::Tensor & out) {
+    return at::_ops::linalg_cond_out::call(self, p, out);
+}
+
+// aten::linalg_cond.p_str(Tensor self, str p) -> Tensor
+inline at::Tensor linalg_cond(const at::Tensor & self, c10::string_view p) {
+    return at::_ops::linalg_cond_p_str::call(self, p);
+}
+
+// aten::linalg_cond.p_str_out(Tensor self, str p, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_cond_out(at::Tensor & out, const at::Tensor & self, c10::string_view p) {
+    return at::_ops::linalg_cond_p_str_out::call(self, p, out);
+}
+// aten::linalg_cond.p_str_out(Tensor self, str p, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_cond_outf(const at::Tensor & self, c10::string_view p, at::Tensor & out) {
+    return at::_ops::linalg_cond_p_str_out::call(self, p, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cond_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cond_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..86407ae1b861494d214f7a57bd33389ef109436b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_cond_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 linalg_cond {
+  using schema = at::Tensor (const at::Tensor &, const ::std::optional<at::Scalar> &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_cond";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_cond(Tensor self, Scalar? p=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const ::std::optional<at::Scalar> & p);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const ::std::optional<at::Scalar> & p);
+};
+
+struct TORCH_API linalg_cond_out {
+  using schema = at::Tensor & (const at::Tensor &, const ::std::optional<at::Scalar> &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_cond";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "linalg_cond.out(Tensor self, Scalar? p=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const ::std::optional<at::Scalar> & p, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const ::std::optional<at::Scalar> & p, at::Tensor & out);
+};
+
+struct TORCH_API linalg_cond_p_str {
+  using schema = at::Tensor (const at::Tensor &, c10::string_view);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_cond";
+  static constexpr const char* overload_name = "p_str";
+  static constexpr const char* schema_str = "linalg_cond.p_str(Tensor self, str p) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, c10::string_view p);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::string_view p);
+};
+
+struct TORCH_API linalg_cond_p_str_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::string_view, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::linalg_cond";
+  static constexpr const char* overload_name = "p_str_out";
+  static constexpr const char* schema_str = "linalg_cond.p_str_out(Tensor self, str p, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, c10::string_view p, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::string_view p, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_det_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_det_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..40429834cd60437b934d09f7ee2dee38566d6c15
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_det_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_det {
+  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_det";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_det(Tensor A) -> Tensor";
+  static at::Tensor call(const at::Tensor & A);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A);
+};
+
+struct TORCH_API linalg_det_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_det";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "linalg_det.out(Tensor A, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & A, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_eigvalsh.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_eigvalsh.h
new file mode 100644
index 0000000000000000000000000000000000000000..1b3f98b736f9e48ed39ed1740144b50b517b819f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_eigvalsh.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_eigvalsh_ops.h>
+
+namespace at {
+
+
+// aten::linalg_eigvalsh(Tensor self, str UPLO="L") -> Tensor
+inline at::Tensor linalg_eigvalsh(const at::Tensor & self, c10::string_view UPLO="L") {
+    return at::_ops::linalg_eigvalsh::call(self, UPLO);
+}
+
+// aten::linalg_eigvalsh.out(Tensor self, str UPLO="L", *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_eigvalsh_out(at::Tensor & out, const at::Tensor & self, c10::string_view UPLO="L") {
+    return at::_ops::linalg_eigvalsh_out::call(self, UPLO, out);
+}
+// aten::linalg_eigvalsh.out(Tensor self, str UPLO="L", *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & linalg_eigvalsh_outf(const at::Tensor & self, c10::string_view UPLO, at::Tensor & out) {
+    return at::_ops::linalg_eigvalsh_out::call(self, UPLO, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_inv_ex_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_inv_ex_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..e296726ad92ca0571689f47687e0ef8cfeb2d595
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_inv_ex_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/linalg_inv_ex_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_linalg_inv_ex_out : public at::meta::structured_linalg_inv_ex {
+void impl(const at::Tensor & A, bool check_errors, const at::Tensor & inverse, const at::Tensor & info);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_ex_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_ex_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..253e3d0b5039420d2d7905110cc48f9736b037e2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_factor_ex_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_ldl_factor_ex(const at::Tensor & self, bool hermitian=false, bool check_errors=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linalg_ldl_factor_ex_out(at::Tensor & LD, at::Tensor & pivots, at::Tensor & info, const at::Tensor & self, bool hermitian=false, bool check_errors=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> linalg_ldl_factor_ex_outf(const at::Tensor & self, bool hermitian, bool check_errors, at::Tensor & LD, at::Tensor & pivots, at::Tensor & info);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_solve_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_solve_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..baa9e3637ebabee0a2a1f09a22b7400277e7b1ef
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_ldl_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_ldl_solve {
+  using schema = at::Tensor (const 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::linalg_ldl_solve";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "linalg_ldl_solve(Tensor LD, Tensor pivots, Tensor B, *, bool hermitian=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & LD, const at::Tensor & pivots, const at::Tensor & B, bool hermitian);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & LD, const at::Tensor & pivots, const at::Tensor & B, bool hermitian);
+};
+
+struct TORCH_API linalg_ldl_solve_out {
+  using schema = at::Tensor & (const 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::linalg_ldl_solve";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "linalg_ldl_solve.out(Tensor LD, Tensor pivots, Tensor B, *, bool hermitian=False, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & LD, const at::Tensor & pivots, const at::Tensor & B, bool hermitian, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & LD, const at::Tensor & pivots, const at::Tensor & B, bool hermitian, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor.h
new file mode 100644
index 0000000000000000000000000000000000000000..21fd47ec3fbfdfacab86590318972e9cc203873f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor.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_factor_ops.h>
+
+namespace at {
+
+
+// aten::linalg_lu_factor(Tensor A, *, bool pivot=True) -> (Tensor LU, Tensor pivots)
+inline ::std::tuple<at::Tensor,at::Tensor> linalg_lu_factor(const at::Tensor & A, bool pivot=true) {
+    return at::_ops::linalg_lu_factor::call(A, pivot);
+}
+
+// aten::linalg_lu_factor.out(Tensor A, *, bool pivot=True, Tensor(a!) LU, Tensor(b!) pivots) -> (Tensor(a!) LU, Tensor(b!) pivots)
+inline ::std::tuple<at::Tensor &,at::Tensor &> linalg_lu_factor_out(at::Tensor & LU, at::Tensor & pivots, const at::Tensor & A, bool pivot=true) {
+    return at::_ops::linalg_lu_factor_out::call(A, pivot, LU, pivots);
+}
+// aten::linalg_lu_factor.out(Tensor A, *, bool pivot=True, Tensor(a!) LU, Tensor(b!) pivots) -> (Tensor(a!) LU, Tensor(b!) pivots)
+inline ::std::tuple<at::Tensor &,at::Tensor &> linalg_lu_factor_outf(const at::Tensor & A, bool pivot, at::Tensor & LU, at::Tensor & pivots) {
+    return at::_ops::linalg_lu_factor_out::call(A, pivot, LU, pivots);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor_ex_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor_ex_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..06a14b24e644ceea3490a402ece19afeb849de0a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_lu_factor_ex_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/linalg_lu_factor_ex_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_linalg_lu_factor_ex_out : public at::meta::structured_linalg_lu_factor_ex {
+void impl(const at::Tensor & A, bool pivot, bool check_errors, const at::Tensor & LU, const at::Tensor & pivots, const at::Tensor & info);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matmul_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matmul_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8ef384282b37fc358c66b289825fa229a0133d2d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_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 linalg_matmul(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & linalg_matmul_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & linalg_matmul_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_exp_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_exp_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..62c89328627c9a60a68d310aeebfc10b012fcd5b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_exp_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 linalg_matrix_exp(const at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_rank_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_rank_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c17f487df1a2f81280323ba8e5e498de2d79c953
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_matrix_rank_compositeimplicitautograd_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 compositeimplicitautograd {
+
+TORCH_API at::Tensor linalg_matrix_rank(const at::Tensor & input, const ::std::optional<at::Tensor> & atol={}, const ::std::optional<at::Tensor> & rtol={}, bool hermitian=false);
+TORCH_API 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);
+TORCH_API 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);
+TORCH_API at::Tensor linalg_matrix_rank(const at::Tensor & self, ::std::optional<double> atol, ::std::optional<double> rtol, bool hermitian=false);
+TORCH_API at::Tensor & linalg_matrix_rank_out(at::Tensor & out, const at::Tensor & self, ::std::optional<double> atol, ::std::optional<double> rtol, bool hermitian=false);
+TORCH_API at::Tensor & linalg_matrix_rank_outf(const at::Tensor & self, ::std::optional<double> atol, ::std::optional<double> rtol, bool hermitian, at::Tensor & out);
+TORCH_API at::Tensor linalg_matrix_rank(const at::Tensor & self, double tol, bool hermitian=false);
+TORCH_API at::Tensor & linalg_matrix_rank_out(at::Tensor & out, const at::Tensor & self, double tol, bool hermitian=false);
+TORCH_API at::Tensor & linalg_matrix_rank_outf(const at::Tensor & self, double tol, bool hermitian, at::Tensor & out);
+TORCH_API at::Tensor linalg_matrix_rank(const at::Tensor & input, const at::Tensor & tol, bool hermitian=false);
+TORCH_API at::Tensor & linalg_matrix_rank_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & tol, bool hermitian=false);
+TORCH_API at::Tensor & linalg_matrix_rank_outf(const at::Tensor & input, const at::Tensor & tol, bool hermitian, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_slogdet_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_slogdet_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..fee5b0c5dd3c0e4f257b16fcdb171b2e89f7e712
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linalg_slogdet_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_slogdet(const at::Tensor & A);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> linalg_slogdet_out(at::Tensor & sign, at::Tensor & logabsdet, const at::Tensor & A);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> linalg_slogdet_outf(const at::Tensor & A, at::Tensor & sign, at::Tensor & logabsdet);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/linspace_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/linspace_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a1165d7d92872c593482ad21e3e2ddbef3d13e7a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/linspace_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 & linspace_out(at::Tensor & out, const at::Scalar & start, const at::Scalar & end, int64_t steps);
+TORCH_API at::Tensor & linspace_outf(const at::Scalar & start, const at::Scalar & end, int64_t steps, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log10_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log10_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..02ac2d1245e5b88d3cfe16e12e87d856e236a5c1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log10_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 log10(const at::Tensor & self);
+TORCH_API at::Tensor & log10_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & log10_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & log10_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log2_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log2_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..dac676d43968b65c609c16d2fcbebef57789cd1a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log2_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/log2_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_log2_out : public at::meta::structured_log2 {
+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/log_normal_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log_normal_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cbf5c2c6369303eeb15f610c06e3969b4c9a70e2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log_normal_meta_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor & log_normal_(at::Tensor & self, double mean=1, double std=2, ::std::optional<at::Generator> generator=::std::nullopt);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..b000453712e04c06349619a5b64fafe013a651cf
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log_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 log {
+  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::log";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "log(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 log_ {
+  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::log_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "log_(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 log_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::log";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "log.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_forward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_forward.h
new file mode 100644
index 0000000000000000000000000000000000000000..8922210343145da95dd1f3a92fc7abd6183d7640
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/log_sigmoid_forward.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_forward_ops.h>
+
+namespace at {
+
+
+// aten::log_sigmoid_forward.output(Tensor self, *, Tensor(a!) output, Tensor(b!) buffer) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> log_sigmoid_forward_out(at::Tensor & output, at::Tensor & buffer, const at::Tensor & self) {
+    return at::_ops::log_sigmoid_forward_output::call(self, output, buffer);
+}
+// aten::log_sigmoid_forward.output(Tensor self, *, Tensor(a!) output, Tensor(b!) buffer) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> log_sigmoid_forward_outf(const at::Tensor & self, at::Tensor & output, at::Tensor & buffer) {
+    return at::_ops::log_sigmoid_forward_output::call(self, output, buffer);
+}
+
+// aten::log_sigmoid_forward(Tensor self) -> (Tensor output, Tensor buffer)
+inline ::std::tuple<at::Tensor,at::Tensor> log_sigmoid_forward(const at::Tensor & self) {
+    return at::_ops::log_sigmoid_forward::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logaddexp_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logaddexp_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a03fe4e01b22c0214234bb89a3077c92808fda56
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logaddexp_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 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 cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logical_and.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_and.h
new file mode 100644
index 0000000000000000000000000000000000000000..4831fa4b476e93de8b3adbb89b0a233ace4c14f7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logical_and.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/logical_and_ops.h>
+
+namespace at {
+
+
+// aten::logical_and(Tensor self, Tensor other) -> Tensor
+inline at::Tensor logical_and(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::logical_and::call(self, other);
+}
+
+// aten::logical_and.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & logical_and_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::logical_and_out::call(self, other, out);
+}
+// aten::logical_and.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & logical_and_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::logical_and_out::call(self, other, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/logspace_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/logspace_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..cb6999a163d0730130581313688e631966dd1cd2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/logspace_native.h
@@ -0,0 +1,29 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor logspace(const at::Scalar & start, const at::Scalar & end, int64_t steps, double base=10.0, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={});
+TORCH_API at::Tensor & logspace_out(const at::Scalar & start, const at::Scalar & end, int64_t steps, double base, at::Tensor & out);
