Add files using upload-large-folder tool

lib/python3.12/site-packages/torch/include/ATen/ops/_add_relu.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/_add_relu_ops.h>
+
+namespace at {
+
+
+// aten::_add_relu.Tensor(Tensor self, Tensor other, *, Scalar alpha=1) -> Tensor
+inline at::Tensor _add_relu(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1) {
+    return at::_ops::_add_relu_Tensor::call(self, other, alpha);
+}
+
+// aten::_add_relu_.Tensor(Tensor(a!) self, Tensor other, *, Scalar alpha=1) -> Tensor(a!)
+inline at::Tensor & _add_relu_(at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1) {
+    return at::_ops::_add_relu__Tensor::call(self, other, alpha);
+}
+
+// aten::_add_relu.out(Tensor self, Tensor other, *, Scalar alpha=1, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _add_relu_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha=1) {
+    return at::_ops::_add_relu_out::call(self, other, alpha, out);
+}
+// aten::_add_relu.out(Tensor self, Tensor other, *, Scalar alpha=1, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _add_relu_outf(const at::Tensor & self, const at::Tensor & other, const at::Scalar & alpha, at::Tensor & out) {
+    return at::_ops::_add_relu_out::call(self, other, alpha, out);
+}
+
+// aten::_add_relu.Scalar(Tensor self, Scalar other, Scalar alpha=1) -> Tensor
+inline at::Tensor _add_relu(const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha=1) {
+    return at::_ops::_add_relu_Scalar::call(self, other, alpha);
+}
+
+// aten::_add_relu_.Scalar(Tensor(a!) self, Scalar other, Scalar alpha=1) -> Tensor(a!)
+inline at::Tensor & _add_relu_(at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha=1) {
+    return at::_ops::_add_relu__Scalar::call(self, other, alpha);
+}
+
+// aten::_add_relu.Scalar_out(Tensor self, Scalar other, Scalar alpha=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _add_relu_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha=1) {
+    return at::_ops::_add_relu_Scalar_out::call(self, other, alpha, out);
+}
+// aten::_add_relu.Scalar_out(Tensor self, Scalar other, Scalar alpha=1, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _add_relu_outf(const at::Tensor & self, const at::Scalar & other, const at::Scalar & alpha, at::Tensor & out) {
+    return at::_ops::_add_relu_Scalar_out::call(self, other, alpha, out);
+}
+
+}

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

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

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

@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API void _backward(const at::Tensor & self, at::TensorList inputs, const ::std::optional<at::Tensor> & gradient={}, ::std::optional<bool> retain_graph=::std::nullopt, bool create_graph=false);
+} // namespace native
+} // namespace at

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

@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _batch_norm_impl_index_backward(int64_t impl_index, const at::Tensor & input, const at::Tensor & grad_output, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & running_mean, const ::std::optional<at::Tensor> & running_var, const ::std::optional<at::Tensor> & save_mean, const ::std::optional<at::Tensor> & save_var_transform, bool train, double eps, ::std::array<bool,3> output_mask, const at::Tensor & reservedSpace);
+} // namespace native
+} // namespace at

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

@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_choose_qparams_per_tensor_ops.h>
+
+namespace at {
+
+
+// aten::_choose_qparams_per_tensor(Tensor self, bool reduce_range=False) -> (float, int)
+inline ::std::tuple<double,int64_t> _choose_qparams_per_tensor(const at::Tensor & self, bool reduce_range=false) {
+    return at::_ops::_choose_qparams_per_tensor::call(self, reduce_range);
+}
+
+}

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

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

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

lib/python3.12/site-packages/torch/include/ATen/ops/_convert_weight_to_int4pack.h

