Add files using upload-large-folder tool

lib/python3.12/site-packages/torch/include/ATen/ops/_addmm_activation_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 _addmm_activation(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1, bool use_gelu=false);
+TORCH_API at::Tensor & _addmm_activation_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, bool use_gelu=false);
+TORCH_API at::Tensor & _addmm_activation_outf(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta, const at::Scalar & alpha, bool use_gelu, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_addmm_activation_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 _addmm_activation(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta=1, const at::Scalar & alpha=1, bool use_gelu=false);
+TORCH_API at::Tensor & _addmm_activation_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, bool use_gelu=false);
+TORCH_API at::Tensor & _addmm_activation_outf(const at::Tensor & self, const at::Tensor & mat1, const at::Tensor & mat2, const at::Scalar & beta, const at::Scalar & alpha, bool use_gelu, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_aminmax.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/_aminmax_ops.h>
+
+namespace at {
+
+
+// aten::_aminmax(Tensor self) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _aminmax(const at::Tensor & self) {
+    return at::_ops::_aminmax::call(self);
+}
+
+// aten::_aminmax.dim(Tensor self, int dim, bool keepdim=False) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _aminmax(const at::Tensor & self, int64_t dim, bool keepdim=false) {
+    return at::_ops::_aminmax_dim::call(self, dim, keepdim);
+}
+
+// aten::_aminmax.out(Tensor self, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _aminmax_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & self) {
+    return at::_ops::_aminmax_out::call(self, out0, out1);
+}
+// aten::_aminmax.out(Tensor self, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _aminmax_outf(const at::Tensor & self, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::_aminmax_out::call(self, out0, out1);
+}
+
+// aten::_aminmax.dim_out(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _aminmax_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & self, int64_t dim, bool keepdim=false) {
+    return at::_ops::_aminmax_dim_out::call(self, dim, keepdim, out0, out1);
+}
+// aten::_aminmax.dim_out(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _aminmax_outf(const at::Tensor & self, int64_t dim, bool keepdim, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::_aminmax_dim_out::call(self, dim, keepdim, out0, out1);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_assert_async_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 _assert_async {
+  using schema = void (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_assert_async";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_assert_async(Tensor self) -> ()";
+  static void call(const at::Tensor & self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+struct TORCH_API _assert_async_msg {
+  using schema = void (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::_assert_async";
+  static constexpr const char* overload_name = "msg";
+  static constexpr const char* schema_str = "_assert_async.msg(Tensor self, str assert_msg) -> ()";
+  static void call(const at::Tensor & self, c10::string_view assert_msg);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::string_view assert_msg);
+};
+
+}} // namespace at::_ops

lib/python3.12/site-packages/torch/include/ATen/ops/_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 void _assert_scalar(const at::Scalar & self, c10::string_view assert_msg);
+
+} // namespace compositeexplicitautograd
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_autocast_to_full_precision_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 _autocast_to_full_precision(const at::Tensor & self, bool cuda_enabled, bool cpu_enabled);
+} // namespace native
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_backward.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/_backward_ops.h>
+
+namespace at {
+
+
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_batch_norm_with_update.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/_batch_norm_with_update_ops.h>
+
+namespace at {
+
+
+// aten::_batch_norm_with_update(Tensor input, Tensor? weight, Tensor? bias, Tensor(a!) running_mean, Tensor(b!) running_var, float momentum, float eps) -> (Tensor, Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> _batch_norm_with_update(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) {
+    return at::_ops::_batch_norm_with_update::call(input, weight, bias, running_mean, running_var, momentum, eps);
+}
+
+// aten::_batch_norm_with_update.out(Tensor input, Tensor? weight, Tensor? bias, Tensor(a!) running_mean, Tensor(b!) running_var, float momentum, float eps, *, Tensor(d!) out, Tensor(e!) save_mean, Tensor(f!) save_invstd, Tensor(g!) reserve) -> (Tensor(d!), Tensor(e!), Tensor(f!), Tensor(g!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _batch_norm_with_update_out(at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd, at::Tensor & reserve, 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) {
+    return at::_ops::_batch_norm_with_update_out::call(input, weight, bias, running_mean, running_var, momentum, eps, out, save_mean, save_invstd, reserve);
+}
+// aten::_batch_norm_with_update.out(Tensor input, Tensor? weight, Tensor? bias, Tensor(a!) running_mean, Tensor(b!) running_var, float momentum, float eps, *, Tensor(d!) out, Tensor(e!) save_mean, Tensor(f!) save_invstd, Tensor(g!) reserve) -> (Tensor(d!), Tensor(e!), Tensor(f!), Tensor(g!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _batch_norm_with_update_outf(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, at::Tensor & running_mean, at::Tensor & running_var, double momentum, double eps, at::Tensor & out, at::Tensor & save_mean, at::Tensor & save_invstd, at::Tensor & reserve) {
+    return at::_ops::_batch_norm_with_update_out::call(input, weight, bias, running_mean, running_var, momentum, eps, out, save_mean, save_invstd, reserve);
+}
+
+// aten::_batch_norm_with_update_functional(Tensor input, Tensor? weight, Tensor? bias, Tensor running_mean, Tensor running_var, float momentum, float eps) -> (Tensor, Tensor, Tensor, Tensor, Tensor running_mean_out, Tensor running_var_out)
+inline ::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) {
+    return at::_ops::_batch_norm_with_update_functional::call(input, weight, bias, running_mean, running_var, momentum, eps);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_cast_Int_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_Int {
+  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_Int";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_cast_Int(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