+TORCH_API at::Tensor & logspace_cuda_out(const at::Scalar & start, const at::Scalar & end, int64_t steps, double base, at::Tensor & out);
+TORCH_API at::Tensor logspace(const at::Tensor & start, const at::Tensor & end, int64_t steps, double base=10.0, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={});
+TORCH_API at::Tensor & logspace_out(const at::Tensor & start, const at::Tensor & end, int64_t steps, double base, at::Tensor & out);
+TORCH_API at::Tensor logspace(const at::Tensor & start, const at::Scalar & end, int64_t steps, double base=10.0, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={});
+TORCH_API at::Tensor & logspace_out(const at::Tensor & start, const at::Scalar & end, int64_t steps, double base, at::Tensor & out);
+TORCH_API at::Tensor logspace(const at::Scalar & start, const at::Tensor & end, int64_t steps, double base=10.0, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={});
+TORCH_API at::Tensor & logspace_out(const at::Scalar & start, const at::Tensor & end, int64_t steps, double base, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7771865c5432478b47eb635c97ea39778ab4a6d9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lshift_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 __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 cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lt_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lt_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..df90696df5931d361757f098ddb861b828ad21a8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lt_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor lt(const at::Tensor & self, const at::Scalar & other);
+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_(at::Tensor & self, const at::Tensor & other);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/lu_unpack_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/lu_unpack_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2e47efdf64994ed15c9a3a65809aebfce2320674
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/lu_unpack_cuda_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> 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 cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/masked_fill_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_fill_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..eb04f83190a57ebe8eb34d584f3648dc2c5830d1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_fill_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 & masked_fill_(at::Tensor & self, const at::Tensor & mask, const at::Scalar & value);
+TORCH_API at::Tensor & masked_fill_(at::Tensor & self, const at::Tensor & mask, const at::Tensor & value);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/masked_fill_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_fill_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..7cbb9f0823f95cfc0f04b08e8a1972c0f8760c44
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_fill_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 masked_fill__Scalar {
+  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::masked_fill_";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "masked_fill_.Scalar(Tensor(a!) self, Tensor mask, Scalar value) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & mask, const at::Scalar & value);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & mask, const at::Scalar & value);
+};
+
+struct TORCH_API masked_fill_Scalar {
+  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::masked_fill";
+  static constexpr const char* overload_name = "Scalar";
+  static constexpr const char* schema_str = "masked_fill.Scalar(Tensor self, Tensor mask, Scalar value) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & mask, const at::Scalar & value);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & mask, const at::Scalar & value);
+};
+
+struct TORCH_API masked_fill__Tensor {
+  using schema = at::Tensor & (at::Tensor &, const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::masked_fill_";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "masked_fill_.Tensor(Tensor(a!) self, Tensor mask, Tensor value) -> Tensor(a!)";
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & mask, const at::Tensor & value);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & mask, const at::Tensor & value);
+};
+
+struct TORCH_API masked_fill_Tensor {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::masked_fill";
+  static constexpr const char* overload_name = "Tensor";
+  static constexpr const char* schema_str = "masked_fill.Tensor(Tensor self, Tensor mask, Tensor value) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & mask, const at::Tensor & value);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & mask, const at::Tensor & value);
+};
+
+struct TORCH_API masked_fill_Scalar_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::masked_fill";
+  static constexpr const char* overload_name = "Scalar_out";
+  static constexpr const char* schema_str = "masked_fill.Scalar_out(Tensor self, Tensor mask, Scalar value, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & mask, const at::Scalar & value, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & mask, const at::Scalar & value, at::Tensor & out);
+};
+
+struct TORCH_API masked_fill_Tensor_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::masked_fill";
+  static constexpr const char* overload_name = "Tensor_out";
+  static constexpr const char* schema_str = "masked_fill.Tensor_out(Tensor self, Tensor mask, Tensor value, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & mask, const at::Tensor & value, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & mask, const at::Tensor & value, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select.h b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select.h
new file mode 100644
index 0000000000000000000000000000000000000000..cec31c278116439ab5be3e7832c993c40632c41c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/masked_select_ops.h>
+
+namespace at {
+
+
+// aten::masked_select.out(Tensor self, Tensor mask, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & masked_select_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & mask) {
+    return at::_ops::masked_select_out::call(self, mask, out);
+}
+// aten::masked_select.out(Tensor self, Tensor mask, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & masked_select_outf(const at::Tensor & self, const at::Tensor & mask, at::Tensor & out) {
+    return at::_ops::masked_select_out::call(self, mask, out);
+}
+
+// aten::masked_select(Tensor self, Tensor mask) -> Tensor
+inline at::Tensor masked_select(const at::Tensor & self, const at::Tensor & mask) {
+    return at::_ops::masked_select::call(self, mask);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..cc97d901e679b75ff979c0b1f6b4710c33a29c91
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/masked_select_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 masked_select(const at::Tensor & self, const at::Tensor & mask);
+TORCH_API at::Tensor & masked_select_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & mask);
+TORCH_API at::Tensor & masked_select_outf(const at::Tensor & self, const at::Tensor & mask, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..54cf78c85c873f3289c6ae1131aba0425753ec34
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_backward.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/max_pool2d_backward_ops.h>
+
+namespace at {
+
+
+// aten::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
+inline at::Tensor max_pool2d_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, bool ceil_mode=false) {
+    return at::_ops::max_pool2d_backward::call(grad_output, self, kernel_size, stride, padding, dilation, ceil_mode);
+}
+
+// aten::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!)
+inline at::Tensor & max_pool2d_backward_out(at::Tensor & out, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, bool ceil_mode=false) {
+    return at::_ops::max_pool2d_backward_out::call(grad_output, self, kernel_size, stride, padding, dilation, ceil_mode, out);
+}
+// aten::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!)
+inline at::Tensor & max_pool2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out) {
+    return at::_ops::max_pool2d_backward_out::call(grad_output, self, kernel_size, stride, padding, dilation, ceil_mode, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_with_indices_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_with_indices_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..5382df32fbd0cd3382d68979d2d164b5241f50e2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool2d_with_indices_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/max_pool2d_with_indices_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_max_pool2d_with_indices_out_cpu : public at::meta::structured_max_pool2d_with_indices {
+void impl(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & out, const at::Tensor & indices);
+};
+struct TORCH_API structured_max_pool2d_with_indices_out_cuda : public at::meta::structured_max_pool2d_with_indices {
+void impl(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, const at::Tensor & out, const at::Tensor & indices);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..8d92a60f691618553a7f73662ca0688abe8051c3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/max_pool3d_with_indices_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API max_pool3d_with_indices_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::max_pool3d_with_indices";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "max_pool3d_with_indices.out(Tensor self, int[3] kernel_size, int[3] stride=[], int[3] padding=0, int[3] dilation=1, bool ceil_mode=False, *, Tensor(a!) out, Tensor(b!) indices) -> (Tensor(a!), Tensor(b!))";
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out, at::Tensor & indices);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out, at::Tensor & indices);
+};
+
+struct TORCH_API max_pool3d_with_indices {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::max_pool3d_with_indices";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "max_pool3d_with_indices(Tensor self, int[3] kernel_size, int[3] stride=[], int[3] padding=0, int[3] dilation=1, bool ceil_mode=False) -> (Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mean_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mean_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3147afdb48cce3d5c3aed4f780273bcb3d3e11dd
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mean_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 mean(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/median_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/median_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..062ae20866c71d27358f8fc6acc46e7a1098d616
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/median_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 & median_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & median_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> median(const at::Tensor & self, int64_t dim, bool keepdim=false);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/min.h b/lib/python3.12/site-packages/torch/include/ATen/ops/min.h
new file mode 100644
index 0000000000000000000000000000000000000000..d2baca13949bf6fd029225c54d66a4a79f72f88f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/min.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/min_ops.h>
+
+namespace at {
+
+
+// aten::min.dim(Tensor self, int dim, bool keepdim=False) -> (Tensor values, Tensor indices)
+inline ::std::tuple<at::Tensor,at::Tensor> min(const at::Tensor & self, int64_t dim, bool keepdim=false) {
+    return at::_ops::min_dim::call(self, dim, keepdim);
+}
+
+// aten::min.dim_min(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) min, Tensor(b!) min_indices) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::std::tuple<at::Tensor &,at::Tensor &> min_out(at::Tensor & min, at::Tensor & min_indices, const at::Tensor & self, int64_t dim, bool keepdim=false) {
+    return at::_ops::min_dim_min::call(self, dim, keepdim, min, min_indices);
+}
+// aten::min.dim_min(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) min, Tensor(b!) min_indices) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::std::tuple<at::Tensor &,at::Tensor &> min_outf(const at::Tensor & self, int64_t dim, bool keepdim, at::Tensor & min, at::Tensor & min_indices) {
+    return at::_ops::min_dim_min::call(self, dim, keepdim, min, min_indices);
+}
+
+// aten::min.names_dim(Tensor self, Dimname dim, bool keepdim=False) -> (Tensor values, Tensor indices)
+inline ::std::tuple<at::Tensor,at::Tensor> min(const at::Tensor & self, at::Dimname dim, bool keepdim=false) {
+    return at::_ops::min_names_dim::call(self, dim, keepdim);
+}
+
+// aten::min.names_dim_min(Tensor self, Dimname dim, bool keepdim=False, *, Tensor(a!) min, Tensor(b!) min_indices) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::std::tuple<at::Tensor &,at::Tensor &> min_out(at::Tensor & min, at::Tensor & min_indices, const at::Tensor & self, at::Dimname dim, bool keepdim=false) {
+    return at::_ops::min_names_dim_min::call(self, dim, keepdim, min, min_indices);
+}
+// aten::min.names_dim_min(Tensor self, Dimname dim, bool keepdim=False, *, Tensor(a!) min, Tensor(b!) min_indices) -> (Tensor(a!) values, Tensor(b!) indices)
+inline ::std::tuple<at::Tensor &,at::Tensor &> min_outf(const at::Tensor & self, at::Dimname dim, bool keepdim, at::Tensor & min, at::Tensor & min_indices) {
+    return at::_ops::min_names_dim_min::call(self, dim, keepdim, min, min_indices);
+}
+
+// aten::min(Tensor self) -> Tensor
+inline at::Tensor min(const at::Tensor & self) {
+    return at::_ops::min::call(self);
+}
+
+// aten::min.unary_out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & min_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::min_unary_out::call(self, out);
+}
+// aten::min.unary_out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & min_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::min_unary_out::call(self, out);
+}
+
+// aten::min.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & min_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::min_out::call(self, other, out);
+}
+// aten::min.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & min_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::min_out::call(self, other, out);
+}
+
+// aten::min.other(Tensor self, Tensor other) -> Tensor
+inline at::Tensor min(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::min_other::call(self, other);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/minimum.h b/lib/python3.12/site-packages/torch/include/ATen/ops/minimum.h
new file mode 100644
index 0000000000000000000000000000000000000000..aa40802df29f1cb8c90a354bff289c4b6c02150b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/minimum.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/minimum_ops.h>
+
+namespace at {
+
+
+// aten::minimum(Tensor self, Tensor other) -> Tensor
+inline at::Tensor minimum(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::minimum::call(self, other);
+}
+
+// aten::minimum.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & minimum_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::minimum_out::call(self, other, out);
+}
+// aten::minimum.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & minimum_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::minimum_out::call(self, other, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_batch_norm_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_batch_norm_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..77a27ead3adf1657ee8ca09a171c6408f1e29a40
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_batch_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 &> miopen_batch_norm_backward_out(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, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> miopen_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);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution.h b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution.h
new file mode 100644
index 0000000000000000000000000000000000000000..45e1e64f9a1aa1b58df0982914466f2fb825f003
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/miopen_convolution_ops.h>
+
+namespace at {
+
+
+// aten::miopen_convolution(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic) -> Tensor
+inline at::Tensor miopen_convolution(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic) {
+    return at::_ops::miopen_convolution::call(self, weight, bias, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor miopen_convolution(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic) {
+    return at::_ops::miopen_convolution::call(self, weight, bias, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic);
+  }
+}
+
+// aten::miopen_convolution(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic) -> Tensor
+inline at::Tensor miopen_convolution_symint(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic) {
+    return at::_ops::miopen_convolution::call(self, weight, bias, padding, stride, dilation, groups, benchmark, deterministic);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor miopen_convolution(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic) {
+    return at::_ops::miopen_convolution::call(self, weight, bias, padding, stride, dilation, groups, benchmark, deterministic);
+  }
+}
+
+// aten::miopen_convolution.out(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & miopen_convolution_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic) {
+    return at::_ops::miopen_convolution_out::call(self, weight, bias, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & miopen_convolution_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic) {
+    return at::_ops::miopen_convolution_out::call(self, weight, bias, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic, out);
+  }
+}
+
+// aten::miopen_convolution.out(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & miopen_convolution_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, at::Tensor & out) {
+    return at::_ops::miopen_convolution_out::call(self, weight, bias, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & miopen_convolution_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, bool benchmark, bool deterministic, at::Tensor & out) {
+    return at::_ops::miopen_convolution_out::call(self, weight, bias, c10::fromIntArrayRefSlow(padding), c10::fromIntArrayRefSlow(stride), c10::fromIntArrayRefSlow(dilation), groups, benchmark, deterministic, out);
