diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_cast_Char_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_cast_Char_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..7ecde69c246125ec003dbe6910f1815281289bab
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_cast_Char_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _cast_Char {
+  using schema = at::Tensor (const at::Tensor &, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_cast_Char")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_cast_Char(Tensor self, bool non_blocking=False) -> Tensor")
+  static at::Tensor call(const at::Tensor & self, bool non_blocking);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, bool non_blocking);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_conj_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_conj_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9d92c548e3cdef8ddb0db51b3875236070c9a3c3
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_conj_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 _conj(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr.h
new file mode 100644
index 0000000000000000000000000000000000000000..f9c6030b39a5b395b996b434e6d47729b99b35a8
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_convert_indices_from_coo_to_csr.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/_convert_indices_from_coo_to_csr_ops.h>
+
+namespace at {
+
+
+// aten::_convert_indices_from_coo_to_csr(Tensor self, int size, *, bool out_int32=False) -> Tensor
+inline at::Tensor _convert_indices_from_coo_to_csr(const at::Tensor & self, int64_t size, bool out_int32=false) {
+    return at::_ops::_convert_indices_from_coo_to_csr::call(self, size, out_int32);
+}
+
+// aten::_convert_indices_from_coo_to_csr.out(Tensor self, int size, *, bool out_int32=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _convert_indices_from_coo_to_csr_out(at::Tensor & out, const at::Tensor & self, int64_t size, bool out_int32=false) {
+    return at::_ops::_convert_indices_from_coo_to_csr_out::call(self, size, out_int32, out);
+}
+// aten::_convert_indices_from_coo_to_csr.out(Tensor self, int size, *, bool out_int32=False, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _convert_indices_from_coo_to_csr_outf(const at::Tensor & self, int64_t size, bool out_int32, at::Tensor & out) {
+    return at::_ops::_convert_indices_from_coo_to_csr_out::call(self, size, out_int32, out);
+}
+
+}
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_convolution_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_convolution_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..f712383b8d6fcd20e6e9ffb1276a098301e2247e
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_convolution_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 _convolution {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, bool, c10::SymIntArrayRef, c10::SymInt, bool, bool, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_convolution")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_convolution(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, SymInt[] output_padding, SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32) -> Tensor")
+  static at::Tensor call(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);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, 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);
+};
+
+struct TORCH_API _convolution_deprecated {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, bool, at::IntArrayRef, c10::SymInt, bool, bool, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_convolution")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "deprecated")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_convolution.deprecated(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, int[] output_padding, SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled) -> Tensor")
+  static at::Tensor call(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, at::IntArrayRef output_padding, c10::SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, 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, at::IntArrayRef output_padding, c10::SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled);
+};
+
+struct TORCH_API _convolution_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, c10::SymIntArrayRef, c10::SymIntArrayRef, c10::SymIntArrayRef, bool, c10::SymIntArrayRef, c10::SymInt, bool, bool, 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::_convolution")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_convolution.out(Tensor input, Tensor weight, Tensor? bias, SymInt[] stride, SymInt[] padding, SymInt[] dilation, bool transposed, SymInt[] output_padding, SymInt groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32, *, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(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);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, 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 at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_empty_per_channel_affine_quantized_cpu_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_empty_per_channel_affine_quantized_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e283b38f5680119946ad48eeac4374a335ec8cd9
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_empty_per_channel_affine_quantized_cpu_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor _empty_per_channel_affine_quantized(at::IntArrayRef size, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous);
+TORCH_API at::Tensor _empty_per_channel_affine_quantized(at::IntArrayRef size, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+TORCH_API at::Tensor _empty_per_channel_affine_quantized_symint(c10::SymIntArrayRef size, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, at::TensorOptions options={}, ::std::optional<at::MemoryFormat> memory_format=c10::MemoryFormat::Contiguous);
+TORCH_API at::Tensor _empty_per_channel_affine_quantized_symint(c10::SymIntArrayRef size, const at::Tensor & scales, const at::Tensor & zero_points, int64_t axis, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+
+} // namespace cpu
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_foreach_sub_compositeexplicitautograd_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_foreach_sub_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2b397b5b73b882ffccef6e1d196ecf92c3fc6dc7
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_foreach_sub_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,34 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::vector<at::Tensor> _foreach_sub(at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_sub_out(at::TensorList out, at::TensorList self, const at::Scalar & scalar);
+TORCH_API void _foreach_sub_outf(at::TensorList self, const at::Scalar & scalar, at::TensorList out);
+TORCH_API void _foreach_sub_(at::TensorList self, const at::Scalar & scalar);
+TORCH_API ::std::vector<at::Tensor> _foreach_sub(at::TensorList self, at::TensorList other, const at::Scalar & alpha=1);
+TORCH_API void _foreach_sub_out(at::TensorList out, at::TensorList self, at::TensorList other, const at::Scalar & alpha=1);
+TORCH_API void _foreach_sub_outf(at::TensorList self, at::TensorList other, const at::Scalar & alpha, at::TensorList out);
+TORCH_API void _foreach_sub_(at::TensorList self, at::TensorList other, const at::Scalar & alpha=1);
+TORCH_API ::std::vector<at::Tensor> _foreach_sub(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_sub_out(at::TensorList out, at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+TORCH_API void _foreach_sub_outf(at::TensorList self, at::ArrayRef<at::Scalar> scalars, at::TensorList out);
+TORCH_API void _foreach_sub_(at::TensorList self, at::ArrayRef<at::Scalar> scalars);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_foreach_zero_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_foreach_zero_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..54444a979d83e0c047e780cb71bae471fba2d1f3
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_foreach_zero_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::vector<at::Tensor> _foreach_zero(at::TensorList self);
+TORCH_API void _foreach_zero_out(at::TensorList self, at::TensorList out);
+TORCH_API void foreach_tensor_zero_slow_(at::TensorList self);
+TORCH_API void foreach_tensor_zero_cuda_(at::TensorList self);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_fused_dropout_cuda_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_fused_dropout_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2993f6a6283355a4dce8620963f08bcfd249093d
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_fused_dropout_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> _fused_dropout(const at::Tensor & self, double p, ::std::optional<at::Generator> generator=::std::nullopt);
+
+} // namespace cuda
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_indices_copy_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_indices_copy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..b7b1613909caadf6122d4fe91e91ccf2581eaf68