lib/python3.12/site-packages/torch/include/ATen/ops/_compute_linear_combination_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 _compute_linear_combination {
+  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::_compute_linear_combination";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_compute_linear_combination(Tensor input, Tensor coefficients) -> Tensor";
+  static at::Tensor call(const at::Tensor & input, const at::Tensor & coefficients);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const at::Tensor & coefficients);
+};
+
+struct TORCH_API _compute_linear_combination_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::_compute_linear_combination";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_compute_linear_combination.out(Tensor input, Tensor coefficients, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & input, const at::Tensor & coefficients, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const at::Tensor & coefficients, at::Tensor & out);
+};
+
+}} // namespace at::_ops

lib/python3.12/site-packages/torch/include/ATen/ops/_convert_weight_to_int4pack_for_cpu_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 _convert_weight_to_int4pack_for_cpu(const at::Tensor & self, int64_t innerKTiles);
+
+} // namespace cpu
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_convolution_compositeexplicitautograd_dispatch.h

@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor _convolution(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32);
+TORCH_API at::Tensor _convolution_symint(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32);
+TORCH_API at::Tensor & _convolution_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32);
+TORCH_API at::Tensor & _convolution_outf(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool transposed, at::IntArrayRef output_padding, int64_t groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32, at::Tensor & out);
+TORCH_API at::Tensor & _convolution_symint_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32);
+TORCH_API at::Tensor & _convolution_symint_outf(const at::Tensor & input, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, bool transposed, c10::SymIntArrayRef output_padding, c10::SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_copy_from_and_resize.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/_copy_from_and_resize_ops.h>
+
+namespace at {
+
+
+// aten::_copy_from_and_resize(Tensor self, Tensor dst) -> Tensor
+inline at::Tensor _copy_from_and_resize(const at::Tensor & self, const at::Tensor & dst) {
+    return at::_ops::_copy_from_and_resize::call(self, dst);
+}
+
+// aten::_copy_from_and_resize.out(Tensor self, Tensor dst, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _copy_from_and_resize_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & dst) {
+    return at::_ops::_copy_from_and_resize_out::call(self, dst, out);
+}
+// aten::_copy_from_and_resize.out(Tensor self, Tensor dst, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _copy_from_and_resize_outf(const at::Tensor & self, const at::Tensor & dst, at::Tensor & out) {
+    return at::_ops::_copy_from_and_resize_out::call(self, dst, out);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_cslt_compress_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 _cslt_compress {
+  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::_cslt_compress";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_cslt_compress(Tensor input) -> Tensor";
+  static at::Tensor call(const at::Tensor & input);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input);
+};
+
+}} // namespace at::_ops

lib/python3.12/site-packages/torch/include/ATen/ops/_cudnn_init_dropout_state.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/_cudnn_init_dropout_state_ops.h>
+
+namespace at {
+
+
+// aten::_cudnn_init_dropout_state(float dropout, bool train, int dropout_seed, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor _cudnn_init_dropout_state(double dropout, bool train, int64_t dropout_seed, at::TensorOptions options) {
+    return at::_ops::_cudnn_init_dropout_state::call(dropout, train, dropout_seed, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::_cudnn_init_dropout_state(float dropout, bool train, int dropout_seed, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor
+inline at::Tensor _cudnn_init_dropout_state(double dropout, bool train, int64_t dropout_seed, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::_cudnn_init_dropout_state::call(dropout, train, dropout_seed, dtype, layout, device, pin_memory);
+}
+
+// aten::_cudnn_init_dropout_state.out(float dropout, bool train, int dropout_seed, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _cudnn_init_dropout_state_out(at::Tensor & out, double dropout, bool train, int64_t dropout_seed) {
+    return at::_ops::_cudnn_init_dropout_state_out::call(dropout, train, dropout_seed, out);
+}
+// aten::_cudnn_init_dropout_state.out(float dropout, bool train, int dropout_seed, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _cudnn_init_dropout_state_outf(double dropout, bool train, int64_t dropout_seed, at::Tensor & out) {
+    return at::_ops::_cudnn_init_dropout_state_out::call(dropout, train, dropout_seed, out);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_cummax_helper_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 _cummax_helper {
+  using schema = void (const at::Tensor &, at::Tensor &, at::Tensor &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_cummax_helper";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_cummax_helper(Tensor self, Tensor(a!) values, Tensor(b!) indices, int dim) -> ()";
+  static void call(const at::Tensor & self, at::Tensor & values, at::Tensor & indices, int64_t dim);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & values, at::Tensor & indices, int64_t dim);
+};
+
+}} // namespace at::_ops