@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_convert_weight_to_int4pack_ops.h>
+
+namespace at {
+
+
+// aten::_convert_weight_to_int4pack(Tensor self, int innerKTiles) -> Tensor
+inline at::Tensor _convert_weight_to_int4pack(const at::Tensor & self, int64_t innerKTiles) {
+    return at::_ops::_convert_weight_to_int4pack::call(self, innerKTiles);
+}
+
+}

lib/python3.12/site-packages/torch/include/ATen/ops/_copy_from.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_ops.h>
+
+namespace at {
+
+
+// aten::_copy_from(Tensor self, Tensor dst, bool non_blocking=False) -> Tensor
+inline at::Tensor _copy_from(const at::Tensor & self, const at::Tensor & dst, bool non_blocking=false) {
+    return at::_ops::_copy_from::call(self, dst, non_blocking);
+}
+
+// aten::_copy_from.out(Tensor self, Tensor dst, bool non_blocking=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _copy_from_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & dst, bool non_blocking=false) {
+    return at::_ops::_copy_from_out::call(self, dst, non_blocking, out);
+}
+// aten::_copy_from.out(Tensor self, Tensor dst, bool non_blocking=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _copy_from_outf(const at::Tensor & self, const at::Tensor & dst, bool non_blocking, at::Tensor & out) {
+    return at::_ops::_copy_from_out::call(self, dst, non_blocking, out);
+}
+
+}

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

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

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

@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> _efficient_attention_backward(const at::Tensor & grad_out_, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const at::Tensor & out, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, int64_t max_seqlen_q, int64_t max_seqlen_k, const at::Tensor & logsumexp, double dropout_p, const at::Tensor & philox_seed, const at::Tensor & philox_offset, int64_t custom_mask_type, bool bias_requires_grad, ::std::optional<double> scale=::std::nullopt, ::std::optional<int64_t> num_splits_key=::std::nullopt, ::std::optional<int64_t> window_size=::std::nullopt, bool shared_storage_dqdkdv=false);
+} // namespace native
+} // namespace at

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

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

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

@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor _embedding_bag_backward_symint(const at::Tensor & grad, const at::Tensor & indices, const at::Tensor & offsets, const at::Tensor & offset2bag, const at::Tensor & bag_size, const at::Tensor & maximum_indices, c10::SymInt num_weights, bool scale_grad_by_freq, int64_t mode, bool sparse, const ::std::optional<at::Tensor> & per_sample_weights, int64_t padding_idx=-1);
+} // namespace native
+} // namespace at

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

@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & _embedding_bag_dense_backward_out(at::Tensor & out, 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_outf(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, at::Tensor & out);
+TORCH_API at::Tensor & _embedding_bag_dense_backward_symint_out(at::Tensor & out, 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);
+TORCH_API at::Tensor & _embedding_bag_dense_backward_symint_outf(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, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at

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

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

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

@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _embedding_bag_dense_backward {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, c10::SymInt, bool, int64_t, const ::std::optional<at::Tensor> &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_embedding_bag_dense_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_embedding_bag_dense_backward(Tensor grad, Tensor indices, Tensor offset2bag, Tensor bag_size, Tensor maximum_indices, SymInt num_weights, bool scale_grad_by_freq, int mode, Tensor? per_sample_weights, int padding_idx=-1) -> Tensor";
+  static at::Tensor call(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);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+struct TORCH_API _embedding_bag_dense_backward_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, c10::SymInt, bool, int64_t, const ::std::optional<at::Tensor> &, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_embedding_bag_dense_backward";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_embedding_bag_dense_backward.out(Tensor grad, Tensor indices, Tensor offset2bag, Tensor bag_size, Tensor maximum_indices, SymInt num_weights, bool scale_grad_by_freq, int mode, Tensor? per_sample_weights, int padding_idx=-1, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(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, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, 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, at::Tensor & out);
+};
+
+}} // namespace at::_ops