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_indices_copy_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _indices_copy_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor _indices_copy(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_mps_convolution_transpose_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_mps_convolution_transpose_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..3807235b771a17b6d8a5dd92fd703d1358ab18d2
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_mps_convolution_transpose_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 & _mps_convolution_transpose_out_symint(const at::Tensor & self, const at::Tensor & weight, c10::SymIntArrayRef padding, c10::SymIntArrayRef output_padding, c10::SymIntArrayRef stride, c10::SymIntArrayRef dilation, c10::SymInt groups, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nested_compute_contiguous_strides_offsets.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nested_compute_contiguous_strides_offsets.h
new file mode 100644
index 0000000000000000000000000000000000000000..742273f596dc71bcca82d3ebdf898f344b8aad44
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nested_compute_contiguous_strides_offsets.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_compute_contiguous_strides_offsets_ops.h>
+
+namespace at {
+
+
+// aten::_nested_compute_contiguous_strides_offsets(Tensor nested_size) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _nested_compute_contiguous_strides_offsets(const at::Tensor & nested_size) {
+    return at::_ops::_nested_compute_contiguous_strides_offsets::call(nested_size);
+}
+
+}
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nested_tensor_softmax_with_shape.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nested_tensor_softmax_with_shape.h
new file mode 100644
index 0000000000000000000000000000000000000000..916d2524729d825a61cb2c39ecb5708dd032ff88
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nested_tensor_softmax_with_shape.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_tensor_softmax_with_shape_ops.h>
+
+namespace at {
+
+
+// aten::_nested_tensor_softmax_with_shape(Tensor self, Tensor query) -> Tensor
+inline at::Tensor _nested_tensor_softmax_with_shape(const at::Tensor & self, const at::Tensor & query) {
+    return at::_ops::_nested_tensor_softmax_with_shape::call(self, query);
+}
+
+}
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nested_view_from_jagged_copy_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nested_view_from_jagged_copy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..095e55404da97b35908c3d286159b39e20e68e2a
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nested_view_from_jagged_copy_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & _nested_view_from_jagged_copy_out(const at::Tensor & self, const at::Tensor & offsets, const at::Tensor & dummy, const ::std::optional<at::Tensor> & lengths, int64_t ragged_idx, const ::std::optional<at::Tensor> & min_seqlen, const ::std::optional<at::Tensor> & max_seqlen, at::Tensor & out);
+TORCH_API at::Tensor _nested_view_from_jagged_copy(const at::Tensor & self, const at::Tensor & offsets, const at::Tensor & dummy, const ::std::optional<at::Tensor> & lengths={}, int64_t ragged_idx=1, const ::std::optional<at::Tensor> & min_seqlen={}, const ::std::optional<at::Tensor> & max_seqlen={});
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nnz_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nnz_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..87995163e077d65de575fb834f38c4768c186e70
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_nnz_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API int64_t _nnz_sparse(const at::Tensor & self);
+TORCH_API int64_t _nnz_sparse_csr(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_scaled_dot_product_flash_attention.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_scaled_dot_product_flash_attention.h
new file mode 100644
index 0000000000000000000000000000000000000000..13b81f4a7dfce0fea842640fb6693e63b414b8ed
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_scaled_dot_product_flash_attention.h
@@ -0,0 +1,30 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_scaled_dot_product_flash_attention_ops.h>
+
+namespace at {
+
+
+// aten::_scaled_dot_product_flash_attention(Tensor query, Tensor key, Tensor value, float dropout_p=0.0, bool is_causal=False, bool return_debug_mask=False, *, float? scale=None) -> (Tensor output, Tensor logsumexp, Tensor cum_seq_q, Tensor cum_seq_k, SymInt max_q, SymInt max_k, Tensor philox_seed, Tensor philox_offset, Tensor debug_attn_mask)
+inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt,at::Tensor,at::Tensor,at::Tensor> _scaled_dot_product_flash_attention(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, double dropout_p=0.0, bool is_causal=false, bool return_debug_mask=false, ::std::optional<double> scale=::std::nullopt) {
+    return at::_ops::_scaled_dot_product_flash_attention::call(query, key, value, dropout_p, is_causal, return_debug_mask, scale);
+}
+
+}
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_sparse_csc_tensor_unsafe_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_sparse_csc_tensor_unsafe_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..33faf0e49c86894695ef19513daf3006096424c4
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_sparse_csc_tensor_unsafe_ops.h
@@ -0,0 +1,28 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API _sparse_csc_tensor_unsafe {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, at::IntArrayRef, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_sparse_csc_tensor_unsafe")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_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")
+  static at::Tensor call(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);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & ccol_indices, const at::Tensor & row_indices, const at::Tensor & values, at::IntArrayRef size, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_sparse_semi_structured_addmm_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_sparse_semi_structured_addmm_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..17b0d175554d274b7faa2ceb0547ab525381b15e
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_sparse_semi_structured_addmm_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_addmm {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Tensor &, const at::Scalar &, const at::Scalar &, ::std::optional<at::ScalarType>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::_sparse_semi_structured_addmm")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "_sparse_semi_structured_addmm(Tensor input, Tensor mat1, Tensor mat1_meta, Tensor mat2, *, Scalar alpha=1, Scalar beta=1, ScalarType? out_dtype=None) -> Tensor")
+  static at::Tensor call(const at::Tensor & input, const at::Tensor & mat1, const at::Tensor & mat1_meta, const at::Tensor & mat2, const at::Scalar & alpha, const at::Scalar & beta, ::std::optional<at::ScalarType> out_dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & input, const at::Tensor & mat1, const at::Tensor & mat1_meta, const at::Tensor & mat2, const at::Scalar & alpha, const at::Scalar & beta, ::std::optional<at::ScalarType> out_dtype);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_test_warn_in_autograd.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_test_warn_in_autograd.h
new file mode 100644
index 0000000000000000000000000000000000000000..279e321f51cbd03212c8c67aff0e88a8da85ad2b
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_test_warn_in_autograd.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_warn_in_autograd_ops.h>
+
+namespace at {
+
+
+// aten::_test_warn_in_autograd(Tensor self) -> Tensor
+inline at::Tensor _test_warn_in_autograd(const at::Tensor & self) {
+    return at::_ops::_test_warn_in_autograd::call(self);
+}
+
+// aten::_test_warn_in_autograd.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _test_warn_in_autograd_out(at::Tensor & out, const at::Tensor & self) {
+    return at::_ops::_test_warn_in_autograd_out::call(self, out);
+}
+// aten::_test_warn_in_autograd.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & _test_warn_in_autograd_outf(const at::Tensor & self, at::Tensor & out) {
+    return at::_ops::_test_warn_in_autograd_out::call(self, out);
+}
+
+}
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_impl_cuda_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_impl_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..acf1f4f4a85f3d2fe0743407032698649234cfb1
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_thnn_fused_lstm_cell_backward_impl_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _thnn_fused_lstm_cell_backward_impl(const ::std::optional<at::Tensor> & grad_hy, const ::std::optional<at::Tensor> & grad_cy, const at::Tensor & cx, const at::Tensor & cy, const at::Tensor & workspace, bool has_bias);
+
+} // namespace cuda
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_unsafe_view_compositeexplicitautograd_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_unsafe_view_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..532036cb8e7370371ba20b2b5bbf9997b5c926e8