lib/python3.12/site-packages/torch/include/ATen/ops/_dimI_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 sparse_dim_sparse(const at::Tensor & self);
+} // namespace native
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_dimI_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 _dimI {
+  using schema = int64_t (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_dimI";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_dimI(Tensor self) -> int";
+  static int64_t call(const at::Tensor & self);
+  static int64_t redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops

lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_dense_backward_cuda_dispatch.h

@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor _embedding_bag_dense_backward(const at::Tensor & grad, const at::Tensor & indices, const at::Tensor & offset2bag, const at::Tensor & bag_size, const at::Tensor & maximum_indices, int64_t num_weights, bool scale_grad_by_freq, int64_t mode, const ::std::optional<at::Tensor> & per_sample_weights, int64_t padding_idx=-1);
+TORCH_API at::Tensor _embedding_bag_dense_backward_symint(const at::Tensor & grad, const at::Tensor & indices, const at::Tensor & offset2bag, const at::Tensor & bag_size, const at::Tensor & maximum_indices, c10::SymInt num_weights, bool scale_grad_by_freq, int64_t mode, const ::std::optional<at::Tensor> & per_sample_weights, int64_t padding_idx=-1);
+
+} // namespace cuda
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_embedding_bag_forward_only.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/_embedding_bag_forward_only_ops.h>
+
+namespace at {
+
+
+// aten::_embedding_bag_forward_only(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)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> _embedding_bag_forward_only(const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq=false, int64_t mode=0, bool sparse=false, const ::std::optional<at::Tensor> & per_sample_weights={}, bool include_last_offset=false, int64_t padding_idx=-1) {
+    return at::_ops::_embedding_bag_forward_only::call(weight, indices, offsets, scale_grad_by_freq, mode, sparse, per_sample_weights, include_last_offset, padding_idx);
+}
+
+// aten::_embedding_bag_forward_only.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!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _embedding_bag_forward_only_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, bool scale_grad_by_freq=false, int64_t mode=0, bool sparse=false, const ::std::optional<at::Tensor> & per_sample_weights={}, bool include_last_offset=false, int64_t padding_idx=-1) {
+    return at::_ops::_embedding_bag_forward_only_out::call(weight, indices, offsets, scale_grad_by_freq, mode, sparse, per_sample_weights, include_last_offset, padding_idx, out0, out1, out2, out3);
+}
+// aten::_embedding_bag_forward_only.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!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> _embedding_bag_forward_only_outf(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) {
+    return at::_ops::_embedding_bag_forward_only_out::call(weight, indices, offsets, scale_grad_by_freq, mode, sparse, per_sample_weights, include_last_offset, padding_idx, out0, out1, out2, out3);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_tensor_affine_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 & _fake_quantize_learnable_per_tensor_affine_out(at::Tensor & out, 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);
+TORCH_API at::Tensor & _fake_quantize_learnable_per_tensor_affine_outf(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t quant_min, int64_t quant_max, double grad_factor, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_fake_quantize_learnable_per_tensor_affine_native.h