+  }
+}
+
+// aten::miopen_convolution.out(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & miopen_convolution_symint_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic) {
+    return at::_ops::miopen_convolution_out::call(self, weight, bias, padding, stride, dilation, groups, benchmark, deterministic, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & miopen_convolution_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic) {
+    return at::_ops::miopen_convolution_out::call(self, weight, bias, padding, stride, dilation, groups, benchmark, deterministic, out);
+  }
+}
+
+// aten::miopen_convolution.out(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & miopen_convolution_symint_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, at::Tensor & out) {
+    return at::_ops::miopen_convolution_out::call(self, weight, bias, padding, stride, dilation, groups, benchmark, deterministic, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & miopen_convolution_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, at::Tensor & out) {
+    return at::_ops::miopen_convolution_out::call(self, weight, bias, padding, stride, dilation, groups, benchmark, deterministic, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_transpose_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_transpose_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8b272d5e8b188094d19660f7e75b41c6e3f8000c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/miopen_convolution_transpose_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 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);
+TORCH_API at::Tensor miopen_convolution_transpose_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);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mish_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mish_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..b4f26e2df49fabcf0960a663b7c498e2a5f477c4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mish_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/mish_backward_ops.h>
+
+namespace at {
+
+
+// aten::mish_backward(Tensor grad_output, Tensor self) -> Tensor
+inline at::Tensor mish_backward(const at::Tensor & grad_output, const at::Tensor & self) {
+    return at::_ops::mish_backward::call(grad_output, self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_adaptive_avg_pool2d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_adaptive_avg_pool2d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..17c978b9216d141e37778da5babe04ddd9f679af
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_adaptive_avg_pool2d_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_adaptive_avg_pool2d(const at::Tensor & self, at::IntArrayRef output_size);
+TORCH_API at::Tensor & mkldnn_adaptive_avg_pool2d_out(const at::Tensor & self, at::IntArrayRef output_size, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..883ef801acd013fe404cb423274505fd9d2b94c7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_linear_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 {
+  using schema = at::Tensor (const at::Tensor &, const 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::mkldnn_linear";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "mkldnn_linear(Tensor self, Tensor weight, Tensor? bias=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias);
+};
+
+struct TORCH_API mkldnn_linear_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mkldnn_linear";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "mkldnn_linear.out(Tensor self, Tensor weight, Tensor? bias=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..6fb21b7d44042223382da7bf8f22a301d7b599a6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API mkldnn_max_pool2d {
+  using schema = at::Tensor (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mkldnn_max_pool2d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "mkldnn_max_pool2d(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+};
+
+struct TORCH_API mkldnn_max_pool2d_out {
+  using schema = at::Tensor & (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mkldnn_max_pool2d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "mkldnn_max_pool2d.out(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_reorder_conv3d_weight_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_reorder_conv3d_weight_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..9c858eca6ace854623bd91c65b4115224d551501
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mkldnn_reorder_conv3d_weight_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_reorder_conv3d_weight {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymInt, at::OptionalSymIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mkldnn_reorder_conv3d_weight";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "mkldnn_reorder_conv3d_weight(Tensor self, SymInt[3] padding=0, SymInt[3] stride=1, SymInt[3] dilation=1, SymInt groups=1, SymInt[]? input_size=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, at::OptionalSymIntArrayRef input_size);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, at::OptionalSymIntArrayRef input_size);
+};
+
+struct TORCH_API mkldnn_reorder_conv3d_weight_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymInt, at::OptionalSymIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::mkldnn_reorder_conv3d_weight";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "mkldnn_reorder_conv3d_weight.out(Tensor self, SymInt[3] padding=0, SymInt[3] stride=1, SymInt[3] dilation=1, SymInt groups=1, SymInt[]? input_size=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, at::OptionalSymIntArrayRef input_size, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, at::OptionalSymIntArrayRef input_size, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mode_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mode_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..c61208a6c1a0e49e67e80df8d2a0fae266bc08f1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mode_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 mode {
+  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::mode";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "mode(Tensor self, int dim=-1, 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 mode_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::mode";
+  static constexpr const char* overload_name = "values";
+  static constexpr const char* schema_str = "mode.values(Tensor self, int dim=-1, 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 mode_dimname {
+  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::mode";
+  static constexpr const char* overload_name = "dimname";
+  static constexpr const char* schema_str = "mode.dimname(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 mode_dimname_out {
+  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::mode";
+  static constexpr const char* overload_name = "dimname_out";
+  static constexpr const char* schema_str = "mode.dimname_out(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);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/movedim_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/movedim_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0543008089c86b9cc6fd7da8cf8fa41e22b0ceab
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/movedim_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 movedim(const at::Tensor & self, at::IntArrayRef source, at::IntArrayRef destination);
+TORCH_API at::Tensor movedim(const at::Tensor & self, int64_t source, int64_t destination);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/mse_loss_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/mse_loss_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..3d96947d3b616aab3c0a017f3d98ee756c737e3b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/mse_loss_backward.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/mse_loss_backward_ops.h>
+
+namespace at {
+
+
+// aten::mse_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, int reduction, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & mse_loss_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction) {
+    return at::_ops::mse_loss_backward_grad_input::call(grad_output, self, target, reduction, grad_input);
+}
+// aten::mse_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, int reduction, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & mse_loss_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction, at::Tensor & grad_input) {
+    return at::_ops::mse_loss_backward_grad_input::call(grad_output, self, target, reduction, grad_input);
+}
+
+// aten::mse_loss_backward(Tensor grad_output, Tensor self, Tensor target, int reduction) -> Tensor
+inline at::Tensor mse_loss_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, int64_t reduction) {
+    return at::_ops::mse_loss_backward::call(grad_output, self, target, reduction);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_batch_norm_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_batch_norm_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..f182453e058b2219e026fa6ddf4891040b2f2a59
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_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/native_batch_norm_backward_ops.h>
+
+namespace at {
+
+
+// aten::native_batch_norm_backward(Tensor grad_out, Tensor input, Tensor? weight, Tensor? running_mean, Tensor? running_var, Tensor? save_mean, Tensor? save_invstd, bool train, float eps, bool[3] output_mask) -> (Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> native_batch_norm_backward(const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask) {
+    return at::_ops::native_batch_norm_backward::call(grad_out, input, weight, running_mean, running_var, save_mean, save_invstd, train, eps, output_mask);
+}
+
+// aten::native_batch_norm_backward.out(Tensor grad_out, Tensor input, Tensor? weight, Tensor? running_mean, Tensor? running_var, Tensor? save_mean, Tensor? save_invstd, bool train, float eps, bool[3] output_mask, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_batch_norm_backward_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask) {
+    return at::_ops::native_batch_norm_backward_out::call(grad_out, input, weight, running_mean, running_var, save_mean, save_invstd, train, eps, output_mask, out0, out1, out2);
+}
+// aten::native_batch_norm_backward.out(Tensor grad_out, Tensor input, Tensor? weight, Tensor? running_mean, Tensor? running_var, Tensor? save_mean, Tensor? save_invstd, bool train, float eps, bool[3] output_mask, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_batch_norm_backward_outf(const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2) {
+    return at::_ops::native_batch_norm_backward_out::call(grad_out, input, weight, running_mean, running_var, save_mean, save_invstd, train, eps, output_mask, out0, out1, out2);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_batch_norm_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_batch_norm_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..71cdf55b97eb8094ce793d4eb1501c700c185622
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_batch_norm_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 ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_batch_norm_backward_out(const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> batch_norm_backward_cpu(const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> batch_norm_backward_cuda(const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> mkldnn_batch_norm_backward(const at::Tensor & grad_out, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_invstd, bool train, double eps, ::std::array<bool,3> output_mask);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_dropout_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_dropout_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..59b64db4884eb81d8252a6784e35c862ec2a6f99
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_dropout_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 &> native_dropout_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & input, double p, ::std::optional<bool> train);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> native_dropout_outf(const at::Tensor & input, double p, ::std::optional<bool> train, at::Tensor & out0, at::Tensor & out1);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_dropout_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_dropout_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..662c8edf857f5c3c8f986c1a9565af36300da2e6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_dropout_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> native_dropout(const at::Tensor & input, double p, ::std::optional<bool> train);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/native_layer_norm_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/native_layer_norm_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..91dd15361146750e6a8e14f318f49c0324ab9731
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/native_layer_norm_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 ::std::tuple<at::Tensor,at::Tensor,at::Tensor> native_layer_norm_backward(const at::Tensor & grad_out, const at::Tensor & input, at::IntArrayRef normalized_shape, const at::Tensor & mean, const at::Tensor & rstd, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, ::std::array<bool,3> output_mask);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> native_layer_norm_backward_symint(const at::Tensor & grad_out, const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const at::Tensor & mean, const at::Tensor & rstd, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, ::std::array<bool,3> output_mask);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/neg_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/neg_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..690d24220b768a96ea9c14035037407b5bcbb617
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/neg_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 neg(const at::Tensor & self);
+TORCH_API at::Tensor & neg_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & neg_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & neg_(at::Tensor & self);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/new_empty_strided_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/new_empty_strided_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a8975973d1e67693f2081f617aa37cbd259ba640
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/new_empty_strided_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API new_empty_strided {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, c10::SymIntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::new_empty_strided";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "new_empty_strided(Tensor self, SymInt[] size, SymInt[] stride, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+struct TORCH_API new_empty_strided_out {
+  using schema = at::Tensor & (const 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::new_empty_strided";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "new_empty_strided.out(Tensor self, SymInt[] size, SymInt[] stride, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef size, c10::SymIntArrayRef stride, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nextafter.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nextafter.h
new file mode 100644
index 0000000000000000000000000000000000000000..4a2808815918f37117a0902b1fb2dd303e5e2ca7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nextafter.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/nextafter_ops.h>
+
+namespace at {
+
+
+// aten::nextafter.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & nextafter_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::nextafter_out::call(self, other, out);
+}
+// aten::nextafter.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & nextafter_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::nextafter_out::call(self, other, out);
+}
+
+// aten::nextafter(Tensor self, Tensor other) -> Tensor
+inline at::Tensor nextafter(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::nextafter::call(self, other);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_forward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_forward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5eea254064db4a8d7a2be0a6f53da04d91beccd4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_forward_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::tuple<at::Tensor,at::Tensor> nll_loss_forward(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> nll_loss_forward_symint(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> nll_loss_forward_out(at::Tensor & output, at::Tensor & total_weight, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> nll_loss_forward_outf(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, at::Tensor & output, at::Tensor & total_weight);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> nll_loss_forward_symint_out(at::Tensor & output, at::Tensor & total_weight, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> nll_loss_forward_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 & output, at::Tensor & total_weight);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_forward_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_forward_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..89751f639b68adc39168c81cc0a01d46d9dc9581
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nll_loss_forward_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 ::std::tuple<at::Tensor,at::Tensor> nll_loss_forward(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> nll_loss_forward_symint(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> nll_loss_forward_out(at::Tensor & output, at::Tensor & total_weight, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> nll_loss_forward_outf(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, int64_t ignore_index, at::Tensor & output, at::Tensor & total_weight);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> nll_loss_forward_symint_out(at::Tensor & output, at::Tensor & total_weight, const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, c10::SymInt ignore_index);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> nll_loss_forward_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 & output, at::Tensor & total_weight);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/norm_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/norm_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..fa9e054b024dd06798871c43c726486072f64958
