Add files using upload-large-folder tool

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_coalesce_native.h

@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _coalesce_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor _coalesce_sparse_cpu(const at::Tensor & self);
+TORCH_API at::Tensor _coalesce_sparse_cuda(const at::Tensor & self);
+} // namespace native
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_compute_linear_combination.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/_compute_linear_combination_ops.h>
+
+namespace at {
+
+
+// aten::_compute_linear_combination(Tensor input, Tensor coefficients) -> Tensor
+inline at::Tensor _compute_linear_combination(const at::Tensor & input, const at::Tensor & coefficients) {
+    return at::_ops::_compute_linear_combination::call(input, coefficients);
+}
+
+// aten::_compute_linear_combination.out(Tensor input, Tensor coefficients, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _compute_linear_combination_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & coefficients) {
+    return at::_ops::_compute_linear_combination_out::call(input, coefficients, out);
+}
+// aten::_compute_linear_combination.out(Tensor input, Tensor coefficients, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _compute_linear_combination_outf(const at::Tensor & input, const at::Tensor & coefficients, at::Tensor & out) {
+    return at::_ops::_compute_linear_combination_out::call(input, coefficients, out);
+}
+
+}

.venv/lib/python3.11/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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_efficientzerotensor_meta_dispatch.h

@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor _efficientzerotensor(at::IntArrayRef size, at::TensorOptions options={});
+TORCH_API at::Tensor _efficientzerotensor(at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+TORCH_API at::Tensor _efficientzerotensor_symint(c10::SymIntArrayRef size, at::TensorOptions options={});
+TORCH_API at::Tensor _efficientzerotensor_symint(c10::SymIntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+
+} // namespace meta
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_embedding_bag_per_sample_weights_backward_cpu_dispatch.h

@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor _embedding_bag_per_sample_weights_backward(const at::Tensor & grad, const at::Tensor & weight, const at::Tensor & indices, const at::Tensor & offsets, const at::Tensor & offset2bag, int64_t mode, int64_t padding_idx=-1);
+
+} // namespace cpu
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_foreach_abs_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_abs(at::TensorList self);
+TORCH_API void _foreach_abs_out(at::TensorList out, at::TensorList self);
+TORCH_API void _foreach_abs_outf(at::TensorList self, at::TensorList out);
+TORCH_API void _foreach_abs_(at::TensorList self);
+
+} // namespace compositeexplicitautograd
+} // namespace at

.venv/lib/python3.11/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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_foreach_mul_ops.h