@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _fake_quantize_learnable_per_tensor_affine_out(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t quant_min, int64_t quant_max, double grad_factor, at::Tensor & out);
+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 native
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_fft_r2c_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 _fft_r2c_mkl(const at::Tensor & self, at::IntArrayRef dim, int64_t normalization, bool onesided);
+TORCH_API at::Tensor & _fft_r2c_mkl_out(const at::Tensor & self, at::IntArrayRef dim, int64_t normalization, bool onesided, at::Tensor & out);
+TORCH_API at::Tensor _fft_r2c_cufft(const at::Tensor & self, at::IntArrayRef dim, int64_t normalization, bool onesided);
+TORCH_API at::Tensor & _fft_r2c_cufft_out(const at::Tensor & self, at::IntArrayRef dim, int64_t normalization, bool onesided, at::Tensor & out);
+} // namespace native
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_fill_mem_eff_dropout_mask_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 & _fill_mem_eff_dropout_mask_(at::Tensor & self, double dropout_p, int64_t seed, int64_t offset);
+} // namespace native
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_foobar_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 & _foobar_out(at::Tensor & out, const at::Tensor & self, bool arg1=true, bool arg2=true, bool arg3=true);
+TORCH_API at::Tensor & _foobar_outf(const at::Tensor & self, bool arg1, bool arg2, bool arg3, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_abs_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_abs_slow(at::TensorList self);
+TORCH_API void _foreach_abs_out(at::TensorList self, at::TensorList out);
+TORCH_API void foreach_tensor_abs_slow_(at::TensorList self);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_abs_cuda(at::TensorList self);
+TORCH_API void foreach_tensor_abs_cuda_(at::TensorList self);
+} // namespace native
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_addcdiv.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_addcdiv_ops.h>
+
+namespace at {
+
+
+// aten::_foreach_addcdiv.Scalar(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Scalar value=1) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_addcdiv(at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, const at::Scalar & value=1) {
+    return at::_ops::_foreach_addcdiv_Scalar::call(self, tensor1, tensor2, value);
+}
+
+// aten::_foreach_addcdiv.ScalarList(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Scalar[] scalars) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_addcdiv(at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, at::ArrayRef<at::Scalar> scalars) {
+    return at::_ops::_foreach_addcdiv_ScalarList::call(self, tensor1, tensor2, scalars);
+}
+
+// aten::_foreach_addcdiv.Tensor(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Tensor scalars) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_addcdiv(at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, const at::Tensor & scalars) {
+    return at::_ops::_foreach_addcdiv_Tensor::call(self, tensor1, tensor2, scalars);
+}
+
+// aten::_foreach_addcdiv_.Scalar(Tensor(a!)[] self, Tensor[] tensor1, Tensor[] tensor2, Scalar value=1) -> ()
+inline void _foreach_addcdiv_(at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, const at::Scalar & value=1) {
+    return at::_ops::_foreach_addcdiv__Scalar::call(self, tensor1, tensor2, value);
+}
+
+// aten::_foreach_addcdiv_.ScalarList(Tensor(a!)[] self, Tensor[] tensor1, Tensor[] tensor2, Scalar[] scalars) -> ()
+inline void _foreach_addcdiv_(at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, at::ArrayRef<at::Scalar> scalars) {
+    return at::_ops::_foreach_addcdiv__ScalarList::call(self, tensor1, tensor2, scalars);
+}
+
+// aten::_foreach_addcdiv_.Tensor(Tensor(a!)[] self, Tensor[] tensor1, Tensor[] tensor2, Tensor scalars) -> ()
+inline void _foreach_addcdiv_(at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, const at::Tensor & scalars) {
+    return at::_ops::_foreach_addcdiv__Tensor::call(self, tensor1, tensor2, scalars);
+}
+
+// aten::_foreach_addcdiv.Scalar_out(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Scalar value=1, *, Tensor(a!)[] out) -> ()
+inline void _foreach_addcdiv_out(at::TensorList out, at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, const at::Scalar & value=1) {
+    return at::_ops::_foreach_addcdiv_Scalar_out::call(self, tensor1, tensor2, value, out);
+}
+// aten::_foreach_addcdiv.Scalar_out(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Scalar value=1, *, Tensor(a!)[] out) -> ()
+inline void _foreach_addcdiv_outf(at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, const at::Scalar & value, at::TensorList out) {
+    return at::_ops::_foreach_addcdiv_Scalar_out::call(self, tensor1, tensor2, value, out);
+}
+
+// aten::_foreach_addcdiv.ScalarList_out(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Scalar[] scalars, *, Tensor(a!)[] out) -> ()
+inline void _foreach_addcdiv_out(at::TensorList out, at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, at::ArrayRef<at::Scalar> scalars) {
+    return at::_ops::_foreach_addcdiv_ScalarList_out::call(self, tensor1, tensor2, scalars, out);
+}
+// aten::_foreach_addcdiv.ScalarList_out(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Scalar[] scalars, *, Tensor(a!)[] out) -> ()
+inline void _foreach_addcdiv_outf(at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, at::ArrayRef<at::Scalar> scalars, at::TensorList out) {
+    return at::_ops::_foreach_addcdiv_ScalarList_out::call(self, tensor1, tensor2, scalars, out);
+}
+
+// aten::_foreach_addcdiv.Tensor_out(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Tensor scalars, *, Tensor(a!)[] out) -> ()
+inline void _foreach_addcdiv_out(at::TensorList out, at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, const at::Tensor & scalars) {
+    return at::_ops::_foreach_addcdiv_Tensor_out::call(self, tensor1, tensor2, scalars, out);
+}
+// aten::_foreach_addcdiv.Tensor_out(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Tensor scalars, *, Tensor(a!)[] out) -> ()
+inline void _foreach_addcdiv_outf(at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, const at::Tensor & scalars, at::TensorList out) {
+    return at::_ops::_foreach_addcdiv_Tensor_out::call(self, tensor1, tensor2, scalars, out);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_lgamma_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_lgamma {
+  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_lgamma";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_lgamma(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_lgamma_ {
+  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_lgamma_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_lgamma_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_lgamma_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_lgamma";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_lgamma.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