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/norm_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_norm_ScalarOpt_dim_dtype : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, at::OptionalScalarRef p, at::IntArrayRef dim, bool keepdim, at::ScalarType dtype);
+};
+struct TORCH_API structured_norm_ScalarOpt_dim : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, at::OptionalScalarRef p, at::IntArrayRef dim, bool keepdim);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/normal_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/normal_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..070aed7ceb156af912ad6a3c68d0c8f150f1c472
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/normal_cuda_dispatch.h
@@ -0,0 +1,32 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause 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 & normal_(at::Tensor & self, double mean=0, double std=1, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor normal(const at::Tensor & mean, double std=1, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor & normal_out(at::Tensor & out, const at::Tensor & mean, double std=1, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor & normal_outf(const at::Tensor & mean, double std, ::std::optional<at::Generator> generator, at::Tensor & out);
+TORCH_API at::Tensor normal(double mean, const at::Tensor & std, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor & normal_out(at::Tensor & out, double mean, const at::Tensor & std, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor & normal_outf(double mean, const at::Tensor & std, ::std::optional<at::Generator> generator, at::Tensor & out);
+TORCH_API at::Tensor normal(const at::Tensor & mean, const at::Tensor & std, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor & normal_out(at::Tensor & out, const at::Tensor & mean, const at::Tensor & std, ::std::optional<at::Generator> generator=::std::nullopt);
+TORCH_API at::Tensor & normal_outf(const at::Tensor & mean, const at::Tensor & std, ::std::optional<at::Generator> generator, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/nuclear_norm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/nuclear_norm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..09ec588d46d0a3173bf14cdeca59ff6cdafe7894
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/nuclear_norm_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor nuclear_norm(const at::Tensor & self, bool keepdim=false);
+TORCH_API at::Tensor & nuclear_norm_out(const at::Tensor & self, bool keepdim, at::Tensor & out);
+TORCH_API at::Tensor nuclear_norm(const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false);
+TORCH_API at::Tensor & nuclear_norm_out(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/numpy_T_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/numpy_T_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..13937a52664c4c391413686fac2b03607f3bfd08
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/numpy_T_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 numpy_T {
+  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::numpy_T";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "numpy_T(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/orgqr_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/orgqr_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ce06d57632bcd494ad3d410680059e6379e7f7c8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/orgqr_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 orgqr(const at::Tensor & self, const at::Tensor & input2);
+TORCH_API at::Tensor & orgqr_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & input2);
+TORCH_API at::Tensor & orgqr_outf(const at::Tensor & self, const at::Tensor & input2, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/ormqr_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/ormqr_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8a4c26e15c2ef55f17b8fd3a02ebdce5cbc4b075
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/ormqr_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 ormqr(const at::Tensor & self, const at::Tensor & input2, const at::Tensor & input3, bool left=true, bool transpose=false);
+TORCH_API at::Tensor & ormqr_out(const at::Tensor & self, const at::Tensor & input2, const at::Tensor & input3, bool left, bool transpose, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/ormqr_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/ormqr_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..30ba1a96b612302dec732706f4143e791da70dbb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/ormqr_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 ormqr_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::ormqr";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "ormqr.out(Tensor self, Tensor input2, Tensor input3, bool left=True, bool transpose=False, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & input2, const at::Tensor & input3, bool left, bool transpose, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & input2, const at::Tensor & input3, bool left, bool transpose, at::Tensor & out);
+};
+
+struct TORCH_API ormqr {
+  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::ormqr";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "ormqr(Tensor self, Tensor input2, Tensor input3, bool left=True, bool transpose=False) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & input2, const at::Tensor & input3, bool left, bool transpose);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & input2, const at::Tensor & input3, bool left, bool transpose);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/pinverse.h b/lib/python3.12/site-packages/torch/include/ATen/ops/pinverse.h
new file mode 100644
index 0000000000000000000000000000000000000000..e60488d2e4417ff5dd719a4132ed3918b3ae4e21
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/pinverse.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/pinverse_ops.h>
+
+namespace at {
+
+
+// aten::pinverse(Tensor self, float rcond=1e-15) -> Tensor
+inline at::Tensor pinverse(const at::Tensor & self, double rcond=1e-15) {
+    return at::_ops::pinverse::call(self, rcond);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/pinverse_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/pinverse_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..060b2acacc81632c4d400b8c39759d5f06dc29b2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/pinverse_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 pinverse(const at::Tensor & self, double rcond=1e-15);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/pixel_unshuffle_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/pixel_unshuffle_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a041121a9f76a6992f97336d9af18776276b0bfe
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/pixel_unshuffle_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API pixel_unshuffle {
+  using schema = at::Tensor (const at::Tensor &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::pixel_unshuffle";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "pixel_unshuffle(Tensor self, int downscale_factor) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, int64_t downscale_factor);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t downscale_factor);
+};
+
+struct TORCH_API pixel_unshuffle_out {
+  using schema = at::Tensor & (const at::Tensor &, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::pixel_unshuffle";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "pixel_unshuffle.out(Tensor self, int downscale_factor, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, int64_t downscale_factor, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t downscale_factor, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/poisson_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/poisson_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..2540c65ff0651a420cf5e0729e25c9c698a52ab7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/poisson_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 poisson {
+  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::poisson";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "poisson(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 poisson_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::poisson";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "poisson.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/positive_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/positive_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..5b498f38b4463ad20d153d185305549eadcc4e01
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/positive_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 positive {
+  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::positive";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "positive(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/prelu_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/prelu_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8d7b5038810e8d7455f05c849ec319853213ff4e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/prelu_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 prelu(const at::Tensor & self, const at::Tensor & weight);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/q_per_channel_scales.h b/lib/python3.12/site-packages/torch/include/ATen/ops/q_per_channel_scales.h
new file mode 100644
index 0000000000000000000000000000000000000000..f62f99a8b93807cdd6fb1efd06d16eed146d1943
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/q_per_channel_scales.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/q_per_channel_scales_ops.h>
+
+namespace at {
+
+
+// aten::q_per_channel_scales(Tensor self) -> Tensor
+inline at::Tensor q_per_channel_scales(const at::Tensor & self) {
+    return at::_ops::q_per_channel_scales::call(self);
+}
+
+// aten::q_per_channel_scales.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & q_per_channel_scales_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::q_per_channel_scales_out::call(self, out);
+}
+// aten::q_per_channel_scales.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & q_per_channel_scales_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::q_per_channel_scales_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/qscheme_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/qscheme_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..0c9a1a6c6925f25338380f13829008150290473f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/qscheme_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 qscheme {
+  using schema = at::QScheme (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::qscheme";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "qscheme(Tensor self) -> QScheme";
+  static at::QScheme call(const at::Tensor & self);
+  static at::QScheme redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_max_pool1d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_max_pool1d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..d537b59bf7f3455659542d6bb3e33ee614e47a12
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/quantized_max_pool1d_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 & quantized_max_pool1d_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 quantized_max_pool1d(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/real_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/real_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1171869e1f0759a2e5712c1b2d5270b959115ec8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/real_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 real(const at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/reciprocal_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/reciprocal_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..7ab826ff00a29fce9e4873d896e5e9f9fbcf32aa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/reciprocal_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_reciprocal : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/refine_names_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/refine_names_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6de924f2068befcfc9115189dfc0de17beb1c95b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/refine_names_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 refine_names(const at::Tensor & self, at::DimnameList names);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad1d_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad1d_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..b2dbbed6a7f938144d2b96331ceb7faeb56b61db
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad1d_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_pad1d : 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/reflection_pad1d_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad1d_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c1e82cf2c091e69f25139f93360f63b36c911bd3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad1d_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 reflection_pad1d(const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor reflection_pad1d_symint(const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & reflection_pad1d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor & reflection_pad1d_outf(const at::Tensor & self, at::IntArrayRef padding, at::Tensor & out);
+TORCH_API at::Tensor & reflection_pad1d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & reflection_pad1d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad1d_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad1d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..aedab64e06f08b694bf624498934b46f7a78c540
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/reflection_pad1d_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/reflection_pad1d_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_reflection_pad1d_out_cpu : public at::meta::structured_reflection_pad1d {
+void impl(const at::Tensor & self, at::ArrayRef<int64_t> padding, const at::Tensor & out);
+};
+struct TORCH_API structured_reflection_pad1d_out_cuda : public at::meta::structured_reflection_pad1d {
+void impl(const at::Tensor & self, at::ArrayRef<int64_t> padding, const at::Tensor & out);
+};
+TORCH_API at::Tensor & reflection_pad1d_out_quantized_cpu(const at::Tensor & self, at::IntArrayRef padding, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/relu_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/relu_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1ca317e8cefce71bbf743637833dde0d68f5568a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/relu_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 relu(const at::Tensor & self);
+TORCH_API at::Tensor & relu_(at::Tensor & self);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/remainder.h b/lib/python3.12/site-packages/torch/include/ATen/ops/remainder.h
new file mode 100644
index 0000000000000000000000000000000000000000..220563e784692e50b3fbc98a1c0c2b3f5e7025b5
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/remainder.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/remainder_ops.h>
+
+namespace at {
+
+
+// aten::remainder.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & remainder_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::remainder_Scalar_out::call(self, other, out);
+}
+// aten::remainder.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & remainder_outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out) {
+    return at::_ops::remainder_Scalar_out::call(self, other, out);
+}
+
+// aten::remainder.Scalar(Tensor self, Scalar other) -> Tensor
+inline at::Tensor remainder(const at::Tensor & self, const at::Scalar & other) {
+    return at::_ops::remainder_Scalar::call(self, other);
+}
+
+// aten::remainder.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & remainder_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::remainder_Tensor_out::call(self, other, out);
+}
+// aten::remainder.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & remainder_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::remainder_Tensor_out::call(self, other, out);
+}
+
+// aten::remainder.Tensor(Tensor self, Tensor other) -> Tensor
+inline at::Tensor remainder(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::remainder_Tensor::call(self, other);
+}
+
+// aten::remainder.Scalar_Tensor(Scalar self, Tensor other) -> Tensor
+inline at::Tensor remainder(const at::Scalar & self, const at::Tensor & other) {
+    return at::_ops::remainder_Scalar_Tensor::call(self, other);
+}
+
+// aten::remainder.Scalar_Tensor_out(Scalar self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & remainder_out(at::Tensor & out, const at::Scalar & self, const at::Tensor & other) {
+    return at::_ops::remainder_Scalar_Tensor_out::call(self, other, out);
+}
+// aten::remainder.Scalar_Tensor_out(Scalar self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & remainder_outf(const at::Scalar & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::remainder_Scalar_Tensor_out::call(self, other, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rename_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rename_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..cf87128a3c686722d773d13756c7c4b2c90f1c8b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rename_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 & rename_(at::Tensor & self, ::std::optional<at::DimnameList> names);
+TORCH_API at::Tensor rename(const at::Tensor & self, ::std::optional<at::DimnameList> names);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/renorm.h b/lib/python3.12/site-packages/torch/include/ATen/ops/renorm.h
new file mode 100644
index 0000000000000000000000000000000000000000..d2acc6985b8c831af097167add95472b530f1dad
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/renorm.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/renorm_ops.h>
+
+namespace at {
+
+
+// aten::renorm.out(Tensor self, Scalar p, int dim, Scalar maxnorm, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & renorm_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & p, int64_t dim, const at::Scalar & maxnorm) {
+    return at::_ops::renorm_out::call(self, p, dim, maxnorm, out);
+}
+// aten::renorm.out(Tensor self, Scalar p, int dim, Scalar maxnorm, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & renorm_outf(const at::Tensor & self, const at::Scalar & p, int64_t dim, const at::Scalar & maxnorm, at::Tensor & out) {
+    return at::_ops::renorm_out::call(self, p, dim, maxnorm, out);