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

@@ -0,0 +1,35 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_min_scalar_kernel_slow(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_clamp_min_Scalar_out(at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+TORCH_API void foreach_tensor_clamp_min_scalar_kernel_slow_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_min_scalar_kernel_cuda(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void foreach_tensor_clamp_min_scalar_kernel_cuda_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_min_list_kernel_slow(at::TensorList self, at::TensorList other);
+TORCH_API void _foreach_clamp_min_List_out(at::TensorList self, at::TensorList other, at::TensorList out);
+TORCH_API void foreach_tensor_clamp_min_list_kernel_slow_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_min_list_kernel_cuda(at::TensorList self, at::TensorList other);
+TORCH_API void foreach_tensor_clamp_min_list_kernel_cuda_(at::TensorList self, at::TensorList other);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_min_scalarlist_kernel_slow(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_clamp_min_ScalarList_out(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+TORCH_API void foreach_tensor_clamp_min_scalarlist_kernel_slow_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_clamp_min_scalarlist_kernel_cuda(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void foreach_tensor_clamp_min_scalarlist_kernel_cuda_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+} // namespace native
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_cos_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_cos_slow(at::TensorList self);
+TORCH_API void _foreach_cos_out(at::TensorList self, at::TensorList out);
+TORCH_API void foreach_tensor_cos_slow_(at::TensorList self);
+TORCH_API ::std::vector<at::Tensor> foreach_tensor_cos_cuda(at::TensorList self);
+TORCH_API void foreach_tensor_cos_cuda_(at::TensorList self);
+} // namespace native
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_foreach_erfc_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_erfc(at::TensorList self);
+TORCH_API void _foreach_erfc_(at::TensorList self);
+
+} // namespace cuda
+} // namespace at

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

@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::vector<at::Tensor> _foreach_norm(at::TensorList self, const at::Scalar & ord=2, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+
+} // namespace cuda
+} // namespace at

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

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

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

@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> _fused_sgd(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, double lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+TORCH_API void _fused_sgd_out(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, double lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale, const ::std::optional<at::Tensor> & found_inf, at::TensorList out);
+TORCH_API void _fused_sgd_kernel_cpu_(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, double lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+TORCH_API void _fused_sgd_kernel_cuda_(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, double lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+TORCH_API ::std::tuple<::std::vector<at::Tensor>,::std::vector<at::Tensor>,::std::vector<at::Tensor>> _fused_sgd(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, const at::Tensor & lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+TORCH_API void _fused_sgd_tensor_lr_out(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, const at::Tensor & lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale, const ::std::optional<at::Tensor> & found_inf, at::TensorList out);
+TORCH_API void _fused_sgd_kernel_cpu_(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, const at::Tensor & lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+TORCH_API void _fused_sgd_kernel_cuda_(at::TensorList self, at::TensorList grads, at::TensorList momentum_buffer_list, double weight_decay, double momentum, const at::Tensor & lr, double dampening, bool nesterov, bool maximize, bool is_first_step, const ::std::optional<at::Tensor> & grad_scale={}, const ::std::optional<at::Tensor> & found_inf={});
+} // namespace native
+} // namespace at

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

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

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

@@ -0,0 +1,44 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_index_put_impl_ops.h>
+
+namespace at {
+
+
+// aten::_index_put_impl_(Tensor(a!) self, Tensor?[] indices, Tensor values, bool accumulate=False, bool unsafe=False) -> Tensor(a!)
+inline at::Tensor & _index_put_impl_(at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false, bool unsafe=false) {
+    return at::_ops::_index_put_impl_::call(self, indices, values, accumulate, unsafe);
+}
+
+// aten::_index_put_impl.out(Tensor self, Tensor?[] indices, Tensor values, bool accumulate=False, bool unsafe=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _index_put_impl_out(at::Tensor & out, const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false, bool unsafe=false) {
+    return at::_ops::_index_put_impl_out::call(self, indices, values, accumulate, unsafe, out);
+}
+// aten::_index_put_impl.out(Tensor self, Tensor?[] indices, Tensor values, bool accumulate=False, bool unsafe=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _index_put_impl_outf(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate, bool unsafe, at::Tensor & out) {
+    return at::_ops::_index_put_impl_out::call(self, indices, values, accumulate, unsafe, out);
+}
+
+// aten::_index_put_impl(Tensor self, Tensor?[] indices, Tensor values, bool accumulate=False, bool unsafe=False) -> Tensor
+inline at::Tensor _index_put_impl(const at::Tensor & self, const c10::List<::std::optional<at::Tensor>> & indices, const at::Tensor & values, bool accumulate=false, bool unsafe=false) {
+    return at::_ops::_index_put_impl::call(self, indices, values, accumulate, unsafe);
+}
+
+}