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log10.h

@@ -0,0 +1,44 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_foreach_log10_ops.h>
+
+namespace at {
+
+
+// aten::_foreach_log10(Tensor[] self) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_log10(at::TensorList self) {
+    return at::_ops::_foreach_log10::call(self);
+}
+
+// aten::_foreach_log10_(Tensor(a!)[] self) -> ()
+inline void _foreach_log10_(at::TensorList self) {
+    return at::_ops::_foreach_log10_::call(self);
+}
+
+// aten::_foreach_log10.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_log10_out(at::TensorList out, at::TensorList self) {
+    return at::_ops::_foreach_log10_out::call(self, out);
+}
+// aten::_foreach_log10.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_log10_outf(at::TensorList self, at::TensorList out) {
+    return at::_ops::_foreach_log10_out::call(self, out);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log10_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_log10_slow(at::TensorList self);
+TORCH_API void _foreach_log10_out(at::TensorList self, at::TensorList out);
+TORCH_API void foreach_tensor_log10_slow_(at::TensorList self);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_log10_cuda(at::TensorList self);
+TORCH_API void foreach_tensor_log10_cuda_(at::TensorList self);
+} // namespace native
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_log_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_log(at::TensorList self);
+TORCH_API void _foreach_log_out(at::TensorList out, at::TensorList self);
+TORCH_API void _foreach_log_outf(at::TensorList self, at::TensorList out);
+TORCH_API void _foreach_log_(at::TensorList self);
+
+} // namespace compositeexplicitautograd
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_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 _foreach_log {
+  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_log";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_log(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_log_ {
+  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_log_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_log_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_log_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_log";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_log.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

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sigmoid.h

@@ -0,0 +1,44 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_foreach_sigmoid_ops.h>
+
+namespace at {
+
+
+// aten::_foreach_sigmoid(Tensor[] self) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_sigmoid(at::TensorList self) {
+    return at::_ops::_foreach_sigmoid::call(self);
+}
+
+// aten::_foreach_sigmoid_(Tensor(a!)[] self) -> ()
+inline void _foreach_sigmoid_(at::TensorList self) {
+    return at::_ops::_foreach_sigmoid_::call(self);
+}
+
+// aten::_foreach_sigmoid.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_sigmoid_out(at::TensorList out, at::TensorList self) {
+    return at::_ops::_foreach_sigmoid_out::call(self, out);
+}
+// aten::_foreach_sigmoid.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_sigmoid_outf(at::TensorList self, at::TensorList out) {
+    return at::_ops::_foreach_sigmoid_out::call(self, out);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sign_cuda_dispatch.h

@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::vector<at::Tensor> _foreach_sign(at::TensorList self);
+TORCH_API void _foreach_sign_(at::TensorList self);
+
+} // namespace cuda
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sinh.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_sinh_ops.h>
+
+namespace at {
+
+
+// aten::_foreach_sinh(Tensor[] self) -> Tensor[]
+inline ::std::vector<at::Tensor> _foreach_sinh(at::TensorList self) {
+    return at::_ops::_foreach_sinh::call(self);
+}
+
+// aten::_foreach_sinh_(Tensor(a!)[] self) -> ()
+inline void _foreach_sinh_(at::TensorList self) {
+    return at::_ops::_foreach_sinh_::call(self);
+}
+
+// aten::_foreach_sinh.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_sinh_out(at::TensorList out, at::TensorList self) {
+    return at::_ops::_foreach_sinh_out::call(self, out);
+}
+// aten::_foreach_sinh.out(Tensor[] self, *, Tensor(a!)[] out) -> ()
+inline void _foreach_sinh_outf(at::TensorList self, at::TensorList out) {
+    return at::_ops::_foreach_sinh_out::call(self, out);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_sub_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_sub(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_sub_out(at::TensorList out, at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_sub_outf(at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+TORCH_API void _foreach_sub_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> _foreach_sub(at::TensorList self, at::TensorList other, const at::Scalar & alpha=1);
+TORCH_API void _foreach_sub_out(at::TensorList out, at::TensorList self, at::TensorList other, const at::Scalar & alpha=1);
+TORCH_API void _foreach_sub_outf(at::TensorList self, at::TensorList other, const at::Scalar & alpha, at::TensorList out);
+TORCH_API void _foreach_sub_(at::TensorList self, at::TensorList other, const at::Scalar & alpha=1);
+TORCH_API ::std::vector<at::Tensor> _foreach_sub(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_sub_out(at::TensorList out, at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_sub_outf(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+TORCH_API void _foreach_sub_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+
+} // namespace compositeexplicitautograd
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_tanh_ops.h