@@ -0,0 +1,149 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _foreach_mul_Scalar {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "Scalar")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul.Scalar(Tensor[] self, Scalar scalar) -> Tensor[]")
+  static ::std::vector<at::Tensor> call(at::TensorList self, const at::Scalar & scalar);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Scalar & scalar);
+};
+
+struct TORCH_API _foreach_mul__Scalar {
+  using schema = void (at::TensorList, const at::Scalar &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul_")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "Scalar")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul_.Scalar(Tensor(a!)[] self, Scalar scalar) -> ()")
+  static void call(at::TensorList self, const at::Scalar & scalar);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Scalar & scalar);
+};
+
+struct TORCH_API _foreach_mul_List {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "List")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul.List(Tensor[] self, Tensor[] other) -> Tensor[]")
+  static ::std::vector<at::Tensor> call(at::TensorList self, at::TensorList other);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::TensorList other);
+};
+
+struct TORCH_API _foreach_mul__List {
+  using schema = void (at::TensorList, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul_")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "List")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul_.List(Tensor(a!)[] self, Tensor[] other) -> ()")
+  static void call(at::TensorList self, at::TensorList other);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::TensorList other);
+};
+
+struct TORCH_API _foreach_mul_ScalarList {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, at::ArrayRef<at::Scalar>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "ScalarList")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul.ScalarList(Tensor[] self, Scalar[] scalars) -> Tensor[]")
+  static ::std::vector<at::Tensor> call(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+};
+
+struct TORCH_API _foreach_mul__ScalarList {
+  using schema = void (at::TensorList, at::ArrayRef<at::Scalar>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul_")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "ScalarList")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul_.ScalarList(Tensor(a!)[] self, Scalar[] scalars) -> ()")
+  static void call(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+};
+
+struct TORCH_API _foreach_mul_Tensor {
+  using schema = ::std::vector<at::Tensor> (at::TensorList, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "Tensor")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul.Tensor(Tensor[] self, Tensor other) -> Tensor[]")
+  static ::std::vector<at::Tensor> call(at::TensorList self, const at::Tensor & other);
+  static ::std::vector<at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Tensor & other);
+};
+
+struct TORCH_API _foreach_mul__Tensor {
+  using schema = void (at::TensorList, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul_")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "Tensor")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul_.Tensor(Tensor(a!)[] self, Tensor other) -> ()")
+  static void call(at::TensorList self, const at::Tensor & other);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Tensor & other);
+};
+
+struct TORCH_API _foreach_mul_Scalar_out {
+  using schema = void (at::TensorList, const at::Scalar &, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "Scalar_out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul.Scalar_out(Tensor[] self, Scalar scalar, *, Tensor(a!)[] out) -> ()")
+  static void call(at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+};
+
+struct TORCH_API _foreach_mul_List_out {
+  using schema = void (at::TensorList, at::TensorList, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "List_out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul.List_out(Tensor[] self, Tensor[] other, *, Tensor(a!)[] out) -> ()")
+  static void call(at::TensorList self, at::TensorList other, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::TensorList other, at::TensorList out);
+};
+
+struct TORCH_API _foreach_mul_ScalarList_out {
+  using schema = void (at::TensorList, at::ArrayRef<at::Scalar>, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "ScalarList_out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul.ScalarList_out(Tensor[] self, Scalar[] scalars, *, Tensor(a!)[] out) -> ()")
+  static void call(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+};
+
+struct TORCH_API _foreach_mul_Tensor_out {
+  using schema = void (at::TensorList, const at::Tensor &, at::TensorList);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_foreach_mul")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "Tensor_out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_foreach_mul.Tensor_out(Tensor[] self, Tensor other, *, Tensor(a!)[] out) -> ()")
+  static void call(at::TensorList self, const at::Tensor & other, at::TensorList out);
+  static void redispatch(c10::DispatchKeySet dispatchKeySet, at::TensorList self, const at::Tensor & other, at::TensorList out);
+};
+
+}} // namespace at::_ops

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_linalg_det_meta_dispatch.h

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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_linalg_eigvals_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 _linalg_eigvals(const at::Tensor & self);
+
+} // namespace cpu
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_make_dual_copy_compositeexplicitautograd_dispatch.h

@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & _make_dual_copy_out(at::Tensor & out, const at::Tensor & primal, const at::Tensor & tangent, int64_t level);
+TORCH_API at::Tensor & _make_dual_copy_outf(const at::Tensor & primal, const at::Tensor & tangent, int64_t level, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_native_batch_norm_legit_no_training_ops.h

@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _native_batch_norm_legit_no_training {
+  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor> (const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const at::Tensor &, const at::Tensor &, double, double);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_native_batch_norm_legit_no_training")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_native_batch_norm_legit_no_training(Tensor input, Tensor? weight, Tensor? bias, Tensor running_mean, Tensor running_var, float momentum, float eps) -> (Tensor, Tensor, Tensor)")
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> call(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, double momentum, double eps);
+  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, double momentum, double eps);
+};
+
+struct TORCH_API _native_batch_norm_legit_no_training_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> (const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const at::Tensor &, const at::Tensor &, double, double, at::Tensor &, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_native_batch_norm_legit_no_training")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_native_batch_norm_legit_no_training.out(Tensor input, Tensor? weight, Tensor? bias, Tensor running_mean, Tensor running_var, float momentum, float eps, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2) -> (Tensor(a!), Tensor(b!), Tensor(c!))")
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> call(const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, double momentum, double eps, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+  static ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, const at::Tensor & running_mean, const at::Tensor & running_var, double momentum, double eps, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+};
+
+}} // namespace at::_ops

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_scaled_dot_product_flash_attention_for_cpu.h