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

@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_is_all_true_ops.h>
+
+namespace at {
+
+
+// aten::_is_all_true(Tensor self) -> Tensor
+inline at::Tensor _is_all_true(const at::Tensor & self) {
+    return at::_ops::_is_all_true::call(self);
+}
+
+}

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

@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _is_any_true {
+  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::_is_any_true";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_is_any_true(Tensor self) -> Tensor";
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+}} // namespace at::_ops

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

@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API void _linalg_check_errors(const at::Tensor & info, c10::string_view api_name, bool is_matrix);
+} // namespace native
+} // namespace at

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

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

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

@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _linalg_slogdet {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_linalg_slogdet";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_linalg_slogdet(Tensor A) -> (Tensor sign, Tensor logabsdet, Tensor LU, Tensor pivots)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & A);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A);
+};
+
+struct TORCH_API _linalg_slogdet_sign {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> (const at::Tensor &, at::Tensor &, at::Tensor &, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_linalg_slogdet";
+  static constexpr const char* overload_name = "sign";
+  static constexpr const char* schema_str = "_linalg_slogdet.sign(Tensor A, *, Tensor(a!) sign, Tensor(b!) logabsdet, Tensor(c!) LU, Tensor(d!) pivots) -> (Tensor(a!) sign, Tensor(b!) logabsdet, Tensor(c!) LU, Tensor(d!) pivots)";
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> call(const at::Tensor & A, at::Tensor & sign, at::Tensor & logabsdet, at::Tensor & LU, at::Tensor & pivots);
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & A, at::Tensor & sign, at::Tensor & logabsdet, at::Tensor & LU, at::Tensor & pivots);
+};
+
+}} // namespace at::_ops

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

@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> _native_batch_norm_legit_functional(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, bool training, double momentum, double eps);
+
+} // namespace compositeexplicitautograd
+} // namespace at

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

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

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

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

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

@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_nested_get_ragged_idx_ops.h>
+
+namespace at {
+
+
+// aten::_nested_get_ragged_idx(Tensor self) -> int
+inline int64_t _nested_get_ragged_idx(const at::Tensor & self) {
+    return at::_ops::_nested_get_ragged_idx::call(self);
+}
+
+}

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

@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API bool _nnpack_available();
+} // namespace native
+} // namespace at

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

@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _nnpack_available {
+  using schema = bool ();
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_nnpack_available";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_nnpack_available() -> bool";
+  static bool call();
+  static bool redispatch(c10::DispatchKeySet dispatchKeySet);
+};
+
+}} // namespace at::_ops

lib/python3.12/site-packages/torch/include/ATen/ops/_resize_output_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 _resize_output(const at::Tensor & self, at::IntArrayRef size, at::Device device);
+TORCH_API at::Tensor _resize_output_symint(const at::Tensor & self, c10::SymIntArrayRef size, at::Device device);
+TORCH_API const at::Tensor & _resize_output_out(const at::Tensor & out, const at::Tensor & self, at::IntArrayRef size, at::Device device);
+TORCH_API const at::Tensor & _resize_output_outf(const at::Tensor & self, at::IntArrayRef size, at::Device device, const at::Tensor & out);
+TORCH_API const at::Tensor & _resize_output_symint_out(const at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef size, at::Device device);
+TORCH_API const at::Tensor & _resize_output_symint_outf(const at::Tensor & self, c10::SymIntArrayRef size, at::Device device, const at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at

lib/python3.12/site-packages/torch/include/ATen/ops/_safe_softmax_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 _safe_softmax(const at::Tensor & self, int64_t dim, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+
+} // namespace compositeexplicitautograd
+} // namespace at