@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _foreach_tanh {
+  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_tanh";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_tanh(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_tanh_ {
+  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_tanh_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_tanh_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_tanh_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_tanh";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_tanh.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

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_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 _foreach_trunc {
+  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_trunc";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_trunc(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_trunc_ {
+  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_trunc_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_foreach_trunc_(Tensor(a!)[] self) -> ()";
+  static void call(at::TensorList self);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self);
+};
+
+struct TORCH_API _foreach_trunc_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_trunc";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_foreach_trunc.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

lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adam.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_adam_ops.h>
+
+namespace at {
+
+
+// aten::_fused_adam_(Tensor(a!)[] self, Tensor(b!)[] grads, Tensor(c!)[] exp_avgs, Tensor(d!)[] exp_avg_sqs, Tensor(e!)[] max_exp_avg_sqs, Tensor[] state_steps, *, float lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None) -> ()
+inline void _fused_adam_(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, double lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={}) {
+    return at::_ops::_fused_adam_::call(self, grads, exp_avgs, exp_avg_sqs, max_exp_avg_sqs, state_steps, lr, beta1, beta2, weight_decay, eps, amsgrad, maximize, grad_scale, found_inf);
+}
+
+// aten::_fused_adam_.tensor_lr(Tensor(a!)[] self, Tensor(b!)[] grads, Tensor(c!)[] exp_avgs, Tensor(d!)[] exp_avg_sqs, Tensor(e!)[] max_exp_avg_sqs, Tensor[] state_steps, *, Tensor lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None) -> ()
+inline void _fused_adam_(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, const at::Tensor & lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={}) {
+    return at::_ops::_fused_adam__tensor_lr::call(self, grads, exp_avgs, exp_avg_sqs, max_exp_avg_sqs, state_steps, lr, beta1, beta2, weight_decay, eps, amsgrad, maximize, grad_scale, found_inf);
+}
+
+// aten::_fused_adam.out(Tensor[] self, Tensor(b!)[] grads, Tensor(c!)[] exp_avgs, Tensor(d!)[] exp_avg_sqs, Tensor(e!)[] max_exp_avg_sqs, Tensor[] state_steps, *, float lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None, Tensor(a!)[] out) -> ()
+inline void _fused_adam_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={}) {
+    return at::_ops::_fused_adam_out::call(self, grads, exp_avgs, exp_avg_sqs, max_exp_avg_sqs, state_steps, lr, beta1, beta2, weight_decay, eps, amsgrad, maximize, grad_scale, found_inf, out);
+}
+// aten::_fused_adam.out(Tensor[] self, Tensor(b!)[] grads, Tensor(c!)[] exp_avgs, Tensor(d!)[] exp_avg_sqs, Tensor(e!)[] max_exp_avg_sqs, Tensor[] state_steps, *, float lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None, Tensor(a!)[] out) -> ()
+inline void _fused_adam_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) {
+    return at::_ops::_fused_adam_out::call(self, grads, exp_avgs, exp_avg_sqs, max_exp_avg_sqs, state_steps, lr, beta1, beta2, weight_decay, eps, amsgrad, maximize, grad_scale, found_inf, out);
+}
+
+// aten::_fused_adam(Tensor[] self, Tensor[] grads, Tensor[] exp_avgs, Tensor[] exp_avg_sqs, Tensor[] max_exp_avg_sqs, Tensor[] state_steps, *, float lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None) -> (Tensor[] self_out, Tensor[] grads_out, Tensor[] exp_avgs_out, Tensor[] exp_avg_sqs_out, Tensor[] max_exp_avg_sqs_out)
+inline ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> _fused_adam(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, double lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={}) {
+    return at::_ops::_fused_adam::call(self, grads, exp_avgs, exp_avg_sqs, max_exp_avg_sqs, state_steps, lr, beta1, beta2, weight_decay, eps, amsgrad, maximize, grad_scale, found_inf);
+}
+
+// aten::_fused_adam.tensor_lr_out(Tensor[] self, Tensor(b!)[] grads, Tensor(c!)[] exp_avgs, Tensor(d!)[] exp_avg_sqs, Tensor(e!)[] max_exp_avg_sqs, Tensor[] state_steps, *, Tensor lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None, Tensor(a!)[] out) -> ()
+inline void _fused_adam_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={}) {
+    return at::_ops::_fused_adam_tensor_lr_out::call(self, grads, exp_avgs, exp_avg_sqs, max_exp_avg_sqs, state_steps, lr, beta1, beta2, weight_decay, eps, amsgrad, maximize, grad_scale, found_inf, out);
+}
+// aten::_fused_adam.tensor_lr_out(Tensor[] self, Tensor(b!)[] grads, Tensor(c!)[] exp_avgs, Tensor(d!)[] exp_avg_sqs, Tensor(e!)[] max_exp_avg_sqs, Tensor[] state_steps, *, Tensor lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None, Tensor(a!)[] out) -> ()
+inline void _fused_adam_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) {
+    return at::_ops::_fused_adam_tensor_lr_out::call(self, grads, exp_avgs, exp_avg_sqs, max_exp_avg_sqs, state_steps, lr, beta1, beta2, weight_decay, eps, amsgrad, maximize, grad_scale, found_inf, out);
+}
+