@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_scaled_dot_product_flash_attention_for_cpu_ops.h>
+
+namespace at {
+
+
+// aten::_scaled_dot_product_flash_attention_for_cpu(Tensor query, Tensor key, Tensor value, float dropout_p=0.0, bool is_causal=False, *, Tensor? attn_mask=None, float? scale=None) -> (Tensor output, Tensor logsumexp)
+inline ::std::tuple<at::Tensor,at::Tensor> _scaled_dot_product_flash_attention_for_cpu(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, double dropout_p=0.0, bool is_causal=false, const ::std::optional<at::Tensor> & attn_mask={}, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_flash_attention_for_cpu::call(query, key, value, dropout_p, is_causal, attn_mask, scale);
+}
+
+}

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_test_ambiguous_defaults.h

@@ -0,0 +1,35 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_test_ambiguous_defaults_ops.h>
+
+namespace at {
+
+
+// aten::_test_ambiguous_defaults.a(Tensor dummy, int a=1, int b=1) -> Tensor
+inline at::Tensor _test_ambiguous_defaults(const at::Tensor & dummy, int64_t a=1, int64_t b=1) {
+    return at::_ops::_test_ambiguous_defaults_a::call(dummy, a, b);
+}
+
+// aten::_test_ambiguous_defaults.b(Tensor dummy, int a=2, str b="2") -> Tensor
+inline at::Tensor _test_ambiguous_defaults(const at::Tensor & dummy, int64_t a, c10::string_view b) {
+    return at::_ops::_test_ambiguous_defaults_b::call(dummy, a, b);
+}
+
+}

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_test_optional_filled_intlist.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/_test_optional_filled_intlist_ops.h>
+
+namespace at {
+
+
+// aten::_test_optional_filled_intlist(Tensor values, int[2]? addends) -> Tensor
+inline at::Tensor _test_optional_filled_intlist(const at::Tensor & values, at::OptionalIntArrayRef addends) {
+    return at::_ops::_test_optional_filled_intlist::call(values, addends);
+}
+
+// aten::_test_optional_filled_intlist.out(Tensor values, int[2]? addends, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _test_optional_filled_intlist_out(at::Tensor & out, const at::Tensor & values, at::OptionalIntArrayRef addends) {
+    return at::_ops::_test_optional_filled_intlist_out::call(values, addends, out);
+}
+// aten::_test_optional_filled_intlist.out(Tensor values, int[2]? addends, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _test_optional_filled_intlist_outf(const at::Tensor & values, at::OptionalIntArrayRef addends, at::Tensor & out) {
+    return at::_ops::_test_optional_filled_intlist_out::call(values, addends, out);
+}
+
+}

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_trilinear_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 _trilinear {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_trilinear")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_trilinear(Tensor i1, Tensor i2, Tensor i3, int[] expand1, int[] expand2, int[] expand3, int[] sumdim, int unroll_dim=1) -> Tensor")
+  static at::Tensor call(const at::Tensor & i1, const at::Tensor & i2, const at::Tensor & i3, at::IntArrayRef expand1, at::IntArrayRef expand2, at::IntArrayRef expand3, at::IntArrayRef sumdim, int64_t unroll_dim);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & i1, const at::Tensor & i2, const at::Tensor & i3, at::IntArrayRef expand1, at::IntArrayRef expand2, at::IntArrayRef expand3, at::IntArrayRef sumdim, int64_t unroll_dim);
+};
+
+struct TORCH_API _trilinear_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_trilinear")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_trilinear.out(Tensor i1, Tensor i2, Tensor i3, int[] expand1, int[] expand2, int[] expand3, int[] sumdim, int unroll_dim=1, *, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & i1, const at::Tensor & i2, const at::Tensor & i3, at::IntArrayRef expand1, at::IntArrayRef expand2, at::IntArrayRef expand3, at::IntArrayRef sumdim, int64_t unroll_dim, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & i1, const at::Tensor & i2, const at::Tensor & i3, at::IntArrayRef expand1, at::IntArrayRef expand2, at::IntArrayRef expand3, at::IntArrayRef sumdim, int64_t unroll_dim, at::Tensor & out);
+};
+
+}} // namespace at::_ops