+}
+
+// aten::renorm(Tensor self, Scalar p, int dim, Scalar maxnorm) -> Tensor
+inline at::Tensor renorm(const at::Tensor & self, const at::Scalar & p, int64_t dim, const at::Scalar & maxnorm) {
+    return at::_ops::renorm::call(self, p, dim, maxnorm);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad1d_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad1d_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..673128eaadfa25260763c9b057158eb84ddf02e1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad1d_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_replication_pad1d : 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/replication_pad2d_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad2d_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4465aa774fce171038b5677c963def97ae61a5bf
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/replication_pad2d_cuda_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor replication_pad2d(const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor replication_pad2d_symint(const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & replication_pad2d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef padding);
+TORCH_API at::Tensor & replication_pad2d_outf(const at::Tensor & self, at::IntArrayRef padding, at::Tensor & out);
+TORCH_API at::Tensor & replication_pad2d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef padding);
+TORCH_API at::Tensor & replication_pad2d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/resize.h b/lib/python3.12/site-packages/torch/include/ATen/ops/resize.h
new file mode 100644
index 0000000000000000000000000000000000000000..718316fd87b011aafbb7ff14a9c2d7b10e5fd652
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/resize.h
@@ -0,0 +1,105 @@
+#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/resize_ops.h>
+
+namespace at {
+
+
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  const at::Tensor & resize_(const at::Tensor & self, at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::resize_::call(self, c10::fromIntArrayRefSlow(size), memory_format);
+  }
+}
+
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  const at::Tensor & resize_(const at::Tensor & self, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::resize_::call(self, size, memory_format);
+  }
+}
+
+// aten::resize.out(Tensor self, SymInt[] size, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline const at::Tensor & resize_out(const at::Tensor & out, const at::Tensor & self, at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::resize_out::call(self, c10::fromIntArrayRefSlow(size), memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  const at::Tensor & resize_out(const at::Tensor & out, const at::Tensor & self, at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::resize_out::call(self, c10::fromIntArrayRefSlow(size), memory_format, out);
+  }
+}
+
+// aten::resize.out(Tensor self, SymInt[] size, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline const at::Tensor & resize_outf(const at::Tensor & self, at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, const at::Tensor & out) {
+    return at::_ops::resize_out::call(self, c10::fromIntArrayRefSlow(size), memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  const at::Tensor & resize_outf(const at::Tensor & self, at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, const at::Tensor & out) {
+    return at::_ops::resize_out::call(self, c10::fromIntArrayRefSlow(size), memory_format, out);
+  }
+}
+
+// aten::resize.out(Tensor self, SymInt[] size, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline const at::Tensor & resize_symint_out(const at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::resize_out::call(self, size, memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  const at::Tensor & resize_out(const at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::resize_out::call(self, size, memory_format, out);
+  }
+}
+
+// aten::resize.out(Tensor self, SymInt[] size, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)
+inline const at::Tensor & resize_symint_outf(const at::Tensor & self, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, const at::Tensor & out) {
+    return at::_ops::resize_out::call(self, size, memory_format, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  const at::Tensor & resize_outf(const at::Tensor & self, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format, const at::Tensor & out) {
+    return at::_ops::resize_out::call(self, size, memory_format, out);
+  }
+}
+
+// aten::resize(Tensor self, SymInt[] size, *, MemoryFormat? memory_format=None) -> Tensor
+inline at::Tensor resize(const at::Tensor & self, at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::resize::call(self, c10::fromIntArrayRefSlow(size), memory_format);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor resize(const at::Tensor & self, at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::resize::call(self, c10::fromIntArrayRefSlow(size), memory_format);
+  }
+}
+
+// aten::resize(Tensor self, SymInt[] size, *, MemoryFormat? memory_format=None) -> Tensor
+inline at::Tensor resize_symint(const at::Tensor & self, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::resize::call(self, size, memory_format);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor resize(const at::Tensor & self, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt) {
+    return at::_ops::resize::call(self, size, memory_format);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/resize_as_sparse_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/resize_as_sparse_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f9e6b7698d647d2a5fbd4e3b7784759429b5e4e0
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/resize_as_sparse_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 & resize_as_sparse_(const at::Tensor & self, const at::Tensor & the_template);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/resize_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/resize_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..6164659f754b3278bbf2ea1aa70e019fc8a9ef96
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/resize_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 const at::Tensor & resize_(const at::Tensor & self, at::IntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API const at::Tensor & resize__symint(const at::Tensor & self, c10::SymIntArrayRef size, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/resolve_conj_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/resolve_conj_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..970c63965615b0d18fff1cc94330ded0dec60a9c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/resolve_conj_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 resolve_conj {
+  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::resolve_conj";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "resolve_conj(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/resolve_neg_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/resolve_neg_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8e591b1bfc3b863c69b48e2b08b3c4187ec73996
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/resolve_neg_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 resolve_neg(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/retains_grad.h b/lib/python3.12/site-packages/torch/include/ATen/ops/retains_grad.h
new file mode 100644
index 0000000000000000000000000000000000000000..f3b565973c14b8f2d591eb349ce8c435a64683c9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/retains_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/retains_grad_ops.h>
+
+namespace at {
+
+
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rms_norm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rms_norm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..c3e5bddbbb558838d8ef739a390a157835ed970f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rms_norm_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 rms_norm_symint(const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight={}, ::std::optional<double> eps=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/round_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/round_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..b5549e84a4769721b090c66ae92c589416213baa
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/round_cpu_dispatch.h
@@ -0,0 +1,30 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor 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 cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_with_noise_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_with_noise_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..dfb4fc555e9b788deb5cb6fcb59711b94641dba2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rrelu_with_noise_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> rrelu_with_noise_functional(const at::Tensor & self, const at::Tensor & noise, const at::Scalar & lower=0.125, const at::Scalar & upper=0.3333333333333333, bool training=false, ::std::optional<at::Generator> generator=::std::nullopt);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/rsqrt_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/rsqrt_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..ae9c9959cb2ba8db74000c2bb08d145b7f766724
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/rsqrt_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 rsqrt(const at::Tensor & self);
+TORCH_API at::Tensor & rsqrt_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & rsqrt_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & rsqrt_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_add_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_add_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3033e1c889e3187594fc085bc31f2de0f7eed04c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_add_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 scatter_add(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src);
+TORCH_API at::Tensor & scatter_add_(at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4561c1f978ecb362ec541c70cddda43556fcfe87
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/scatter_compositeexplicitautogradnonfunctional_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 compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor scatter(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src);
+TORCH_API at::Tensor & scatter_(at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src);
+TORCH_API at::Tensor scatter(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Scalar & value);
+TORCH_API at::Tensor & scatter_(at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Scalar & value);
+TORCH_API at::Tensor scatter(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, c10::string_view reduce);
+TORCH_API at::Tensor & scatter_(at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Tensor & src, c10::string_view reduce);
+TORCH_API at::Tensor scatter(const at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Scalar & value, c10::string_view reduce);
+TORCH_API at::Tensor & scatter_(at::Tensor & self, int64_t dim, const at::Tensor & index, const at::Scalar & value, c10::string_view reduce);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/select_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/select_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..069061a2d2a14e4123548471cfb0d7eef2306b8e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/select_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_Dimname {
+  using schema = at::Tensor (const at::Tensor &, at::Dimname, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::select";
+  static constexpr const char* overload_name = "Dimname";
+  static constexpr const char* schema_str = "select.Dimname(Tensor(a) self, Dimname dim, int index) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, at::Dimname dim, int64_t index);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Dimname dim, int64_t index);
+};
+
+struct TORCH_API select_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";
+  static constexpr const char* overload_name = "int";
+  static constexpr const char* schema_str = "select.int(Tensor(a) self, int dim, SymInt index) -> Tensor(a)";
+  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);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/select_scatter_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/select_scatter_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f428a1d5c816f147904bef6bd61780ad852bce10
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/select_scatter_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_scatter(const at::Tensor & self, const at::Tensor & src, int64_t dim, int64_t index);
+TORCH_API at::Tensor select_scatter_symint(const at::Tensor & self, const at::Tensor & src, int64_t dim, c10::SymInt index);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..53a0b4584cdc3f1db237364867ab4ec98e60f980
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sigmoid_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 sigmoid_backward(const at::Tensor & grad_output, const at::Tensor & output);
+TORCH_API at::Tensor & sigmoid_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & output);
+TORCH_API at::Tensor & sigmoid_backward_outf(const at::Tensor & grad_output, const at::Tensor & output, at::Tensor & grad_input);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sign.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sign.h
new file mode 100644
index 0000000000000000000000000000000000000000..0585cfa20e21f8262b40852e9a39f945418e7867
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sign.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/sign_ops.h>
+
+namespace at {
+
+
+// aten::sign(Tensor self) -> Tensor
+inline at::Tensor sign(const at::Tensor & self) {
+    return at::_ops::sign::call(self);
+}
+
+// aten::sign.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & sign_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::sign_out::call(self, out);
+}
+// aten::sign.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & sign_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::sign_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/signbit_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/signbit_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..eecd921d7307e9c76edce50cdc87a0f093210b64
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/signbit_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 signbit(const at::Tensor & self);
+TORCH_API at::Tensor & signbit_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & signbit_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/signbit_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/signbit_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e1210ea2aae6503bcedd287583cbfe09438fcf8a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/signbit_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 signbit(const at::Tensor & self);
+TORCH_API at::Tensor & signbit_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & signbit_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/silu_backward_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/silu_backward_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a3c7bf327ded56b22e5ade6809baa7146cb194d8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/silu_backward_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor silu_backward(const at::Tensor & grad_output, const at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/silu_backward_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/silu_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d984b785b39e00e01c138705ffe39f9471d8a5f9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/silu_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 silu_backward(const at::Tensor & grad_output, const at::Tensor & self);
+TORCH_API at::Tensor & silu_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self);
+TORCH_API at::Tensor & silu_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/sinc_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sinc_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..97c2167b9d293f218d17596ed2e08d1a9e146ae2
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sinc_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 sinc(const at::Tensor & self);
+TORCH_API at::Tensor & sinc_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & sinc_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & sinc_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sinc_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sinc_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..dcd9762ae3581007abff1b51b0bf632e0e23d2a1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sinc_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 sinc {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sinc";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "sinc(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+struct TORCH_API sinc_ {
+  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::sinc_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "sinc_(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 sinc_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::sinc";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "sinc.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/size_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/size_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0b8a3fc58301808f83e630b80f3ebe252b0ea31d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/size_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 int64_t size(const at::Tensor & self, int64_t dim);
+TORCH_API int64_t size(const at::Tensor & self, at::Dimname dim);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slice_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..227d1462b0e7f16e4083fe2c1d02f44b5f8618eb
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_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 slice(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_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 compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slice_copy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_copy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..a3e31e45b76e1523eef271159832a96d5852e251
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_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 & slice_copy_Tensor_out_symint(const at::Tensor & self, int64_t dim, ::std::optional<c10::SymInt> start, ::std::optional<c10::SymInt> end, c10::SymInt step, at::Tensor & out);