+// aten::_fused_adam.tensor_lr(Tensor[] self, Tensor[] grads, Tensor[] exp_avgs, Tensor[] exp_avg_sqs, Tensor[] max_exp_avg_sqs, Tensor[] state_steps, *, Tensor lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None) -> (Tensor[] self_out, Tensor[] grads_out, Tensor[] exp_avgs_out, Tensor[] exp_avg_sqs_out, Tensor[] max_exp_avg_sqs_out)
+inline ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> _fused_adam(at::TensorList self, at::TensorList grads, at::TensorList exp_avgs, at::TensorList exp_avg_sqs, at::TensorList max_exp_avg_sqs, at::TensorList state_steps, const at::Tensor & lr, double beta1, double beta2, double weight_decay, double eps, bool amsgrad, bool maximize, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={}) {
+    return at::_ops::_fused_adam_tensor_lr::call(self, grads, exp_avgs, exp_avg_sqs, max_exp_avg_sqs, state_steps, lr, beta1, beta2, weight_decay, eps, amsgrad, maximize, grad_scale, found_inf);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_fused_adamw_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 _fused_adamw_ {
+  using schema = void (at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, double, double, double, double, double, bool, bool, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fused_adamw_";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_fused_adamw_(Tensor(a!)[] self, Tensor(b!)[] grads, Tensor(c!)[] exp_avgs, Tensor(d!)[] exp_avg_sqs, Tensor(e!)[] max_exp_avg_sqs, Tensor[] state_steps, *, float lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None) -> ()";
+  static void call(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);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+struct TORCH_API _fused_adamw__tensor_lr {
+  using schema = void (at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, const at::Tensor &, double, double, double, double, bool, bool, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fused_adamw_";
+  static constexpr const char* overload_name = "tensor_lr";
+  static constexpr const char* schema_str = "_fused_adamw_.tensor_lr(Tensor(a!)[] self, Tensor(b!)[] grads, Tensor(c!)[] exp_avgs, Tensor(d!)[] exp_avg_sqs, Tensor(e!)[] max_exp_avg_sqs, Tensor[] state_steps, *, Tensor lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None) -> ()";
+  static void call(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);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+struct TORCH_API _fused_adamw_out {
+  using schema = void (at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, double, double, double, double, double, bool, bool, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fused_adamw";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_fused_adamw.out(Tensor[] self, Tensor(b!)[] grads, Tensor(c!)[] exp_avgs, Tensor(d!)[] exp_avg_sqs, Tensor(e!)[] max_exp_avg_sqs, Tensor[] state_steps, *, float lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None, Tensor(a!)[] out) -> ()";
+  static void call(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);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+struct TORCH_API _fused_adamw {
+  using schema = ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> (at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, double, double, double, double, double, bool, bool, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fused_adamw";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_fused_adamw(Tensor[] self, Tensor[] grads, Tensor[] exp_avgs, Tensor[] exp_avg_sqs, Tensor[] max_exp_avg_sqs, Tensor[] state_steps, *, float lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None) -> (Tensor[] self_out, Tensor[] grads_out, Tensor[] exp_avgs_out, Tensor[] exp_avg_sqs_out, Tensor[] max_exp_avg_sqs_out)";
+  static ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> call(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);
+  static ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+struct TORCH_API _fused_adamw_tensor_lr_out {
+  using schema = void (at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, const at::Tensor &, double, double, double, double, bool, bool, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fused_adamw";
+  static constexpr const char* overload_name = "tensor_lr_out";
+  static constexpr const char* schema_str = "_fused_adamw.tensor_lr_out(Tensor[] self, Tensor(b!)[] grads, Tensor(c!)[] exp_avgs, Tensor(d!)[] exp_avg_sqs, Tensor(e!)[] max_exp_avg_sqs, Tensor[] state_steps, *, Tensor lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None, Tensor(a!)[] out) -> ()";
+  static void call(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);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+struct TORCH_API _fused_adamw_tensor_lr {
+  using schema = ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> (at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, at::TensorList, const at::Tensor &, double, double, double, double, bool, bool, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_fused_adamw";
+  static constexpr const char* overload_name = "tensor_lr";
+  static constexpr const char* schema_str = "_fused_adamw.tensor_lr(Tensor[] self, Tensor[] grads, Tensor[] exp_avgs, Tensor[] exp_avg_sqs, Tensor[] max_exp_avg_sqs, Tensor[] state_steps, *, Tensor lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None) -> (Tensor[] self_out, Tensor[] grads_out, Tensor[] exp_avgs_out, Tensor[] exp_avg_sqs_out, Tensor[] max_exp_avg_sqs_out)";
+  static ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> call(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);
+  static ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> redispatch(c10::DispatchKeySet dispatchKeySet, 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 at::_ops