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_triton_multi_head_attention_cuda_dispatch.h

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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_upsample_bilinear2d_aa_backward_cuda_dispatch.h

@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor _upsample_bilinear2d_aa_backward(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor _upsample_bilinear2d_aa_backward_symint(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_bilinear2d_aa_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_bilinear2d_aa_backward_outf(const at::Tensor & grad_output, at::IntArrayRef output_size, at::IntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input);
+TORCH_API at::Tensor & _upsample_bilinear2d_aa_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor & _upsample_bilinear2d_aa_backward_symint_outf(const at::Tensor & grad_output, c10::SymIntArrayRef output_size, c10::SymIntArrayRef input_size, bool align_corners, ::std::optional<double> scales_h, ::std::optional<double> scales_w, at::Tensor & grad_input);
+
+} // namespace cuda
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_upsample_nearest_exact1d_meta.h

@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured__upsample_nearest_exact1d : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, at::ArrayRef<int64_t> output_size, ::std::optional<double> scales);
+};
+
+} // namespace native
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_validate_sparse_coo_tensor_args.h

@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_validate_sparse_coo_tensor_args_ops.h>
+
+namespace at {
+
+
+// aten::_validate_sparse_coo_tensor_args(Tensor indices, Tensor values, int[] size, bool? is_coalesced=None) -> ()
+inline void _validate_sparse_coo_tensor_args(const at::Tensor & indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<bool> is_coalesced=::std::nullopt) {
+    return at::_ops::_validate_sparse_coo_tensor_args::call(indices, values, size, is_coalesced);
+}
+
+}

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/all.h

@@ -0,0 +1,81 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/all_ops.h>
+
+namespace at {
+
+
+// aten::all.dim(Tensor self, int dim, bool keepdim=False) -> Tensor
+inline at::Tensor all(const at::Tensor & self, int64_t dim, bool keepdim=false) {
+    return at::_ops::all_dim::call(self, dim, keepdim);
+}
+
+// aten::all.dims(Tensor self, int[]? dim=None, bool keepdim=False) -> Tensor
+inline at::Tensor all(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim=false) {
+    return at::_ops::all_dims::call(self, dim, keepdim);
+}
+
+// aten::all.out(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & all_out(at::Tensor & out, const at::Tensor & self, int64_t dim, bool keepdim=false) {
+    return at::_ops::all_out::call(self, dim, keepdim, out);
+}
+// aten::all.out(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & all_outf(const at::Tensor & self, int64_t dim, bool keepdim, at::Tensor & out) {
+    return at::_ops::all_out::call(self, dim, keepdim, out);
+}
+
+// aten::all.dims_out(Tensor self, int[]? dim=None, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & all_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim=false) {
+    return at::_ops::all_dims_out::call(self, dim, keepdim, out);
+}
+// aten::all.dims_out(Tensor self, int[]? dim=None, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & all_outf(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim, at::Tensor & out) {
+    return at::_ops::all_dims_out::call(self, dim, keepdim, out);
+}
+
+// aten::all.dimname(Tensor self, Dimname dim, bool keepdim=False) -> Tensor
+inline at::Tensor all(const at::Tensor & self, at::Dimname dim, bool keepdim=false) {
+    return at::_ops::all_dimname::call(self, dim, keepdim);
+}
+
+// aten::all.dimname_out(Tensor self, Dimname dim, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & all_out(at::Tensor & out, const at::Tensor & self, at::Dimname dim, bool keepdim=false) {
+    return at::_ops::all_dimname_out::call(self, dim, keepdim, out);
+}
+// aten::all.dimname_out(Tensor self, Dimname dim, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & all_outf(const at::Tensor & self, at::Dimname dim, bool keepdim, at::Tensor & out) {
+    return at::_ops::all_dimname_out::call(self, dim, keepdim, out);
+}
+
+// aten::all(Tensor self) -> Tensor
+inline at::Tensor all(const at::Tensor & self) {
+    return at::_ops::all::call(self);
+}
+
+// aten::all.all_out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & all_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::all_all_out::call(self, out);
+}
+// aten::all.all_out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & all_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::all_all_out::call(self, out);
+}
+
+}