+TORCH_API at::Tensor slice_copy_Tensor_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 native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slice_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..e7bf6e5a59a2585296f47693fb44377acf0ed754
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_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 slice(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);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slice_scatter_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_scatter_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..721837de96b90eab6ca3256221fb56b98330c979
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slice_scatter_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_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);
+TORCH_API 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);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slogdet_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slogdet_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..50d8c75edfbfd34c8c2a94390a1cb8a775b85144
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slogdet_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> slogdet(const at::Tensor & self);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> slogdet_out(const at::Tensor & self, at::Tensor & sign, at::Tensor & logabsdet);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_transpose2d_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_transpose2d_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e23f1f6bd34915d302da42131f49fb1353ed900f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/slow_conv_transpose2d_cuda_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor slow_conv_transpose2d(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias={}, at::IntArrayRef stride=1, at::IntArrayRef padding=0, at::IntArrayRef output_padding=0, at::IntArrayRef dilation=1);
+TORCH_API at::Tensor slow_conv_transpose2d_symint(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef padding=c10::SymInt(0), c10::SymIntArrayRef output_padding=c10::SymInt(0), c10::SymIntArrayRef dilation=c10::SymInt(1));
+TORCH_API at::Tensor & slow_conv_transpose2d_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias={}, at::IntArrayRef stride=1, at::IntArrayRef padding=0, at::IntArrayRef output_padding=0, at::IntArrayRef dilation=1);
+TORCH_API at::Tensor & slow_conv_transpose2d_outf(const at::Tensor & self, const at::Tensor & weight, at::IntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef output_padding, at::IntArrayRef dilation, at::Tensor & out);
+TORCH_API at::Tensor & slow_conv_transpose2d_symint_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias={}, c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef padding=c10::SymInt(0), c10::SymIntArrayRef output_padding=c10::SymInt(0), c10::SymIntArrayRef dilation=c10::SymInt(1));
+TORCH_API at::Tensor & slow_conv_transpose2d_symint_outf(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef kernel_size, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef dilation, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/smooth_l1_loss_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/smooth_l1_loss_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..18a9966ee3818f2e9f8a9298161b8a1c1e3b8f04
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/smooth_l1_loss_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API smooth_l1_loss_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, int64_t, double, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::smooth_l1_loss";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "smooth_l1_loss.out(Tensor self, Tensor target, int reduction=Mean, float beta=1.0, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & target, int64_t reduction, double beta, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & target, int64_t reduction, double beta, at::Tensor & out);
+};
+
+struct TORCH_API smooth_l1_loss {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, int64_t, double);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::smooth_l1_loss";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "smooth_l1_loss(Tensor self, Tensor target, int reduction=Mean, float beta=1.0) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & target, int64_t reduction, double beta);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & target, int64_t reduction, double beta);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/softplus_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/softplus_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..76674116588fd5dd0d9488cbe83fb08199b479e6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/softplus_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 softplus(const at::Tensor & self, const at::Scalar & beta=1, const at::Scalar & threshold=20);
+TORCH_API at::Tensor & softplus_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & beta=1, const at::Scalar & threshold=20);
+TORCH_API at::Tensor & softplus_outf(const at::Tensor & self, const at::Scalar & beta, const at::Scalar & threshold, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_compressed_tensor.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_compressed_tensor.h
new file mode 100644
index 0000000000000000000000000000000000000000..753f17d21d8fa5939befb38893ee8c282bfe9a87
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_compressed_tensor.h
@@ -0,0 +1,78 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/sparse_compressed_tensor_ops.h>
+
+namespace at {
+
+
+// aten::sparse_compressed_tensor.comp_plain_value_size(Tensor compressed_indices, Tensor plain_indices, Tensor values, SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_compressed_tensor(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, at::IntArrayRef size, at::TensorOptions options) {
+    return at::_ops::sparse_compressed_tensor_comp_plain_value_size::call(compressed_indices, plain_indices, values, c10::fromIntArrayRefSlow(size), c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor sparse_compressed_tensor(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, at::IntArrayRef size, at::TensorOptions options) {
+    return at::_ops::sparse_compressed_tensor_comp_plain_value_size::call(compressed_indices, plain_indices, values, c10::fromIntArrayRefSlow(size), c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+  }
+}
+
+// aten::sparse_compressed_tensor.comp_plain_value_size(Tensor compressed_indices, Tensor plain_indices, Tensor values, SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline 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) {
+    return at::_ops::sparse_compressed_tensor_comp_plain_value_size::call(compressed_indices, plain_indices, values, c10::fromIntArrayRefSlow(size), dtype, layout, device, pin_memory);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  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) {
+    return at::_ops::sparse_compressed_tensor_comp_plain_value_size::call(compressed_indices, plain_indices, values, c10::fromIntArrayRefSlow(size), dtype, layout, device, pin_memory);
+  }
+}
+
+// aten::sparse_compressed_tensor.comp_plain_value_size(Tensor compressed_indices, Tensor plain_indices, Tensor values, SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_compressed_tensor_symint(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, c10::SymIntArrayRef size, at::TensorOptions options) {
+    return at::_ops::sparse_compressed_tensor_comp_plain_value_size::call(compressed_indices, plain_indices, values, size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor sparse_compressed_tensor(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, c10::SymIntArrayRef size, at::TensorOptions options) {
+    return at::_ops::sparse_compressed_tensor_comp_plain_value_size::call(compressed_indices, plain_indices, values, size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+  }
+}
+
+// aten::sparse_compressed_tensor.comp_plain_value_size(Tensor compressed_indices, Tensor plain_indices, Tensor values, SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_compressed_tensor_symint(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::sparse_compressed_tensor_comp_plain_value_size::call(compressed_indices, plain_indices, values, size, dtype, layout, device, pin_memory);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor sparse_compressed_tensor(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::sparse_compressed_tensor_comp_plain_value_size::call(compressed_indices, plain_indices, values, size, dtype, layout, device, pin_memory);
+  }
+}
+
+// aten::sparse_compressed_tensor.comp_plain_value(Tensor compressed_indices, Tensor plain_indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor sparse_compressed_tensor(const at::Tensor & compressed_indices, const at::Tensor & plain_indices, const at::Tensor & values, at::TensorOptions options) {
+    return at::_ops::sparse_compressed_tensor_comp_plain_value::call(compressed_indices, plain_indices, values, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::sparse_compressed_tensor.comp_plain_value(Tensor compressed_indices, Tensor plain_indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline 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) {
+    return at::_ops::sparse_compressed_tensor_comp_plain_value::call(compressed_indices, plain_indices, values, dtype, layout, device, pin_memory);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_coo_tensor_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_coo_tensor_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..0573dfe2a70a29992a6890c6de36897f9c34b14c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_coo_tensor_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 sparse_coo_tensor(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_coo_tensor_size_out(at::IntArrayRef size, at::Tensor & out);
+TORCH_API at::Tensor sparse_coo_tensor(const at::Tensor & indices, const at::Tensor & values, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={}, ::std::optional<bool> is_coalesced=::std::nullopt);
+TORCH_API at::Tensor sparse_coo_tensor(const at::Tensor & indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={}, ::std::optional<bool> is_coalesced=::std::nullopt);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_sampled_addmm.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_sampled_addmm.h
new file mode 100644
index 0000000000000000000000000000000000000000..759c3978c68561f7cd3b95ca7a03de61b94b9ca9
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_sampled_addmm.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_sampled_addmm_ops.h>
+
+namespace at {
+
+
+// aten::sparse_sampled_addmm.out(Tensor self, Tensor mat1, Tensor mat2, *, Scalar beta=1, Scalar alpha=1, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & sparse_sampled_addmm_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1) {
+    return at::_ops::sparse_sampled_addmm_out::call(self, mat1, mat2, beta, alpha, out);
+}
+// aten::sparse_sampled_addmm.out(Tensor self, Tensor mat1, Tensor mat2, *, Scalar beta=1, Scalar alpha=1, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & sparse_sampled_addmm_outf(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta, const at::Scalar & alpha, at::Tensor & out) {
+    return at::_ops::sparse_sampled_addmm_out::call(self, mat1, mat2, beta, alpha, out);
+}
+
+// aten::sparse_sampled_addmm(Tensor self, Tensor mat1, Tensor mat2, *, Scalar beta=1, Scalar alpha=1) -> Tensor
+inline at::Tensor sparse_sampled_addmm(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1) {
+    return at::_ops::sparse_sampled_addmm::call(self, mat1, mat2, beta, alpha);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_sampled_addmm_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_sampled_addmm_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4dedea44099006ead54ec06347ad0d70b3132701
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sparse_sampled_addmm_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 sparse_sampled_addmm_sparse_csr_cpu(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & sparse_sampled_addmm_out_sparse_csr_cpu(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta, const at::Scalar & alpha, at::Tensor & out);
+TORCH_API at::Tensor sparse_sampled_addmm_sparse_csr_cuda(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & sparse_sampled_addmm_out_sparse_csr_cuda(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta, const at::Scalar & alpha, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_bessel_j0_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_bessel_j0_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..a58b8709f6a0360a242d4945bdfaa7ff4c73cea3
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_bessel_j0_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_bessel_j0 : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_entr_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_entr_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..f5018e78c62bd2beabcbc1d96c18b13eeb8f13b4
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_entr_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_entr : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_erfinv.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_erfinv.h
new file mode 100644
index 0000000000000000000000000000000000000000..0385c1fc7015e9568a5e7c04beeb69cbd3e75319
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_erfinv.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/special_erfinv_ops.h>
+
+namespace at {
+
+
+// aten::special_erfinv(Tensor self) -> Tensor
+inline at::Tensor special_erfinv(const at::Tensor & self) {
+    return at::_ops::special_erfinv::call(self);
+}
+
+// aten::special_erfinv.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_erfinv_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::special_erfinv_out::call(self, out);
+}
+// aten::special_erfinv.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_erfinv_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::special_erfinv_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_i0e_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i0e_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..eed4e1cff11d87ac33d0b6eb36bddfa644c45f31
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i0e_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_i0e(const at::Tensor & self);
+TORCH_API at::Tensor & special_i0e_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & special_i0e_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_i1_cpu_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i1_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7504ffc7a966306c574a6fd629595cacfe293b7b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i1_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor special_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 cpu
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_i1_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i1_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..399bdf09023646929cf54fd3671ea5c77ad4193b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_i1_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_i1 {
+  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_i1";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_i1(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_i1_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_i1";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_i1.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_log_ndtr.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_log_ndtr.h
new file mode 100644
index 0000000000000000000000000000000000000000..27845aafbcfa60a94cd7e22b2eadb9c847bd0d9f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_log_ndtr.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/special_log_ndtr_ops.h>
+
+namespace at {
+
+
+// aten::special_log_ndtr(Tensor self) -> Tensor
+inline at::Tensor special_log_ndtr(const at::Tensor & self) {
+    return at::_ops::special_log_ndtr::call(self);
+}
+
+// aten::special_log_ndtr.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_log_ndtr_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::special_log_ndtr_out::call(self, out);
+}
+// aten::special_log_ndtr.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_log_ndtr_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::special_log_ndtr_out::call(self, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtr_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtr_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..83f8103e9590315b062e1f2738ef142ab110839f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtr_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor special_ndtr(const at::Tensor & self);
+TORCH_API at::Tensor & special_ndtr_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & special_ndtr_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtri_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtri_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..397b8eebf6a5c193fe9edd8393c36566a4bc4bd6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_ndtri_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_ndtri(const at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k0_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k0_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..013862d9cbc14d2b52b3ba91a8b1c896c90e124a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k0_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_scaled_modified_bessel_k0(const at::Tensor & x);
+TORCH_API at::Tensor & special_scaled_modified_bessel_k0_out(at::Tensor & out, const at::Tensor & x);
+TORCH_API at::Tensor & special_scaled_modified_bessel_k0_outf(const at::Tensor & x, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k0_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k0_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..ccd732b4cb08f4aa3b988ff16d5b07e9a84e4b98
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_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_scaled_modified_bessel_k0_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_special_scaled_modified_bessel_k0_out : public at::meta::structured_special_scaled_modified_bessel_k0 {
+void impl(const at::Tensor & x, const at::Tensor & out);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k0_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k0_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..11d6b7848923e99b8dd57c551fe8c9c0badcc957