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_unsafe_view_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 _unsafe_view(const at::Tensor & self, at::IntArrayRef size);
+TORCH_API at::Tensor _unsafe_view_symint(const at::Tensor & self, c10::SymIntArrayRef size);
+TORCH_API at::Tensor & _unsafe_view_out(at::Tensor & out, const at::Tensor & self, at::IntArrayRef size);
+TORCH_API at::Tensor & _unsafe_view_outf(const at::Tensor & self, at::IntArrayRef size, at::Tensor & out);
+TORCH_API at::Tensor & _unsafe_view_symint_out(at::Tensor & out, const at::Tensor & self, c10::SymIntArrayRef size);
+TORCH_API at::Tensor & _unsafe_view_symint_outf(const at::Tensor & self, c10::SymIntArrayRef size, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_weight_norm_interface.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_weight_norm_interface.h
new file mode 100644
index 0000000000000000000000000000000000000000..eb781cb320930edea247f53973fb4d2b96357e43
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_weight_norm_interface.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/_weight_norm_interface_ops.h>
+
+namespace at {
+
+
+// aten::_weight_norm_interface(Tensor v, Tensor g, int dim=0) -> (Tensor, Tensor)
+inline ::std::tuple<at::Tensor,at::Tensor> _weight_norm_interface(const at::Tensor & v, const at::Tensor & g, int64_t dim=0) {
+    return at::_ops::_weight_norm_interface::call(v, g, dim);
+}
+
+// aten::_weight_norm_interface.out(Tensor v, Tensor g, int dim=0, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _weight_norm_interface_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & v, const at::Tensor & g, int64_t dim=0) {
+    return at::_ops::_weight_norm_interface_out::call(v, g, dim, out0, out1);
+}
+// aten::_weight_norm_interface.out(Tensor v, Tensor g, int dim=0, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))
+inline ::std::tuple<at::Tensor &,at::Tensor &> _weight_norm_interface_outf(const at::Tensor & v, const at::Tensor & g, int64_t dim, at::Tensor & out0, at::Tensor & out1) {
+    return at::_ops::_weight_norm_interface_out::call(v, g, dim, out0, out1);
+}
+
+}
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_weight_norm_interface_backward_cuda_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_weight_norm_interface_backward_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4d0f12a3d040af148407c30630901ca4f5742ad4
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/_weight_norm_interface_backward_cuda_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cuda {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor> _weight_norm_interface_backward(const at::Tensor & grad_w, const at::Tensor & saved_v, const at::Tensor & saved_g, const at::Tensor & saved_norms, int64_t dim);
+
+} // namespace cuda
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward_compositeexplicitautogradnonfunctional_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..8bc988555d4df387f3c54e96f577e88c87f38ac0
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/adaptive_max_pool2d_backward_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor adaptive_max_pool2d_backward(const at::Tensor & grad_output, const at::Tensor & self, const at::Tensor & indices);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/alias_compositeexplicitautograd_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/alias_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..4ad23ce0577d552682209502825545fe9cfd7ee2
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/alias_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 alias(const at::Tensor & self);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/binary_cross_entropy_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/binary_cross_entropy_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..541c3aefcaa4b33617e419e5475358885411993b
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/binary_cross_entropy_native.h
@@ -0,0 +1,24 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor binary_cross_entropy_cpu(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean);
+TORCH_API at::Tensor & binary_cross_entropy_out_cpu(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, at::Tensor & out);
+TORCH_API at::Tensor binary_cross_entropy_cuda(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight={}, int64_t reduction=at::Reduction::Mean);
+TORCH_API at::Tensor & binary_cross_entropy_out_cuda(const at::Tensor & self, const at::Tensor & target, const ::std::optional<at::Tensor> & weight, int64_t reduction, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/cat_meta_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/cat_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..7c094d7f83bd88fb770e504923190a5639687223
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/cat_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor cat(const at::ITensorListRef & tensors, int64_t dim=0);
+TORCH_API at::Tensor & cat_out(at::Tensor & out, const at::ITensorListRef & tensors, int64_t dim=0);
+TORCH_API at::Tensor & cat_outf(const at::ITensorListRef & tensors, int64_t dim, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/cudnn_convolution_relu_compositeexplicitautograd_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/cudnn_convolution_relu_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..70896ded145e186a83930f630976c2ccd59c1b7d
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/cudnn_convolution_relu_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 & cudnn_convolution_relu_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, int64_t groups);
+TORCH_API at::Tensor & cudnn_convolution_relu_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, int64_t groups, at::Tensor & out);
+TORCH_API at::Tensor & cudnn_convolution_relu_symint_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, c10::SymInt groups);
+TORCH_API at::Tensor & cudnn_convolution_relu_symint_outf(const at::Tensor & self, const at::Tensor & weight, const ::std::optional<at::Tensor> & bias, c10::SymIntArrayRef stride, c10::SymIntArrayRef padding, c10::SymIntArrayRef dilation, c10::SymInt groups, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/expm1_meta.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/expm1_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..7c2e4a836efa76a8151261e1a0c592e3ab6fb809
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/expm1_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_expm1 : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_cachemask_compositeexplicitautograd_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_cachemask_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..3758201c755cdaf3e0d04e0eacb533969e1e4392
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_cachemask_compositeexplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautograd {
+
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> fake_quantize_per_channel_affine_cachemask_out(at::Tensor & out0, at::Tensor & out1, const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &> fake_quantize_per_channel_affine_cachemask_outf(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max, at::Tensor & out0, at::Tensor & out1);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_cachemask_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_cachemask_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..5de31e13b13a9de84316a42b3d0533246cecd019
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fake_quantize_per_channel_affine_cachemask_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API fake_quantize_per_channel_affine_cachemask {
+  using schema = ::std::tuple<at::Tensor,at::Tensor> (const at::Tensor &, const at::Tensor &, const at::Tensor &, int64_t, int64_t, int64_t);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::fake_quantize_per_channel_affine_cachemask")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "fake_quantize_per_channel_affine_cachemask(Tensor self, Tensor scale, Tensor zero_point, int axis, int quant_min, int quant_max) -> (Tensor output, Tensor mask)")
+  static ::std::tuple<at::Tensor,at::Tensor> call(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max);
+  static ::std::tuple<at::Tensor,at::Tensor> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max);
+};
+
+struct TORCH_API fake_quantize_per_channel_affine_cachemask_out {
+  using schema = ::std::tuple<at::Tensor &,at::Tensor &> (const at::Tensor &, const at::Tensor &, const at::Tensor &, int64_t, int64_t, int64_t, at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::fake_quantize_per_channel_affine_cachemask")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "fake_quantize_per_channel_affine_cachemask.out(Tensor self, Tensor scale, Tensor zero_point, int axis, int quant_min, int quant_max, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))")
+  static ::std::tuple<at::Tensor &,at::Tensor &> call(const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max, at::Tensor & out0, at::Tensor & out1);