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

@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _scaled_dot_product_flash_attention_for_cpu_backward {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, double, bool, const ::std::optional<at::Tensor> &, ::std::optional<double>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_scaled_dot_product_flash_attention_for_cpu_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_scaled_dot_product_flash_attention_for_cpu_backward(Tensor grad_out, Tensor query, Tensor key, Tensor value, Tensor out, Tensor logsumexp, float dropout_p, bool is_causal, *, Tensor? attn_mask=None, float? scale=None) -> (Tensor grad_query, Tensor grad_key, Tensor grad_value)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & grad_out, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const at::Tensor & out, const at::Tensor & logsumexp, double dropout_p, bool is_causal, const ::std::optional<at::Tensor> & attn_mask, ::std::optional<double> scale);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_out, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const at::Tensor & out, const at::Tensor & logsumexp, double dropout_p, bool is_causal, const ::std::optional<at::Tensor> & attn_mask, ::std::optional<double> scale);
+};
+
+}} // namespace at::_ops

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

@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor _sparse_coo_tensor_unsafe_symint(const at::Tensor & indices, const at::Tensor & values, c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={}, ::std::optional<bool> is_coalesced=::std::nullopt);
+} // namespace native
+} // namespace at

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

@@ -0,0 +1,34 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_sparse_csc_tensor_unsafe_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_csc_tensor_unsafe(Tensor ccol_indices, Tensor row_indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor _sparse_csc_tensor_unsafe(const at::Tensor & ccol_indices, const at::Tensor & row_indices, const at::Tensor & values, at::IntArrayRef size, at::TensorOptions options={}) {
+    return at::_ops::_sparse_csc_tensor_unsafe::call(ccol_indices, row_indices, values, size, c10::optTypeMetaToScalarType(options.dtype_opt()), options.layout_opt(), options.device_opt(), options.pinned_memory_opt());
+}
+// aten::_sparse_csc_tensor_unsafe(Tensor ccol_indices, Tensor row_indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor
+inline at::Tensor _sparse_csc_tensor_unsafe(const at::Tensor & ccol_indices, const at::Tensor & row_indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory) {
+    return at::_ops::_sparse_csc_tensor_unsafe::call(ccol_indices, row_indices, values, size, dtype, layout, device, pin_memory);
+}
+
+}

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

@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_sparse_csr_sum_ops.h>
+
+namespace at {
+
+
+// aten::_sparse_csr_sum.dim_dtype(Tensor self, int[1] dim, bool keepdim=False, *, ScalarType? dtype=None) -> Tensor
+inline at::Tensor _sparse_csr_sum(const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt) {
+    return at::_ops::_sparse_csr_sum_dim_dtype::call(self, dim, keepdim, dtype);
+}
+
+// aten::_sparse_csr_sum.dim_dtype_out(Tensor self, int[1] dim, bool keepdim=False, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_csr_sum_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt) {
+    return at::_ops::_sparse_csr_sum_dim_dtype_out::call(self, dim, keepdim, dtype, out);
+}
+// aten::_sparse_csr_sum.dim_dtype_out(Tensor self, int[1] dim, bool keepdim=False, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _sparse_csr_sum_outf(const at::Tensor & self, at::IntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out) {
+    return at::_ops::_sparse_csr_sum_dim_dtype_out::call(self, dim, keepdim, dtype, out);
+}
+
+}