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_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_scaled_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_scaled_modified_bessel_k0";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_scaled_modified_bessel_k0(Tensor x) -> Tensor";
+  static at::Tensor call(const at::Tensor & x);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & x);
+};
+
+struct TORCH_API special_scaled_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_scaled_modified_bessel_k0";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_scaled_modified_bessel_k0.out(Tensor x, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & x, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & x, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k1_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k1_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..9c2b3c888900c60eef5e14396aa6cad9fd9a4c2b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_scaled_modified_bessel_k1_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_scaled_modified_bessel_k1 : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & x);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_u.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_u.h
new file mode 100644
index 0000000000000000000000000000000000000000..c565cd20079cb8461169470da95b19db0bd64049
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_shifted_chebyshev_polynomial_u.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_shifted_chebyshev_polynomial_u_ops.h>
+
+namespace at {
+
+
+// aten::special_shifted_chebyshev_polynomial_u(Tensor x, Tensor n) -> Tensor
+inline at::Tensor special_shifted_chebyshev_polynomial_u(const at::Tensor & x, const at::Tensor & n) {
+    return at::_ops::special_shifted_chebyshev_polynomial_u::call(x, n);
+}
+
+// aten::special_shifted_chebyshev_polynomial_u.x_scalar(Scalar x, Tensor n) -> Tensor
+inline at::Tensor special_shifted_chebyshev_polynomial_u(const at::Scalar & x, const at::Tensor & n) {
+    return at::_ops::special_shifted_chebyshev_polynomial_u_x_scalar::call(x, n);
+}
+
+// aten::special_shifted_chebyshev_polynomial_u.n_scalar(Tensor x, Scalar n) -> Tensor
+inline at::Tensor special_shifted_chebyshev_polynomial_u(const at::Tensor & x, const at::Scalar & n) {
+    return at::_ops::special_shifted_chebyshev_polynomial_u_n_scalar::call(x, n);
+}
+
+// aten::special_shifted_chebyshev_polynomial_u.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_shifted_chebyshev_polynomial_u_out(at::Tensor & out, const at::Tensor & x, const at::Tensor & n) {
+    return at::_ops::special_shifted_chebyshev_polynomial_u_out::call(x, n, out);
+}
+// aten::special_shifted_chebyshev_polynomial_u.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_shifted_chebyshev_polynomial_u_outf(const at::Tensor & x, const at::Tensor & n, at::Tensor & out) {
+    return at::_ops::special_shifted_chebyshev_polynomial_u_out::call(x, n, out);
+}
+
+// aten::special_shifted_chebyshev_polynomial_u.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_shifted_chebyshev_polynomial_u_out(at::Tensor & out, const at::Scalar & x, const at::Tensor & n) {
+    return at::_ops::special_shifted_chebyshev_polynomial_u_x_scalar_out::call(x, n, out);
+}
+// aten::special_shifted_chebyshev_polynomial_u.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_shifted_chebyshev_polynomial_u_outf(const at::Scalar & x, const at::Tensor & n, at::Tensor & out) {
+    return at::_ops::special_shifted_chebyshev_polynomial_u_x_scalar_out::call(x, n, out);
+}
+
+// aten::special_shifted_chebyshev_polynomial_u.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_shifted_chebyshev_polynomial_u_out(at::Tensor & out, const at::Tensor & x, const at::Scalar & n) {
+    return at::_ops::special_shifted_chebyshev_polynomial_u_n_scalar_out::call(x, n, out);
+}
+// aten::special_shifted_chebyshev_polynomial_u.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_shifted_chebyshev_polynomial_u_outf(const at::Tensor & x, const at::Scalar & n, at::Tensor & out) {
+    return at::_ops::special_shifted_chebyshev_polynomial_u_n_scalar_out::call(x, n, out);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_spherical_bessel_j0_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_spherical_bessel_j0_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..53db55b7031105be8d010ea6e3a8ef5fcf6bdb9a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_spherical_bessel_j0_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_spherical_bessel_j0 : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & x);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlogy_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlogy_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..75c527395e50d93d051598909149e6daeab9ca6b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/special_xlogy_ops.h
@@ -0,0 +1,83 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API special_xlogy {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_xlogy";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "special_xlogy(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 special_xlogy_self_scalar {
+  using schema = at::Tensor (const at::Scalar &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_xlogy";
+  static constexpr const char* overload_name = "self_scalar";
+  static constexpr const char* schema_str = "special_xlogy.self_scalar(Scalar self, Tensor other) -> Tensor";
+  static at::Tensor call(const at::Scalar & self, const at::Tensor & other);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & self, const at::Tensor & other);
+};
+
+struct TORCH_API special_xlogy_other_scalar {
+  using schema = at::Tensor (const at::Tensor &, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_xlogy";
+  static constexpr const char* overload_name = "other_scalar";
+  static constexpr const char* schema_str = "special_xlogy.other_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 special_xlogy_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_xlogy";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "special_xlogy.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 special_xlogy_self_scalar_out {
+  using schema = at::Tensor & (const at::Scalar &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_xlogy";
+  static constexpr const char* overload_name = "self_scalar_out";
+  static constexpr const char* schema_str = "special_xlogy.self_scalar_out(Scalar self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Scalar & self, const at::Tensor & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Scalar & self, const at::Tensor & other, at::Tensor & out);
+};
+
+struct TORCH_API special_xlogy_other_scalar_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Scalar &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::special_xlogy";
+  static constexpr const char* overload_name = "other_scalar_out";
+  static constexpr const char* schema_str = "special_xlogy.other_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);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/split_with_sizes_copy_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/split_with_sizes_copy_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..d41fc5896ea79126b7b82c9f236f4991ac927077
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/split_with_sizes_copy_cuda_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API void split_with_sizes_copy_out(at::TensorList out, const at::Tensor & self, at::IntArrayRef split_sizes, int64_t dim=0);
+TORCH_API void split_with_sizes_copy_outf(const at::Tensor & self, at::IntArrayRef split_sizes, int64_t dim, at::TensorList out);
+TORCH_API void split_with_sizes_copy_symint_out(at::TensorList out, const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim=0);
+TORCH_API void split_with_sizes_copy_symint_outf(const at::Tensor & self, c10::SymIntArrayRef split_sizes, int64_t dim, at::TensorList out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sqrt_compositeexplicitautogradnonfunctional_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sqrt_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..5c78f87d09af69fad98ed358c18bb47c61f7a84d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sqrt_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 sqrt(const at::Tensor & self);
+TORCH_API at::Tensor & sqrt_(at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/square_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/square_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..82e8a9f3909816ffca6d42ea8319030a2f9064b8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/square_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 square(const at::Tensor & self);
+TORCH_API at::Tensor & square_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & square_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & square_(at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/stride.h b/lib/python3.12/site-packages/torch/include/ATen/ops/stride.h
new file mode 100644
index 0000000000000000000000000000000000000000..3861df3bcf90e6539aa24b44f732d0a5e817e897
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/stride.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/stride_ops.h>
+
+namespace at {
+
+
+// aten::stride.int(Tensor self, int dim) -> int
+inline int64_t __dispatch_stride(const at::Tensor & self, int64_t dim) {
+    return at::_ops::stride_int::call(self, dim);
+}
+
+// aten::stride.Dimname(Tensor self, Dimname dim) -> int
+inline int64_t stride(const at::Tensor & self, at::Dimname dim) {
+    return at::_ops::stride_Dimname::call(self, dim);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sym_numel.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sym_numel.h
new file mode 100644
index 0000000000000000000000000000000000000000..bf5e612f9b14791f9ace54ed8041a34c682ad990
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sym_numel.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/sym_numel_ops.h>
+
+namespace at {
+
+
+// aten::sym_numel(Tensor self) -> SymInt
+inline c10::SymInt __dispatch_sym_numel(const at::Tensor & self) {
+    return at::_ops::sym_numel::call(self);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/sym_stride.h b/lib/python3.12/site-packages/torch/include/ATen/ops/sym_stride.h
new file mode 100644
index 0000000000000000000000000000000000000000..13c9edd8304a654a1ad78d50ef166a44628f6755
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/sym_stride.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/sym_stride_ops.h>
+
+namespace at {
+
+
+// aten::sym_stride.int(Tensor self, int dim) -> SymInt
+inline c10::SymInt __dispatch_sym_stride(const at::Tensor & self, int64_t dim) {
+    return at::_ops::sym_stride_int::call(self, dim);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/t_copy_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/t_copy_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..49d4f876de01faf504d80cb6137b7429830c553b
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/t_copy_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & t_copy_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & t_copy_outf(const at::Tensor & self, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/take_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/take_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..68e3727e44bb341665c6218bdab80302b2a2a6e7
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/take_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 take(const at::Tensor & self, const at::Tensor & index);
+TORCH_API at::Tensor & take_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & index);
+TORCH_API at::Tensor & take_outf(const at::Tensor & self, const at::Tensor & index, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..8e4092d377a55121da6ddb38cf1dbe4e80af238c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_backward_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/tanh_backward_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_tanh_backward_out : public at::meta::structured_tanh_backward {
+void impl(const at::Tensor & grad_output, const at::Tensor & output, const at::Tensor & grad_input);
+};
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..729c490a0328035dea4e34daaf771e63954c842a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/tanh_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 tanh(const at::Tensor & self);
+TORCH_API at::Tensor & tanh_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & tanh_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & tanh_(at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..6430b8023e05eae06ddd1aa38c95c3ee715d87c1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_backward_native.h
@@ -0,0 +1,29 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/threshold_backward_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_threshold_backward_out : public at::meta::structured_threshold_backward {
+void impl(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold, const at::Tensor & grad_input);
+};
+TORCH_API at::Tensor threshold_backwards_nested(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold);
+TORCH_API at::Tensor threshold_backward_sparse(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold);
+TORCH_API at::Tensor & threshold_backward_sparse_out(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold, at::Tensor & grad_input);
+TORCH_API at::Tensor threshold_backward_sparse_compressed(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold);
+TORCH_API at::Tensor & threshold_backward_sparse_compressed_out(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold, at::Tensor & grad_input);
+TORCH_API at::Tensor mkldnn_relu_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Scalar & threshold);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..17f263cf7853d94049a661e8d29741e5e8ecc073
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/threshold_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_threshold : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Scalar & threshold, const at::Scalar & value);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/to_dense_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/to_dense_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..5d306c3a4ed7ea5b4dcc699786710bf9f2bb3a3f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/to_dense_backward_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API to_dense_backward {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::to_dense_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "to_dense_backward(Tensor grad, Tensor input, bool? masked_grad=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & grad, const at::Tensor & input, ::std::optional<bool> masked_grad);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad, const at::Tensor & input, ::std::optional<bool> masked_grad);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/trace_backward.h b/lib/python3.12/site-packages/torch/include/ATen/ops/trace_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..5bb440ba8256047abd82c568b4ceae37b5204053
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/trace_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/trace_backward_ops.h>
+
+namespace at {
+
+
+// aten::trace_backward(Tensor grad, SymInt[] sizes) -> Tensor
+inline at::Tensor trace_backward(const at::Tensor & grad, at::IntArrayRef sizes) {
+    return at::_ops::trace_backward::call(grad, c10::fromIntArrayRefSlow(sizes));
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor trace_backward(const at::Tensor & grad, at::IntArrayRef sizes) {
+    return at::_ops::trace_backward::call(grad, c10::fromIntArrayRefSlow(sizes));
+  }
+}
+
+// aten::trace_backward(Tensor grad, SymInt[] sizes) -> Tensor
+inline at::Tensor trace_backward_symint(const at::Tensor & grad, c10::SymIntArrayRef sizes) {
+    return at::_ops::trace_backward::call(grad, sizes);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor trace_backward(const at::Tensor & grad, c10::SymIntArrayRef sizes) {
+    return at::_ops::trace_backward::call(grad, sizes);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/transpose_compositeimplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/transpose_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..30fe07181bba8cd6093401257cb5664f86e9ca6d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/transpose_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 transpose(const at::Tensor & self, at::Dimname dim0, at::Dimname dim1);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/transpose_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/transpose_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..e72f7a8b1ec1cc415a40ae4e276b60140b332325
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/transpose_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 transpose(const at::Tensor & self, int64_t dim0, int64_t dim1);
+TORCH_API at::Tensor transpose_nested(const at::Tensor & self, int64_t dim0, int64_t dim1);
+TORCH_API at::Tensor & transpose_(at::Tensor & self, int64_t dim0, int64_t dim1);
+TORCH_API at::Tensor transpose(const at::Tensor & self, at::Dimname dim0, at::Dimname dim1);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/triangular_solve_cuda_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/triangular_solve_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..bf2f57766b0c156aae72d26d376d7d22733ddba8
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/triangular_solve_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> triangular_solve(const at::Tensor & self, const at::Tensor & A, bool upper=true, bool transpose=false, bool unitriangular=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> triangular_solve_out(at::Tensor & X, at::Tensor & M, const at::Tensor & self, const at::Tensor & A, bool upper=true, bool transpose=false, bool unitriangular=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> triangular_solve_outf(const at::Tensor & self, const at::Tensor & A, bool upper, bool transpose, bool unitriangular, at::Tensor & X, at::Tensor & M);
+
+} // namespace cuda
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/tril_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/tril_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..714abca1ee0fb90bb08dd07a2442bf42639e9f30