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/avg_pool2d_backward_ops.h

@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API avg_pool2d_backward_grad_input {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, bool, ::std::optional<int64_t>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::avg_pool2d_backward")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "grad_input")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "avg_pool2d_backward.grad_input(Tensor grad_output, Tensor self, int[2] kernel_size, int[2] stride, int[2] padding, bool ceil_mode, bool count_include_pad, int? divisor_override, *, Tensor(a!) grad_input) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override, at::Tensor & grad_input);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override, at::Tensor & grad_input);
+};
+
+struct TORCH_API avg_pool2d_backward {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool, bool, ::std::optional<int64_t>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::avg_pool2d_backward")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "avg_pool2d_backward(Tensor grad_output, Tensor self, int[2] kernel_size, int[2] stride, int[2] padding, bool ceil_mode, bool count_include_pad, int? divisor_override) -> Tensor")
+  static at::Tensor call(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode, bool count_include_pad, ::std::optional<int64_t> divisor_override);
+};
+
+}} // namespace at::_ops

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/baddbmm_cuda_dispatch.h

@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API at::Tensor baddbmm(const at::Tensor & self, const at::Tensor & batch1, const at::Tensor & batch2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & baddbmm_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & batch1, const at::Tensor & batch2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+TORCH_API at::Tensor & baddbmm_outf(const at::Tensor & self, const at::Tensor & batch1, const at::Tensor & batch2, const at::Scalar & beta, const at::Scalar & alpha, at::Tensor & out);
+TORCH_API at::Tensor & baddbmm_(at::Tensor & self, const at::Tensor & batch1, const at::Tensor & batch2, const at::Scalar & beta=1, const at::Scalar & alpha=1);
+
+} // namespace cuda
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/cauchy.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/cauchy_ops.h>
+
+namespace at {
+
+
+// aten::cauchy.out(Tensor self, float median=0, float sigma=1, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cauchy_out(at::Tensor & out, const at::Tensor & self, double median=0, double sigma=1, ::std::optional<at::Generator> generator=::std::nullopt) {
+    return at::_ops::cauchy_out::call(self, median, sigma, generator, out);
+}
+// aten::cauchy.out(Tensor self, float median=0, float sigma=1, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & cauchy_outf(const at::Tensor & self, double median, double sigma, ::std::optional<at::Generator> generator, at::Tensor & out) {
+    return at::_ops::cauchy_out::call(self, median, sigma, generator, out);
+}
+
+// aten::cauchy(Tensor self, float median=0, float sigma=1, *, Generator? generator=None) -> Tensor
+inline at::Tensor cauchy(const at::Tensor & self, double median=0, double sigma=1, ::std::optional<at::Generator> generator=::std::nullopt) {
+    return at::_ops::cauchy::call(self, median, sigma, generator);
+}
+
+}

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/combinations_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 combinations(const at::Tensor & self, int64_t r=2, bool with_replacement=false);
+} // namespace native
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/copy_sparse_to_sparse_native.h

@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor copy_sparse_to_sparse(const at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+TORCH_API at::Tensor & copy_sparse_to_sparse_out(const at::Tensor & self, const at::Tensor & src, bool non_blocking, at::Tensor & out);
+TORCH_API at::Tensor & copy_sparse_(at::Tensor & self, const at::Tensor & src, bool non_blocking=false);
+} // namespace native
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/cumsum_compositeexplicitautogradnonfunctional_dispatch.h