+  static ::std::tuple<at::Tensor &,at::Tensor &> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & scale, const at::Tensor & zero_point, int64_t axis, int64_t quant_min, int64_t quant_max, at::Tensor & out0, at::Tensor & out1);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fake_quantize_per_tensor_affine_cachemask_backward_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fake_quantize_per_tensor_affine_cachemask_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..488472a741ee6059b47f6e092c27d01135e85322
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fake_quantize_per_tensor_affine_cachemask_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 fake_quantize_per_tensor_affine_cachemask_backward(const at::Tensor & grad, const at::Tensor & mask);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fft_fft_compositeimplicitautograd_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fft_fft_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..845e5ad621989dbf896ce048631b0089d91ab3ee
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fft_fft_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,28 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor fft_fft(const at::Tensor & self, ::std::optional<int64_t> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor fft_fft_symint(const at::Tensor & self, ::std::optional<c10::SymInt> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_fft_out(at::Tensor & out, const at::Tensor & self, ::std::optional<int64_t> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_fft_outf(const at::Tensor & self, ::std::optional<int64_t> n, int64_t dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+TORCH_API at::Tensor & fft_fft_symint_out(at::Tensor & out, const at::Tensor & self, ::std::optional<c10::SymInt> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_fft_symint_outf(const at::Tensor & self, ::std::optional<c10::SymInt> n, int64_t dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fft_irfft2.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fft_irfft2.h
new file mode 100644
index 0000000000000000000000000000000000000000..67f1dceea37136983cf3b6fecc3a88498ea59c10
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fft_irfft2.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/fft_irfft2_ops.h>
+
+namespace at {
+
+
+// aten::fft_irfft2(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None) -> Tensor
+inline at::Tensor fft_irfft2(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_irfft2::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, int64_t>::value>>
+  at::Tensor fft_irfft2(const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_irfft2::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm);
+  }
+}
+
+// aten::fft_irfft2(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None) -> Tensor
+inline at::Tensor fft_irfft2_symint(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_irfft2::call(self, s, dim, norm);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, c10::SymInt>::value>>
+  at::Tensor fft_irfft2(const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_irfft2::call(self, s, dim, norm);
+  }
+}
+
+// aten::fft_irfft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_irfft2_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_irfft2_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, int64_t>::value>>
+  at::Tensor & fft_irfft2_out(at::Tensor & out, const at::Tensor & self, at::OptionalIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_irfft2_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_irfft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_irfft2_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_irfft2_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, int64_t>::value>>
+  at::Tensor & fft_irfft2_outf(const at::Tensor & self, at::OptionalIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_irfft2_out::call(self, s.has_value() ? ::std::make_optional(c10::fromIntArrayRefSlow(*s)) : ::std::nullopt, dim, norm, out);
+  }
+}
+
+// aten::fft_irfft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_irfft2_symint_out(at::Tensor & out, const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_irfft2_out::call(self, s, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, c10::SymInt>::value>>
+  at::Tensor & fft_irfft2_out(at::Tensor & out, const at::Tensor & self, at::OptionalSymIntArrayRef s=::std::nullopt, at::IntArrayRef dim={-2,-1}, ::std::optional<c10::string_view> norm=::std::nullopt) {
+    return at::_ops::fft_irfft2_out::call(self, s, dim, norm, out);
+  }
+}
+
+// aten::fft_irfft2.out(Tensor self, SymInt[1]? s=None, int[1] dim=[-2,-1], str? norm=None, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & fft_irfft2_symint_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_irfft2_out::call(self, s, dim, norm, out);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, c10::SymInt>::value>>
+  at::Tensor & fft_irfft2_outf(const at::Tensor & self, at::OptionalSymIntArrayRef s, at::IntArrayRef dim, ::std::optional<c10::string_view> norm, at::Tensor & out) {
+    return at::_ops::fft_irfft2_out::call(self, s, dim, norm, out);
+  }
+}
+
+}
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fft_rfft_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fft_rfft_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..a9b0c196f0b675f67ad0a7adc9b8aa90aecd601d
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fft_rfft_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor fft_rfft_symint(const at::Tensor & self, ::std::optional<c10::SymInt> n=::std::nullopt, int64_t dim=-1, ::std::optional<c10::string_view> norm=::std::nullopt);
+TORCH_API at::Tensor & fft_rfft_symint_out(const at::Tensor & self, ::std::optional<c10::SymInt> n, int64_t dim, ::std::optional<c10::string_view> norm, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fix_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fix_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..2cab32804671df5c0aff98604b83b6a5d7e7114c
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/fix_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 fix {
+  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::fix")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "fix(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 fix_ {
+  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::fix_")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "fix_(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 fix_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::fix")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "fix.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/grid_sampler_2d_compositeexplicitautograd_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/grid_sampler_2d_compositeexplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..1f87bc0f0a5b07c21e4a312fc00f1880f6b344e6
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/grid_sampler_2d_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 & grid_sampler_2d_out(at::Tensor & out, const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners);
+TORCH_API at::Tensor & grid_sampler_2d_outf(const at::Tensor & input, const at::Tensor & grid, int64_t interpolation_mode, int64_t padding_mode, bool align_corners, at::Tensor & out);
+
+} // namespace compositeexplicitautograd
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/hypot.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/hypot.h
new file mode 100644
index 0000000000000000000000000000000000000000..9728e49095ea04e000e9ea8f125faf5208de1e55
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/hypot.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/hypot_ops.h>
+
+namespace at {
+
+
+// aten::hypot.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hypot_out(at::Tensor & out, const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::hypot_out::call(self, other, out);
+}
+// aten::hypot.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & hypot_outf(const at::Tensor & self, const at::Tensor & other, at::Tensor & out) {
+    return at::_ops::hypot_out::call(self, other, out);
+}
+
+// aten::hypot(Tensor self, Tensor other) -> Tensor
+inline at::Tensor hypot(const at::Tensor & self, const at::Tensor & other) {
+    return at::_ops::hypot::call(self, other);
+}
+
+}
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/hypot_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/hypot_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..8c26720867c36b21bdca8b0012c2e5f8ec8c5f07
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/hypot_ops.h
@@ -0,0 +1,50 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API hypot_out {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::hypot")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "hypot.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other, at::Tensor & out);
+};
+
+struct TORCH_API hypot {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::hypot")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "hypot(Tensor self, Tensor other) -> Tensor")
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & other);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & other);
+};
+
+struct TORCH_API hypot_ {
+  using schema = at::Tensor & (at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::hypot_")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "hypot_(Tensor(a!) self, Tensor other) -> Tensor(a!)")