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/tril_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_tril : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, int64_t diagonal);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/trunc_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/trunc_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..d1e2726a564464e636dc395a2576ba62c848550a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/trunc_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_trunc : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/trunc_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/trunc_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..98d6565fec5dc546da71878a74d1e8c7b8966355
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/trunc_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 trunc {
+  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::trunc";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "trunc(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 trunc_ {
+  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::trunc_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "trunc_(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 trunc_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::trunc";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "trunc.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/unsafe_split_with_sizes_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_split_with_sizes_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..ac7ec21702b473c729f60313a5d78f9a22ba379a
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unsafe_split_with_sizes_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 unsafe_split_with_sizes {
+  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::unsafe_split_with_sizes";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "unsafe_split_with_sizes(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 unsafe_split_with_sizes_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::unsafe_split_with_sizes";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "unsafe_split_with_sizes.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/unsqueeze.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unsqueeze.h
new file mode 100644
index 0000000000000000000000000000000000000000..5abbe8932433d8b6c0e1b1d358b1644dfc8f5e78
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unsqueeze.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/unsqueeze_ops.h>
+
+namespace at {
+
+
+// aten::unsqueeze(Tensor(a) self, int dim) -> Tensor(a)
+inline at::Tensor unsqueeze(const at::Tensor & self, int64_t dim) {
+    return at::_ops::unsqueeze::call(self, dim);
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unsqueeze_compositeexplicitautograd_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unsqueeze_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0fcea185ba7b6577fa61e1a23b084f5a0ebe6dc6
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unsqueeze_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor unsqueeze(const at::Tensor & self, int64_t dim);
+TORCH_API at::Tensor & unsqueeze_(at::Tensor & self, int64_t dim);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/unsqueeze_copy_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/unsqueeze_copy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..45b4240bf5f32ad22773c96e4a3ed01febf8e02e
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/unsqueeze_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 & unsqueeze_copy_out(const at::Tensor & self, int64_t dim, at::Tensor & out);
+TORCH_API at::Tensor unsqueeze_copy(const at::Tensor & self, int64_t dim);
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bicubic2d_backward_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bicubic2d_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..21d40396362b7d12a744f341b78c61ef284ec412
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bicubic2d_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_bicubic2d_backward_grad_input {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, 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_backward";
+  static constexpr const char* overload_name = "grad_input";
+  static constexpr const char* schema_str = "upsample_bicubic2d_backward.grad_input(Tensor grad_output, SymInt[2] output_size, SymInt[4] input_size, bool align_corners, 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, bool align_corners, ::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, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input);
+};
+
+struct TORCH_API upsample_bicubic2d_backward {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, 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_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "upsample_bicubic2d_backward(Tensor grad_output, SymInt[2] output_size, SymInt[4] input_size, bool align_corners, 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, bool align_corners, ::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, 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_bicubic2d_meta_dispatch.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bicubic2d_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..62d48b676ab4eee38f2f0ea3dc695a4c8fbb3f05
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_bicubic2d_meta_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor upsample_bicubic2d(const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor upsample_bicubic2d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_bicubic2d_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_bicubic2d_outf(const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+TORCH_API at::Tensor & upsample_bicubic2d_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & upsample_bicubic2d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_backward_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_backward_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..a554043f3121c9eeffae0f0e505ffa260d0db779
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_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_upsample_linear1d_backward : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & grad_output, at::ArrayRef<int64_t> output_size, at::ArrayRef<int64_t> input_size, bool align_corners, ::std::optional<double> scales);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_backward_native.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..7d5194edd437f4a797c274299cf54af745f1d49c
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_linear1d_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_linear1d_backward_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_upsample_linear1d_backward_out_cpu : public at::meta::structured_upsample_linear1d_backward {
+void impl(const at::Tensor & grad_output, at::ArrayRef<int64_t> output_size, at::ArrayRef<int64_t> input_size, bool align_corners, ::std::optional<double> scales, const at::Tensor & grad_input);
+};
+struct TORCH_API structured_upsample_linear1d_backward_out_cuda : public at::meta::structured_upsample_linear1d_backward {
+void impl(const at::Tensor & grad_output, at::ArrayRef<int64_t> output_size, at::ArrayRef<int64_t> input_size, bool align_corners, ::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_nearest1d_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest1d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f86005e6b8f49d6b129213796184d53a8e53bc7f
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest1d_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_nearest1d_vec {
+  using schema = at::Tensor (const at::Tensor &, at::OptionalSymIntArrayRef, ::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_nearest1d";
+  static constexpr const char* overload_name = "vec";
+  static constexpr const char* schema_str = "upsample_nearest1d.vec(Tensor input, SymInt[]? output_size, float[]? scale_factors) -> Tensor";
+  static at::Tensor call(const at::Tensor & input, at::OptionalSymIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, at::OptionalSymIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors);
+};
+
+struct TORCH_API upsample_nearest1d_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, ::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_nearest1d";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "upsample_nearest1d.out(Tensor self, SymInt[1] output_size, float? scales=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales, at::Tensor & out);
+};
+
+struct TORCH_API upsample_nearest1d {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, ::std::optional<double>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::upsample_nearest1d";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "upsample_nearest1d(Tensor self, SymInt[1] output_size, float? scales=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales);
+};
+
+}} // namespace at::_ops
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest2d.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest2d.h
new file mode 100644
index 0000000000000000000000000000000000000000..d0375ee76fcf10cde91991a4b777eba2e2723d76
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest2d.h
@@ -0,0 +1,157 @@
+#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_nearest2d_ops.h>
+
+namespace at {
+
+
+// aten::upsample_nearest2d.vec(Tensor input, SymInt[]? output_size, float[]? scale_factors) -> Tensor
+inline at::Tensor upsample_nearest2d(const at::Tensor & input, at::OptionalIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest2d_vec::call(input, output_size.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*output_size)) : ::std::nullopt, scale_factors);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor upsample_nearest2d(const at::Tensor & input, at::OptionalIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest2d_vec::call(input, output_size.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*output_size)) : ::std::nullopt, scale_factors);
+  }
+}
+
+// aten::upsample_nearest2d.vec(Tensor input, SymInt[]? output_size, float[]? scale_factors) -> Tensor
+inline at::Tensor upsample_nearest2d_symint(const at::Tensor & input, at::OptionalSymIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest2d_vec::call(input, output_size, scale_factors);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor upsample_nearest2d(const at::Tensor & input, at::OptionalSymIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest2d_vec::call(input, output_size, scale_factors);
+  }
+}
+
+// aten::upsample_nearest2d.out(Tensor self, SymInt[2] output_size, float? scales_h=None, float? scales_w=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest2d_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) {
+    return at::_ops::upsample_nearest2d_out::call(self, c10::fromIntArrayRefSlow(output_size), scales_h, scales_w, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & upsample_nearest2d_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) {
+    return at::_ops::upsample_nearest2d_out::call(self, c10::fromIntArrayRefSlow(output_size), scales_h, scales_w, out);
+  }
+}
+
+// aten::upsample_nearest2d.out(Tensor self, SymInt[2] output_size, float? scales_h=None, float? scales_w=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest2d_outf(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out) {
+    return at::_ops::upsample_nearest2d_out::call(self, c10::fromIntArrayRefSlow(output_size), scales_h, scales_w, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & upsample_nearest2d_outf(const at::Tensor & self, at::IntArrayRef output_size, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out) {
+    return at::_ops::upsample_nearest2d_out::call(self, c10::fromIntArrayRefSlow(output_size), scales_h, scales_w, out);
+  }
+}
+
+// aten::upsample_nearest2d.out(Tensor self, SymInt[2] output_size, float? scales_h=None, float? scales_w=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest2d_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) {
+    return at::_ops::upsample_nearest2d_out::call(self, output_size, scales_h, scales_w, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & upsample_nearest2d_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) {
+    return at::_ops::upsample_nearest2d_out::call(self, output_size, scales_h, scales_w, out);
+  }
+}
+
+// aten::upsample_nearest2d.out(Tensor self, SymInt[2] output_size, float? scales_h=None, float? scales_w=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest2d_symint_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out) {
+    return at::_ops::upsample_nearest2d_out::call(self, output_size, scales_h, scales_w, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & upsample_nearest2d_outf(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & out) {
+    return at::_ops::upsample_nearest2d_out::call(self, output_size, scales_h, scales_w, out);
+  }
+}
+
+// aten::upsample_nearest2d(Tensor self, SymInt[2] output_size, float? scales_h=None, float? scales_w=None) -> Tensor
+inline 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) {
+    return at::_ops::upsample_nearest2d::call(self, c10::fromIntArrayRefSlow(output_size), scales_h, scales_w);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  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) {
+    return at::_ops::upsample_nearest2d::call(self, c10::fromIntArrayRefSlow(output_size), scales_h, scales_w);
+  }
+}
+
+// aten::upsample_nearest2d(Tensor self, SymInt[2] output_size, float? scales_h=None, float? scales_w=None) -> Tensor
+inline 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) {
+    return at::_ops::upsample_nearest2d::call(self, output_size, scales_h, scales_w);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor upsample_nearest2d(const at::Tensor & self, c10::SymIntArrayRef output_size, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt) {
+    return at::_ops::upsample_nearest2d::call(self, output_size, scales_h, scales_w);
+  }
+}
+
+// aten::upsample_nearest2d.vec_out(Tensor input, SymInt[]? output_size, float[]? scale_factors, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest2d_out(at::Tensor & out, const at::Tensor & input, at::OptionalIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest2d_vec_out::call(input, output_size.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*output_size)) : ::std::nullopt, scale_factors, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & upsample_nearest2d_out(at::Tensor & out, const at::Tensor & input, at::OptionalIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest2d_vec_out::call(input, output_size.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*output_size)) : ::std::nullopt, scale_factors, out);
+  }
+}
+
+// aten::upsample_nearest2d.vec_out(Tensor input, SymInt[]? output_size, float[]? scale_factors, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest2d_outf(const at::Tensor & input, at::OptionalIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors, at::Tensor & out) {
+    return at::_ops::upsample_nearest2d_vec_out::call(input, output_size.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*output_size)) : ::std::nullopt, scale_factors, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
+  at::Tensor & upsample_nearest2d_outf(const at::Tensor & input, at::OptionalIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors, at::Tensor & out) {
+    return at::_ops::upsample_nearest2d_vec_out::call(input, output_size.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*output_size)) : ::std::nullopt, scale_factors, out);
+  }
+}
+
+// aten::upsample_nearest2d.vec_out(Tensor input, SymInt[]? output_size, float[]? scale_factors, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest2d_symint_out(at::Tensor & out, const at::Tensor & input, at::OptionalSymIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest2d_vec_out::call(input, output_size, scale_factors, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & upsample_nearest2d_out(at::Tensor & out, const at::Tensor & input, at::OptionalSymIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors) {
+    return at::_ops::upsample_nearest2d_vec_out::call(input, output_size, scale_factors, out);
+  }
+}
+
+// aten::upsample_nearest2d.vec_out(Tensor input, SymInt[]? output_size, float[]? scale_factors, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & upsample_nearest2d_symint_outf(const at::Tensor & input, at::OptionalSymIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors, at::Tensor & out) {
+    return at::_ops::upsample_nearest2d_vec_out::call(input, output_size, scale_factors, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
+  at::Tensor & upsample_nearest2d_outf(const at::Tensor & input, at::OptionalSymIntArrayRef output_size, ::std::optional<at::ArrayRef<double>> scale_factors, at::Tensor & out) {
+    return at::_ops::upsample_nearest2d_vec_out::call(input, output_size, scale_factors, out);
+  }
+}
+
+}
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest2d_backward_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest2d_backward_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..eef0d3653a9d984189567f04a52c95e2ed984891
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_nearest2d_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_upsample_nearest2d_backward : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & grad_output, at::ArrayRef<int64_t> output_size, at::ArrayRef<int64_t> input_size, ::std::optional<double> scales_h, ::std::optional<double> scales_w);
+};
+
+} // namespace native
+} // namespace at
diff --git a/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_trilinear3d_meta.h b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_trilinear3d_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..9180f967c26f0712c0c4651c15cd495111a58f4d
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/upsample_trilinear3d_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_trilinear3d : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, at::ArrayRef<int64_t> output_size, bool align_corners, ::std::optional<double> scales_d, ::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/view_as_ops.h b/lib/python3.12/site-packages/torch/include/ATen/ops/view_as_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..6a4c7c69451c564895461443b7d6cc3f4376c5e1
--- /dev/null
+++ b/lib/python3.12/site-packages/torch/include/ATen/ops/view_as_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API view_as {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::view_as";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "view_as(Tensor(a) self, Tensor other) -> Tensor(a)";
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & other);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other);
+};
+
+}} // namespace at::_ops