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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/embedding_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 embedding(const at::Tensor & weight, const at::Tensor & indices, int64_t padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false);
+TORCH_API at::Tensor embedding_symint(const at::Tensor & weight, const at::Tensor & indices, c10::SymInt padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false);
+TORCH_API at::Tensor & embedding_out(at::Tensor & out, const at::Tensor & weight, const at::Tensor & indices, int64_t padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false);
+TORCH_API at::Tensor & embedding_outf(const at::Tensor & weight, const at::Tensor & indices, int64_t padding_idx, bool scale_grad_by_freq, bool sparse, at::Tensor & out);
+TORCH_API at::Tensor & embedding_symint_out(at::Tensor & out, const at::Tensor & weight, const at::Tensor & indices, c10::SymInt padding_idx=-1, bool scale_grad_by_freq=false, bool sparse=false);
+TORCH_API at::Tensor & embedding_symint_outf(const at::Tensor & weight, const at::Tensor & indices, c10::SymInt padding_idx, bool scale_grad_by_freq, bool sparse, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/float_power_native.h

@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor float_power(const at::Tensor & self, const at::Tensor & exponent);
+TORCH_API at::Tensor & float_power_out(const at::Tensor & self, const at::Tensor & exponent, at::Tensor & out);
+TORCH_API at::Tensor & float_power_(at::Tensor & self, const at::Tensor & exponent);
+TORCH_API at::Tensor float_power(const at::Scalar & self, const at::Tensor & exponent);
+TORCH_API at::Tensor & float_power_out(const at::Scalar & self, const at::Tensor & exponent, at::Tensor & out);
+TORCH_API at::Tensor float_power(const at::Tensor & self, const at::Scalar & exponent);
+TORCH_API at::Tensor & float_power_out(const at::Tensor & self, const at::Scalar & exponent, at::Tensor & out);
+TORCH_API at::Tensor & float_power_(at::Tensor & self, const at::Scalar & exponent);
+} // namespace native
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fractional_max_pool2d_native.h

@@ -0,0 +1,26 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+#include <ATen/ops/fractional_max_pool2d_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_fractional_max_pool2d_out_cpu : public at::meta::structured_fractional_max_pool2d {
+void impl(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & random_samples, const at::Tensor & output, const at::Tensor & indices);
+};
+struct TORCH_API structured_fractional_max_pool2d_out_cuda : public at::meta::structured_fractional_max_pool2d {
+void impl(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef output_size, const at::Tensor & random_samples, const at::Tensor & output, const at::Tensor & indices);
+};
+} // namespace native
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/hardshrink_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 hardshrink(const at::Tensor & self, const at::Scalar & lambd=0.5);
+TORCH_API at::Tensor & hardshrink_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & lambd=0.5);
+TORCH_API at::Tensor & hardshrink_outf(const at::Tensor & self, const at::Scalar & lambd, at::Tensor & out);
+
+} // namespace cuda
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/hardsigmoid_backward_compositeexplicitautogradnonfunctional_dispatch.h

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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/hardsigmoid_cuda_dispatch.h

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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/hardswish_cuda_dispatch.h

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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/is_leaf_native.h

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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/kthvalue_compositeimplicitautograd_dispatch.h

@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> kthvalue(const at::Tensor & self, int64_t k, at::Dimname dim, bool keepdim=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> kthvalue_out(at::Tensor & values, at::Tensor & indices, const at::Tensor & self, int64_t k, at::Dimname dim, bool keepdim=false);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> kthvalue_outf(const at::Tensor & self, int64_t k, at::Dimname dim, bool keepdim, at::Tensor & values, at::Tensor & indices);
+
+} // namespace compositeimplicitautograd
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/ldexp_native.h

@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor ldexp(const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & ldexp_out(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+TORCH_API at::Tensor & ldexp_(at::Tensor & self, const at::Tensor & other);
+} // namespace native
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/leaky_relu_meta_dispatch.h