+  static at::Tensor & call(at::Tensor & self, const at::Tensor & other);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, at::Tensor & self, const at::Tensor & other);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/int_repr_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/int_repr_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4f7b48a27c104aa4e6b279edf222fbb0fb54ed2d
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/int_repr_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 & int_repr_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor int_repr_quantized_cpu(const at::Tensor & self);
+TORCH_API at::Tensor int_repr_quantized_cuda(const at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/int_repr_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/int_repr_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..ceded25cb0c0b7cf304d849660100691dd74d3ed
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/int_repr_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 int_repr {
+  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::int_repr")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "int_repr(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 int_repr_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::int_repr")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "int_repr.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/linalg_cross_meta.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/linalg_cross_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..c4c25818425c2ba8d331e079b4d70bf625dc03c5
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/linalg_cross_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_linalg_cross : public at::impl::MetaBase {
+    
+    
+    void meta(const at::Tensor & self, const at::Tensor & other, int64_t dim);
+};
+
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/log2_cpu_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/log2_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f885a233fbed2c46b8d73c615b32dc327be5cae5
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/log2_cpu_dispatch.h
@@ -0,0 +1,26 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API at::Tensor log2(const at::Tensor & self);
+TORCH_API at::Tensor & log2_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & log2_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & log2_(at::Tensor & self);
+
+} // namespace cpu
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/lshift_meta_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/lshift_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..e75d05f6734819a6c04965dabd8bcf4b3b737568
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/lshift_meta_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor & __ilshift__(at::Tensor & self, const at::Scalar & other);
+TORCH_API at::Tensor & __ilshift__(at::Tensor & self, const at::Tensor & other);
+
+} // namespace meta
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/lu_unpack_cpu_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/lu_unpack_cpu_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..33a48f21e220762c6041677d52a654dffef51ad9
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/lu_unpack_cpu_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace cpu {
+
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> lu_unpack(const at::Tensor & LU_data, const at::Tensor & LU_pivots, bool unpack_data=true, bool unpack_pivots=true);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> lu_unpack_out(at::Tensor & P, at::Tensor & L, at::Tensor & U, const at::Tensor & LU_data, const at::Tensor & LU_pivots, bool unpack_data=true, bool unpack_pivots=true);
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> lu_unpack_outf(const at::Tensor & LU_data, const at::Tensor & LU_pivots, bool unpack_data, bool unpack_pivots, at::Tensor & P, at::Tensor & L, at::Tensor & U);
+
+} // namespace cpu
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/max_pool1d_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/max_pool1d_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..64f82cf6d56d603a4746618010c692f20f55c9f5
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/max_pool1d_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 max_pool1d {
+  using schema = at::Tensor (const at::Tensor &, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, at::IntArrayRef, bool);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::max_pool1d")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "max_pool1d(Tensor self, int[1] kernel_size, int[1] stride=[], int[1] padding=0, int[1] dilation=1, bool ceil_mode=False) -> Tensor")
+  static at::Tensor call(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/max_pool2d_compositeimplicitautograd_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/max_pool2d_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0da0108e932711216fc2381fb6433058ba00dc37
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/max_pool2d_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor max_pool2d(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, bool ceil_mode=false);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/mean_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/mean_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..45cb16a4c7a924a1acc1a7dadf541d6da6045326
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/mean_ops.h
@@ -0,0 +1,83 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API mean {
+  using schema = at::Tensor (const at::Tensor &, ::std::optional<at::ScalarType>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::mean")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "mean(Tensor self, *, ScalarType? dtype=None) -> Tensor")
+  static at::Tensor call(const at::Tensor & self, ::std::optional<at::ScalarType> dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, ::std::optional<at::ScalarType> dtype);
+};
+
+struct TORCH_API mean_dtype_out {
+  using schema = at::Tensor & (const at::Tensor &, ::std::optional<at::ScalarType>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::mean")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "dtype_out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "mean.dtype_out(Tensor self, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+};
+
+struct TORCH_API mean_dim {
+  using schema = at::Tensor (const at::Tensor &, at::OptionalIntArrayRef, bool, ::std::optional<at::ScalarType>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::mean")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "dim")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "mean.dim(Tensor self, int[1]? dim, bool keepdim=False, *, ScalarType? dtype=None) -> Tensor")
+  static at::Tensor call(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype);
+};
+
+struct TORCH_API mean_out {
+  using schema = at::Tensor & (const at::Tensor &, at::OptionalIntArrayRef, bool, ::std::optional<at::ScalarType>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::mean")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "mean.out(Tensor self, int[1]? dim, bool keepdim=False, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+};
+
+struct TORCH_API mean_names_dim {
+  using schema = at::Tensor (const at::Tensor &, at::DimnameList, bool, ::std::optional<at::ScalarType>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::mean")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "names_dim")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "mean.names_dim(Tensor self, Dimname[1] dim, bool keepdim=False, *, ScalarType? dtype=None) -> Tensor")
+  static at::Tensor call(const at::Tensor & self, at::DimnameList dim, bool keepdim, ::std::optional<at::ScalarType> dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::DimnameList dim, bool keepdim, ::std::optional<at::ScalarType> dtype);
+};
+
+struct TORCH_API mean_names_out {
+  using schema = at::Tensor & (const at::Tensor &, at::DimnameList, bool, ::std::optional<at::ScalarType>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::mean")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "names_out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "mean.names_out(Tensor self, Dimname[1] dim, bool keepdim=False, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, at::DimnameList dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::DimnameList dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/mish_backward_compositeimplicitautograd_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/mish_backward_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..a150e80649b0b6d42d767d9210c770e98d21aaf7
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/mish_backward_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor mish_backward(const at::Tensor & grad_output, const at::Tensor & self);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..e5f4ab54a5c50477cabbfcb7c1a31ff048306d53
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/mkldnn_max_pool2d_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor & mkldnn_max_pool2d_out(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation, bool ceil_mode, at::Tensor & out);
+TORCH_API at::Tensor mkldnn_max_pool2d(const at::Tensor & self, at::IntArrayRef kernel_size, at::IntArrayRef stride={}, at::IntArrayRef padding=0, at::IntArrayRef dilation=1, bool ceil_mode=false);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/nanmean_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/nanmean_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..a8b285ddb53fa6fed13dea2ec8e89726d45d7536