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

@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _sparse_mm_reduce_impl_backward {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, c10::string_view, const at::Tensor &, ::std::array<bool,2>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sparse_mm_reduce_impl_backward";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_sparse_mm_reduce_impl_backward(Tensor self, Tensor grad_out, Tensor weight, str reduce, Tensor arg_out, bool[2] output_mask) -> (Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self, const at::Tensor & grad_out, const at::Tensor & weight, c10::string_view reduce, const at::Tensor & arg_out, ::std::array<bool,2> output_mask);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & grad_out, const at::Tensor & weight, c10::string_view reduce, const at::Tensor & arg_out, ::std::array<bool,2> output_mask);
+};
+
+}} // namespace at::_ops

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

@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _sparse_semi_structured_apply {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sparse_semi_structured_apply";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_sparse_semi_structured_apply(Tensor input, Tensor thread_masks) -> (Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & input, const at::Tensor & thread_masks);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const at::Tensor & thread_masks);
+};
+
+}} // namespace at::_ops

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

@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor _sparse_semi_structured_linear(const at::Tensor & input, const at::Tensor & weight, const at::Tensor & meta, const ::std::optional<at::Tensor> & bias={}, ::std::optional<c10::string_view> activation=::std::nullopt, ::std::optional<at::ScalarType> out_dtype=::std::nullopt);
+} // namespace native
+} // namespace at

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

@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _sparse_semi_structured_tile {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, c10::string_view, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_sparse_semi_structured_tile";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_sparse_semi_structured_tile(Tensor input, str algorithm=\"\", bool use_cutlass=True) -> (Tensor, Tensor, Tensor, Tensor, Tensor)";
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & input, c10::string_view algorithm, bool use_cutlass);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, c10::string_view algorithm, bool use_cutlass);
+};
+
+}} // namespace at::_ops

lib/python3.12/site-packages/torch/include/ATen/ops/_spdiags_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 _spdiags {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, at::IntArrayRef, ::std::optional<at::Layout>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_spdiags";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_spdiags(Tensor diagonals, Tensor offsets, int[] shape, Layout? layout=None) -> Tensor";
+  static at::Tensor call(const at::Tensor & diagonals, const at::Tensor & offsets, at::IntArrayRef shape, ::std::optional<at::Layout> layout);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & diagonals, const at::Tensor & offsets, at::IntArrayRef shape, ::std::optional<at::Layout> layout);
+};
+
+struct TORCH_API _spdiags_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::IntArrayRef, ::std::optional<at::Layout>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_spdiags";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_spdiags.out(Tensor diagonals, Tensor offsets, int[] shape, Layout? layout=None, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & diagonals, const at::Tensor & offsets, at::IntArrayRef shape, ::std::optional<at::Layout> layout, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & diagonals, const at::Tensor & offsets, at::IntArrayRef shape, ::std::optional<at::Layout> layout, at::Tensor & out);
+};
+
+}} // namespace at::_ops

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

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

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

@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _test_optional_intlist {
+  using schema = at::Tensor (const at::Tensor &, at::OptionalIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_test_optional_intlist";
+  static constexpr const char* overload_name = "";
+  static constexpr const char* schema_str = "_test_optional_intlist(Tensor values, int[]? addends) -> Tensor";
+  static at::Tensor call(const at::Tensor & values, at::OptionalIntArrayRef addends);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & values, at::OptionalIntArrayRef addends);
+};
+
+struct TORCH_API _test_optional_intlist_out {
+  using schema = at::Tensor & (const at::Tensor &, at::OptionalIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  static constexpr const char* name = "aten::_test_optional_intlist";
+  static constexpr const char* overload_name = "out";
+  static constexpr const char* schema_str = "_test_optional_intlist.out(Tensor values, int[]? addends, *, Tensor(a!) out) -> Tensor(a!)";
+  static at::Tensor & call(const at::Tensor & values, at::OptionalIntArrayRef addends, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & values, at::OptionalIntArrayRef addends, at::Tensor & out);
+};
+
+}} // namespace at::_ops