@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor leaky_relu(const at::Tensor & self, const at::Scalar & negative_slope=0.01);
+TORCH_API at::Tensor & leaky_relu_out(at::Tensor & out, const at::Tensor & self, const at::Scalar & negative_slope=0.01);
+TORCH_API at::Tensor & leaky_relu_outf(const at::Tensor & self, const at::Scalar & negative_slope, at::Tensor & out);
+TORCH_API at::Tensor & leaky_relu_(at::Tensor & self, const at::Scalar & negative_slope=0.01);
+
+} // namespace meta
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/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 lgamma_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_STR_INL_EXCEPT_WIN_CUDA(name, "aten::lgamma")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "lgamma.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+struct TORCH_API lgamma_ {
+  using schema = at::Tensor & (at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::lgamma_")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "lgamma_(Tensor(a!) self) -> Tensor(a!)")
+  static at::Tensor & call(at::Tensor & self);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self);
+};
+
+struct TORCH_API lgamma {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::lgamma")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "lgamma(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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/linalg_diagonal.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/linalg_diagonal_ops.h>
+
+namespace at {
+
+
+// aten::linalg_diagonal(Tensor(a) A, *, int offset=0, int dim1=-2, int dim2=-1) -> Tensor(a)
+inline at::Tensor linalg_diagonal(const at::Tensor & A, int64_t offset=0, int64_t dim1=-2, int64_t dim2=-1) {
+    return at::_ops::linalg_diagonal::call(A, offset, dim1, dim2);
+}
+
+}

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/linalg_qr_cuda_dispatch.h

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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/linalg_svd_compositeimplicitautograd_dispatch.h

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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/logsumexp_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 logsumexp(const at::Tensor & self, at::IntArrayRef dim, bool keepdim=false);
+
+} // namespace compositeexplicitautograd
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/lshift_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 & __lshift___out(at::Tensor & out, const at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & __lshift___outf(const at::Tensor & self, const at::Scalar & other, at::Tensor & out);
+TORCH_API at::Tensor & __lshift___out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other);
+TORCH_API at::Tensor & __lshift___outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/mean_compositeexplicitautogradnonfunctional_dispatch.h

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

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/miopen_convolution_transpose_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 miopen_convolution_transpose {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymInt, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::miopen_convolution_transpose")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "miopen_convolution_transpose(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, SymInt[] output_padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic) -> Tensor")
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic);
+};
+
+struct TORCH_API miopen_convolution_transpose_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymInt, bool, bool, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::miopen_convolution_transpose")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "miopen_convolution_transpose.out(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, SymInt[] output_padding, SymInt[] stride, SymInt[] dilation, SymInt groups, bool benchmark, bool deterministic, *, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, bool benchmark, bool deterministic, at::Tensor & out);
+};
+
+}} // namespace at::_ops

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/miopen_rnn.h

@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/miopen_rnn_ops.h>
+
+namespace at {
+
+
+// aten::miopen_rnn(Tensor input, Tensor[] weight, int weight_stride0, Tensor hx, Tensor? cx, int mode, int hidden_size, int num_layers, bool batch_first, float dropout, bool train, bool bidirectional, int[] batch_sizes, Tensor? dropout_state) -> (Tensor, Tensor, Tensor, Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,at::Tensor> miopen_rnn(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, int64_t mode, int64_t hidden_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, at::IntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state) {
+    return at::_ops::miopen_rnn::call(input, weight, weight_stride0, hx, cx, mode, hidden_size, num_layers, batch_first, dropout, train, bidirectional, batch_sizes, dropout_state);
+}
+
+// aten::miopen_rnn.out(Tensor input, Tensor[] weight, int weight_stride0, Tensor hx, Tensor? cx, int mode, int hidden_size, int num_layers, bool batch_first, float dropout, bool train, bool bidirectional, int[] batch_sizes, Tensor? dropout_state, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2, Tensor(d!) out3, Tensor(e!) out4) -> (Tensor(a!), Tensor(b!), Tensor(c!), Tensor(d!), Tensor(e!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> miopen_rnn_out(at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, at::Tensor & out4, const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, int64_t mode, int64_t hidden_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, at::IntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state) {
+    return at::_ops::miopen_rnn_out::call(input, weight, weight_stride0, hx, cx, mode, hidden_size, num_layers, batch_first, dropout, train, bidirectional, batch_sizes, dropout_state, out0, out1, out2, out3, out4);
+}
+// aten::miopen_rnn.out(Tensor input, Tensor[] weight, int weight_stride0, Tensor hx, Tensor? cx, int mode, int hidden_size, int num_layers, bool batch_first, float dropout, bool train, bool bidirectional, int[] batch_sizes, Tensor? dropout_state, *, Tensor(a!) out0, Tensor(b!) out1, Tensor(c!) out2, Tensor(d!) out3, Tensor(e!) out4) -> (Tensor(a!), Tensor(b!), Tensor(c!), Tensor(d!), Tensor(e!))
+inline ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &,at::Tensor &> miopen_rnn_outf(const at::Tensor & input, at::TensorList weight, int64_t weight_stride0, const at::Tensor & hx, const ::std::optional<at::Tensor> & cx, int64_t mode, int64_t hidden_size, int64_t num_layers, bool batch_first, double dropout, bool train, bool bidirectional, at::IntArrayRef batch_sizes, const ::std::optional<at::Tensor> & dropout_state, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2, at::Tensor & out3, at::Tensor & out4) {
+    return at::_ops::miopen_rnn_out::call(input, weight, weight_stride0, hx, cx, mode, hidden_size, num_layers, batch_first, dropout, train, bidirectional, batch_sizes, dropout_state, out0, out1, out2, out3, out4);
+}
+
+}