lib/python3.12/site-packages/torch/include/ATen/ops/_is_all_true_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 _is_all_true(const at::Tensor & self);
+
+} // namespace compositeexplicitautograd
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_jagged_to_padded_dense_forward_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 _jagged_to_padded_dense_forward {
+  using schema = at::Tensor (const at::Tensor &, at::TensorList, c10::SymIntArrayRef, double);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_jagged_to_padded_dense_forward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_jagged_to_padded_dense_forward(Tensor values, Tensor[] offsets, SymInt[] max_lengths, float padding_value=0.0) -> Tensor";
+  static at::Tensor call(const at::Tensor & values, at::TensorList offsets, c10::SymIntArrayRef max_lengths, double padding_value);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & values, at::TensorList offsets, c10::SymIntArrayRef max_lengths, double padding_value);
+};
+
+}} // namespace at::_ops

lib/python3.12/site-packages/torch/include/ATen/ops/_linalg_det_cpu_dispatch.h

@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _linalg_det(const at::Tensor & A);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _linalg_det_out(at::Tensor & result, at::Tensor & LU, at::Tensor & pivots, const at::Tensor & A);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> _linalg_det_outf(const at::Tensor & A, at::Tensor & result, at::Tensor & LU, at::Tensor & pivots);
+
+} // namespace cpu
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_local_scalar_dense_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::Scalar _local_scalar_dense_cpu(const at::Tensor & self);
+TORCH_API at::Scalar _local_scalar_dense_cuda(const at::Tensor & self);
+} // namespace native
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_local_scalar_dense_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 _local_scalar_dense {
+  using schema = at::Scalar (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_local_scalar_dense";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_local_scalar_dense(Tensor self) -> Scalar";
+  static at::Scalar call(const at::Tensor & self);
+  static at::Scalar redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops

lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_backward_data_compositeexplicitautogradnonfunctional_dispatch.h

@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor _log_softmax_backward_data(const at::Tensor & grad_output, const at::Tensor & output, int64_t dim, at::ScalarType input_dtype);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_log_softmax_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 _log_softmax(const at::Tensor & self, int64_t dim, bool half_to_float);
+TORCH_API at::Tensor & _log_softmax_out(at::Tensor & out, const at::Tensor & self, int64_t dim, bool half_to_float);
+TORCH_API at::Tensor & _log_softmax_outf(const at::Tensor & self, int64_t dim, bool half_to_float, at::Tensor & out);
+
+} // namespace cpu
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_masked_scale.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_scale_ops.h>
+
+namespace at {
+
+
+// aten::_masked_scale(Tensor self, Tensor mask, float scale) -> Tensor
+inline at::Tensor _masked_scale(const at::Tensor & self, const at::Tensor & mask, double scale) {
+    return at::_ops::_masked_scale::call(self, mask, scale);
+}
+
+// aten::_masked_scale.out(Tensor self, Tensor mask, float scale, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _masked_scale_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & mask, double scale) {
+    return at::_ops::_masked_scale_out::call(self, mask, scale, out);
+}
+// aten::_masked_scale.out(Tensor self, Tensor mask, float scale, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _masked_scale_outf(const at::Tensor & self, const at::Tensor & mask, double scale, at::Tensor & out) {
+    return at::_ops::_masked_scale_out::call(self, mask, scale, out);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_mkldnn_reshape.h

@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_mkldnn_reshape_ops.h>
+
+namespace at {
+
+
+// aten::_mkldnn_reshape(Tensor self, int[] shape) -> Tensor
+inline at::Tensor _mkldnn_reshape(const at::Tensor & self, at::IntArrayRef shape) {
+    return at::_ops::_mkldnn_reshape::call(self, shape);
+}
+
+// aten::_mkldnn_reshape.out(Tensor self, int[] shape, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _mkldnn_reshape_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef shape) {
+    return at::_ops::_mkldnn_reshape_out::call(self, shape, out);
+}
+// aten::_mkldnn_reshape.out(Tensor self, int[] shape, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _mkldnn_reshape_outf(const at::Tensor & self, at::IntArrayRef shape, at::Tensor & out) {
+    return at::_ops::_mkldnn_reshape_out::call(self, shape, out);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_nested_compute_contiguous_strides_offsets_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> _nested_compute_contiguous_strides_offsets(const at::Tensor & nested_size);
+} // namespace native
+} // namespace at