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/nanmean_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor nanmean(const at::Tensor & self, at::OptionalIntArrayRef dim=::std::nullopt, bool keepdim=false, ::std::optional<at::ScalarType> dtype=::std::nullopt);
+TORCH_API at::Tensor & nanmean_out(const at::Tensor & self, at::OptionalIntArrayRef dim, bool keepdim, ::std::optional<at::ScalarType> dtype, at::Tensor & out);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/prelu_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/prelu_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..4987293fa35db02fb7aa8a969204965942551de8
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/prelu_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 prelu {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::prelu")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "prelu(Tensor self, Tensor weight) -> Tensor")
+  static at::Tensor call(const at::Tensor & self, const at::Tensor & weight);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, const at::Tensor & weight);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/randint_like_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/randint_like_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..cb8151061a2f7e74e582aa16e957e032c22d4f8e
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/randint_like_ops.h
@@ -0,0 +1,61 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API randint_like {
+  using schema = at::Tensor (const at::Tensor &, c10::SymInt, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>, ::std::optional<at::MemoryFormat>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::randint_like")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "randint_like(Tensor self, SymInt high, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor")
+  static at::Tensor call(const at::Tensor & self, c10::SymInt high, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymInt high, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+};
+
+struct TORCH_API randint_like_low_dtype {
+  using schema = at::Tensor (const at::Tensor &, c10::SymInt, c10::SymInt, ::std::optional<at::ScalarType>, ::std::optional<at::Layout>, ::std::optional<at::Device>, ::std::optional<bool>, ::std::optional<at::MemoryFormat>);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::randint_like")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "low_dtype")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "randint_like.low_dtype(Tensor self, SymInt low, SymInt high, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor")
+  static at::Tensor call(const at::Tensor & self, c10::SymInt low, c10::SymInt high, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymInt low, c10::SymInt high, ::std::optional<at::ScalarType> dtype, ::std::optional<at::Layout> layout, ::std::optional<at::Device> device, ::std::optional<bool> pin_memory, ::std::optional<at::MemoryFormat> memory_format);
+};
+
+struct TORCH_API randint_like_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymInt, ::std::optional<at::MemoryFormat>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::randint_like")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "randint_like.out(Tensor self, SymInt high, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, c10::SymInt high, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymInt high, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+};
+
+struct TORCH_API randint_like_low_dtype_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymInt, c10::SymInt, ::std::optional<at::MemoryFormat>, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::randint_like")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "low_dtype_out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "randint_like.low_dtype_out(Tensor self, SymInt low, SymInt high, *, MemoryFormat? memory_format=None, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, c10::SymInt low, c10::SymInt high, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymInt low, c10::SymInt high, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/reciprocal_meta_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/reciprocal_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..35b127f2ec222def7a12a20b7beadd6f8316e96b
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/reciprocal_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 reciprocal(const at::Tensor & self);
+TORCH_API at::Tensor & reciprocal_out(at::Tensor & out, const at::Tensor & self);
+TORCH_API at::Tensor & reciprocal_outf(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & reciprocal_(at::Tensor & self);
+
+} // namespace meta
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/replication_pad2d_backward.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/replication_pad2d_backward.h
new file mode 100644
index 0000000000000000000000000000000000000000..e58e7de257c34e85d45f636c8b8670729ae98ffd
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/replication_pad2d_backward.h
@@ -0,0 +1,91 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/replication_pad2d_backward_ops.h>
+
+namespace at {
+
+
+// aten::replication_pad2d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & replication_pad2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, c10::fromIntArrayRefSlow(padding), grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, int64_t>::value>>
+  at::Tensor & replication_pad2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, c10::fromIntArrayRefSlow(padding), grad_input);
+  }
+}
+
+// aten::replication_pad2d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & replication_pad2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding, at::Tensor & grad_input) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, c10::fromIntArrayRefSlow(padding), grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, int64_t>::value>>
+  at::Tensor & replication_pad2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding, at::Tensor & grad_input) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, c10::fromIntArrayRefSlow(padding), grad_input);
+  }
+}
+
+// aten::replication_pad2d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & replication_pad2d_backward_symint_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, padding, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, c10::SymInt>::value>>
+  at::Tensor & replication_pad2d_backward_out(at::Tensor & grad_input, const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, padding, grad_input);
+  }
+}
+
+// aten::replication_pad2d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!)
+inline at::Tensor & replication_pad2d_backward_symint_outf(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & grad_input) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, padding, grad_input);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, c10::SymInt>::value>>
+  at::Tensor & replication_pad2d_backward_outf(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & grad_input) {
+    return at::_ops::replication_pad2d_backward_grad_input::call(grad_output, self, padding, grad_input);
+  }
+}
+
+// aten::replication_pad2d_backward(Tensor grad_output, Tensor self, SymInt[4] padding) -> Tensor
+inline at::Tensor replication_pad2d_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward::call(grad_output, self, c10::fromIntArrayRefSlow(padding));
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, int64_t>::value>>
+  at::Tensor replication_pad2d_backward(const at::Tensor & grad_output, const at::Tensor & self, at::IntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward::call(grad_output, self, c10::fromIntArrayRefSlow(padding));
+  }
+}
+
+// aten::replication_pad2d_backward(Tensor grad_output, Tensor self, SymInt[4] padding) -> Tensor
+inline at::Tensor replication_pad2d_backward_symint(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward::call(grad_output, self, padding);
+}
+namespace symint {
+  template <typename T, typename = std::enable_if_t<std::is_same<T, c10::SymInt>::value>>
+  at::Tensor replication_pad2d_backward(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding) {
+    return at::_ops::replication_pad2d_backward::call(grad_output, self, padding);
+  }
+}
+
+}
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/replication_pad3d_backward_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/replication_pad3d_backward_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..8278ea07582228854b2fe728a27877fe8c081db8
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/replication_pad3d_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 replication_pad3d_backward_grad_input {
+  using schema = at::Tensor & (const at::Tensor &, const at::Tensor &, c10::SymIntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::replication_pad3d_backward")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "grad_input")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "replication_pad3d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[6] padding, *, Tensor(a!) grad_input) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & grad_input);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding, at::Tensor & grad_input);
+};
+
+struct TORCH_API replication_pad3d_backward {
+  using schema = at::Tensor (const at::Tensor &, const at::Tensor &, c10::SymIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::replication_pad3d_backward")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "replication_pad3d_backward(Tensor grad_output, Tensor self, SymInt[6] padding) -> Tensor")