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/mkldnn_reorder_conv2d_weight_compositeexplicitautograd_dispatch.h

@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API at::Tensor & mkldnn_reorder_conv2d_weight_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef padding=0, at::IntArrayRef stride=1, at::IntArrayRef dilation=1, int64_t groups=1, at::OptionalIntArrayRef input_size=::std::nullopt);
+TORCH_API at::Tensor & mkldnn_reorder_conv2d_weight_outf(const at::Tensor & self, at::IntArrayRef padding, at::IntArrayRef stride, at::IntArrayRef dilation, int64_t groups, at::OptionalIntArrayRef input_size, at::Tensor & out);
+TORCH_API at::Tensor & mkldnn_reorder_conv2d_weight_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef padding=c10::SymInt(0), c10::SymIntArrayRef stride=c10::SymInt(1), c10::SymIntArrayRef dilation=c10::SymInt(1), c10::SymInt groups=1, at::OptionalSymIntArrayRef input_size=::std::nullopt);
+TORCH_API at::Tensor & mkldnn_reorder_conv2d_weight_symint_outf(const at::Tensor & self, c10::SymIntArrayRef padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, at::OptionalSymIntArrayRef input_size, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/multi_margin_loss_backward_ops.h

@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API multi_margin_loss_backward_grad_input {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Scalar &, const at::Scalar &, const ::std::optional<at::Tensor> &, int64_t, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::multi_margin_loss_backward")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "grad_input")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "multi_margin_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Scalar p, Scalar margin, Tensor? weight=None, int reduction=Mean, *, Tensor(a!) grad_input) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const at::Scalar & p, const at::Scalar & margin, const ::std::optional<at::Tensor> & weight, int64_t reduction, at::Tensor & grad_input);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const at::Scalar & p, const at::Scalar & margin, const ::std::optional<at::Tensor> & weight, int64_t reduction, at::Tensor & grad_input);
+};
+
+struct TORCH_API multi_margin_loss_backward {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Scalar &, const at::Scalar &, const ::std::optional<at::Tensor> &, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::multi_margin_loss_backward")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "multi_margin_loss_backward(Tensor grad_output, Tensor self, Tensor target, Scalar p, Scalar margin, Tensor? weight=None, int reduction=Mean) -> Tensor")
+  static at::Tensor call(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const at::Scalar & p, const at::Scalar & margin, const ::std::optional<at::Tensor> & weight, int64_t reduction);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & target, const at::Scalar & p, const at::Scalar & margin, const ::std::optional<at::Tensor> & weight, int64_t reduction);
+};
+
+}} // namespace at::_ops

.venv/lib/python3.11/site-packages/torch/include/ATen/ops/native_layer_norm_native.h

@@ -0,0 +1,25 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> math_native_layer_norm(const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> native_layer_norm_out_symint(const at::Tensor & input, c10::SymIntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> layer_norm_cpu(const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> layer_norm_cuda(const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> nested_layer_norm(const at::Tensor & input, at::IntArrayRef normalized_shape, const ::std::optional<at::Tensor> & weight, const ::std::optional<at::Tensor> & bias, double eps);
+} // namespace native
+} // namespace at