+  static at::Tensor call(const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & grad_output, const at::Tensor & self, c10::SymIntArrayRef padding);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/requires_grad_compositeimplicitautograd_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/requires_grad_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..50a6aafd48f3e2814b60b16ebecab6a273772b3d
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/requires_grad_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,23 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor & requires_grad_(at::Tensor & self, bool requires_grad=true);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/roll_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/roll_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..58644c7c76734c072edbea9732faa555bf4bc665
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/roll_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 roll {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef, at::IntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::roll")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "roll(Tensor self, SymInt[1] shifts, int[1] dims=[]) -> Tensor")
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef shifts, at::IntArrayRef dims);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef shifts, at::IntArrayRef dims);
+};
+
+struct TORCH_API roll_out {
+  using schema = at::Tensor & (const at::Tensor &, c10::SymIntArrayRef, at::IntArrayRef, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::roll")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "roll.out(Tensor self, SymInt[1] shifts, int[1] dims=[], *, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, c10::SymIntArrayRef shifts, at::IntArrayRef dims, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef shifts, at::IntArrayRef dims, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/sigmoid_compositeexplicitautogradnonfunctional_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/sigmoid_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..2e614a6eb1dde781783a148f64d11fb159543b71
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/sigmoid_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 sigmoid(const at::Tensor & self);
+TORCH_API at::Tensor & sigmoid_(at::Tensor & self);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/sin_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/sin_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..519ccf00bd1e986b9476b85bb2f26bc16cc9a6bd
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/sin_native.h
@@ -0,0 +1,30 @@
+#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/sin_meta.h>
+
+namespace at {
+namespace native {
+struct TORCH_API structured_sin_out : public at::meta::structured_sin {
+void impl(const at::Tensor & self, const at::Tensor & out);
+};
+TORCH_API at::Tensor sin_nested(const at::Tensor & self);
+TORCH_API at::Tensor sin_sparse(const at::Tensor & self);
+TORCH_API at::Tensor & sin_sparse_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & sin_sparse_(at::Tensor & self);
+TORCH_API at::Tensor sin_sparse_csr(const at::Tensor & self);
+TORCH_API at::Tensor & sin_sparse_csr_out(const at::Tensor & self, at::Tensor & out);
+TORCH_API at::Tensor & sin_sparse_csr_(at::Tensor & self);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/size.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/size.h
new file mode 100644
index 0000000000000000000000000000000000000000..6916fa2d8c4ad6a4ef5dcd0c8c10142ddf807b4f
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/size.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/size_ops.h>
+
+namespace at {
+
+
+// aten::size.int(Tensor self, int dim) -> int
+inline int64_t __dispatch_size(const at::Tensor & self, int64_t dim) {
+    return at::_ops::size_int::call(self, dim);
+}
+
+// aten::size.Dimname(Tensor self, Dimname dim) -> int
+inline int64_t size(const at::Tensor & self, at::Dimname dim) {
+    return at::_ops::size_Dimname::call(self, dim);
+}
+
+}
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/slice_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/slice_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..bb4dfe862dd02ca55cf69c204b814a2fb6658bc2
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/slice_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 slice_Tensor {
+  using schema = at::Tensor (const at::Tensor &, int64_t, ::std::optional<c10::SymInt>, ::std::optional<c10::SymInt>, c10::SymInt);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::slice")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "Tensor")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "slice.Tensor(Tensor(a) self, int dim=0, SymInt? start=None, SymInt? end=None, SymInt step=1) -> Tensor(a)")
+  static at::Tensor call(const at::Tensor & self, int64_t dim, ::std::optional<c10::SymInt> start, ::std::optional<c10::SymInt> end, c10::SymInt step);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, int64_t dim, ::std::optional<c10::SymInt> start, ::std::optional<c10::SymInt> end, c10::SymInt step);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_w_meta_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_w_meta_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..24920a15075e9c124c872fba20e37722c584519d
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_chebyshev_polynomial_w_meta_dispatch.h
@@ -0,0 +1,25 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace meta {
+
+TORCH_API at::Tensor special_chebyshev_polynomial_w(const at::Tensor & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_chebyshev_polynomial_w_out(at::Tensor & out, const at::Tensor & x, const at::Tensor & n);
+TORCH_API at::Tensor & special_chebyshev_polynomial_w_outf(const at::Tensor & x, const at::Tensor & n, at::Tensor & out);
+
+} // namespace meta
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_i1_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_i1_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..669760680ecd229e35860a74c86cb8f5b992e0c0
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_i1_ops.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Operator.h
+
+#include <tuple>
+#include <vector>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+namespace _ops {
+
+
+struct TORCH_API special_i1 {
+  using schema = at::Tensor (const at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::special_i1")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "special_i1(Tensor self) -> Tensor")
+  static at::Tensor call(const at::Tensor & self);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self);
+};
+
+struct TORCH_API special_i1_out {
+  using schema = at::Tensor & (const at::Tensor &, at::Tensor &);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::special_i1")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "special_i1.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_i1e_meta.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_i1e_meta.h
new file mode 100644
index 0000000000000000000000000000000000000000..045b0a2377bc492ba1f95602afae88b003ad7aba
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_i1e_meta.h
@@ -0,0 +1,27 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeMetaFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/TensorIterator.h>
+#include <ATen/TensorMeta.h>
+#include <tuple>
+#include <vector>
+
+namespace at {
+namespace meta {
+
+struct TORCH_API structured_special_i1e : public TensorIteratorBase {
+    
+    
+    void meta(const at::Tensor & self);
+};
+
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_round.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_round.h
new file mode 100644
index 0000000000000000000000000000000000000000..57664be671696c68e5a5b936cbf3f9c0d42df089
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/special_round.h
@@ -0,0 +1,39 @@
+#pragma once
+
+// @generated by torchgen/gen.py from Function.h
+
+#include <ATen/Context.h>
+#include <ATen/DeviceGuard.h>
+#include <ATen/TensorUtils.h>
+#include <ATen/TracerMode.h>
+#include <ATen/core/Generator.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+
+
+
+#include <ATen/ops/special_round_ops.h>
+
+namespace at {
+
+
+// aten::special_round(Tensor self, *, int decimals=0) -> Tensor
+inline at::Tensor special_round(const at::Tensor & self, int64_t decimals=0) {
+    return at::_ops::special_round::call(self, decimals);
+}
+
+// aten::special_round.out(Tensor self, *, int decimals=0, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_round_out(at::Tensor & out, const at::Tensor & self, int64_t decimals=0) {
+    return at::_ops::special_round_out::call(self, decimals, out);
+}
+// aten::special_round.out(Tensor self, *, int decimals=0, Tensor(a!) out) -> Tensor(a!)
+inline at::Tensor & special_round_outf(const at::Tensor & self, int64_t decimals, at::Tensor & out) {
+    return at::_ops::special_round_out::call(self, decimals, out);
+}
+
+}
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/tan_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/tan_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..0c56be6077a1ad4a15fff82b5404957c8ad515aa
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/tan_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 tan {
+  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::tan")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "tan(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 tan_ {
+  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::tan_")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "tan_(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 tan_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::tan")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "out")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "tan.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!)")
+  static at::Tensor & call(const at::Tensor & self, at::Tensor & out);
+  static at::Tensor & redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::Tensor & out);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/to_mkldnn_backward_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/to_mkldnn_backward_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..4e7d221203e58040da48e652ed4a599b47abe912
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/to_mkldnn_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 to_mkldnn_backward(const at::Tensor & grad, const at::Tensor & input);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/trapz_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/trapz_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..9dc40b45cecc91a6bef3bc064a6c6ed7f482829d
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/trapz_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor trapz(const at::Tensor & y, const at::Tensor & x, int64_t dim=-1);
+TORCH_API at::Tensor trapz(const at::Tensor & y, double dx=1, int64_t dim=-1);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/unflatten_dense_tensors_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/unflatten_dense_tensors_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..286c395c698d4d559c6316eb2fef957093304e7d
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/unflatten_dense_tensors_native.h
@@ -0,0 +1,21 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::vector<at::Tensor> unflatten_dense_tensors(const at::Tensor & flat, at::TensorList tensors);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/unique_dim_consecutive_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/unique_dim_consecutive_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..499281b18be8a63da9d64d23d4bd8d2fc97c59ff
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/unique_dim_consecutive_native.h
@@ -0,0 +1,23 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API ::std::tuple<at::Tensor &,at::Tensor &,at::Tensor &> unique_dim_consecutive_out(const at::Tensor & self, int64_t dim, bool return_inverse, bool return_counts, at::Tensor & out0, at::Tensor & out1, at::Tensor & out2);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> unique_dim_consecutive_cpu(const at::Tensor & self, int64_t dim, bool return_inverse=false, bool return_counts=false);
+TORCH_API ::std::tuple<at::Tensor,at::Tensor,at::Tensor> unique_dim_consecutive_cuda(const at::Tensor & self, int64_t dim, bool return_inverse=false, bool return_counts=false);
+} // namespace native
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/upsample_bilinear2d_compositeexplicitautogradnonfunctional_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/upsample_bilinear2d_compositeexplicitautogradnonfunctional_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..0799f171959c6300582acbf1e1574b793eaaf34f
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/upsample_bilinear2d_compositeexplicitautogradnonfunctional_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeexplicitautogradnonfunctional {
+
+TORCH_API at::Tensor upsample_bilinear2d(const at::Tensor & self, at::IntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+TORCH_API at::Tensor upsample_bilinear2d_symint(const at::Tensor & self, c10::SymIntArrayRef output_size, bool align_corners, ::std::optional<double> scales_h=::std::nullopt, ::std::optional<double> scales_w=::std::nullopt);
+
+} // namespace compositeexplicitautogradnonfunctional
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/upsample_bilinear2d_compositeimplicitautograd_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/upsample_bilinear2d_compositeimplicitautograd_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..c554723846bfe5e9321d68bc3e16297445a291f8
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/upsample_bilinear2d_compositeimplicitautograd_dispatch.h
@@ -0,0 +1,24 @@
+#pragma once
+// @generated by torchgen/gen.py from DispatchKeyFunction.h
+
+// NB: The implementing C++ file is RegisterDispatchKey.cpp
+
+// The only #includes we need are for custom classes that have defaults in the C++ API
+#include <c10/core/MemoryFormat.h>
+#include <c10/core/Scalar.h>
+#include <ATen/core/Reduction.h>
+
+// Forward declarations of any types needed in the operator signatures.
+// We can't directly include these classes because it will cause circular include dependencies.
+// This file is included by TensorBody.h, which defines the Tensor class.
+#include <ATen/core/ATen_fwd.h>
+
+namespace at {
+
+namespace compositeimplicitautograd {
+
+TORCH_API at::Tensor upsample_bilinear2d(const at::Tensor & input, at::OptionalIntArrayRef output_size, bool align_corners, ::std::optional<at::ArrayRef<double>> scale_factors);
+TORCH_API at::Tensor upsample_bilinear2d_symint(const at::Tensor & input, at::OptionalSymIntArrayRef output_size, bool align_corners, ::std::optional<at::ArrayRef<double>> scale_factors);
+
+} // namespace compositeimplicitautograd
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/view_as_real_cuda_dispatch.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/view_as_real_cuda_dispatch.h
new file mode 100644
index 0000000000000000000000000000000000000000..f73e2a9b5f1aede026ad7cd32488f728cd2eb2d6
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/view_as_real_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 view_as_real(const at::Tensor & self);
+
+} // namespace cuda
+} // namespace at
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/view_ops.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/view_ops.h
new file mode 100644
index 0000000000000000000000000000000000000000..a13519b0c39bf434905b95994d56797103bdc45c
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/view_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 view {
+  using schema = at::Tensor (const at::Tensor &, c10::SymIntArrayRef);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::view")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "view(Tensor(a) self, SymInt[] size) -> Tensor(a)")
+  static at::Tensor call(const at::Tensor & self, c10::SymIntArrayRef size);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, c10::SymIntArrayRef size);
+};
+
+struct TORCH_API view_dtype {
+  using schema = at::Tensor (const at::Tensor &, at::ScalarType);
+  using ptr_schema = schema*;
+  // See Note [static constexpr char* members for windows NVCC]
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(name, "aten::view")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(overload_name, "dtype")
+  STATIC_CONSTEXPR_STR_INL_EXCEPT_WIN_CUDA(schema_str, "view.dtype(Tensor(a) self, ScalarType dtype) -> Tensor(a)")
+  static at::Tensor call(const at::Tensor & self, at::ScalarType dtype);
+  static at::Tensor redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & self, at::ScalarType dtype);
+};
+
+}} // namespace at::_ops
diff --git a/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/zeros_like_native.h b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/zeros_like_native.h
new file mode 100644
index 0000000000000000000000000000000000000000..eb0d1cd08b388cc473cccf67e0bfc50f0eec0cb0
--- /dev/null
+++ b/.venv/lib/python3.11/site-packages/torch/include/ATen/ops/zeros_like_native.h
@@ -0,0 +1,22 @@
+#pragma once
+
+// @generated by torchgen/gen.py from NativeFunction.h
+
+#include <c10/core/Scalar.h>
+#include <c10/core/Storage.h>
+#include <c10/core/TensorOptions.h>
+#include <c10/util/Deprecated.h>
+#include <optional>
+#include <c10/core/QScheme.h>
+#include <ATen/core/Reduction.h>
+#include <ATen/core/Tensor.h>
+#include <tuple>
+#include <vector>
+
+
+namespace at {
+namespace native {
+TORCH_API at::Tensor zeros_like(const at::Tensor & self, ::std::optional<at::ScalarType> dtype={}, ::std::optional<at::Layout> layout={}, ::std::optional<at::Device> device={}, ::std::optional<bool> pin_memory={}, ::std::optional<at::MemoryFormat> memory_format=::std::nullopt);
+TORCH_API at::Tensor & zeros_like_out(const at::Tensor & self, ::std::optional<at::MemoryFormat> memory_format, at::Tensor & out);
+} // namespace native
+} // namespace at