Hugging Face's logo

koichi12
/
llm_tutorial

Model card Files
xet
Community
koichi12 commited on
Commit
f3e6968
·
verified ·
1 Parent(s): 104185d

Add files using upload-large-folder tool

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .venv/lib/python3.11/site-packages/torch/include/ATen/ATen.h +37 -0
  2. .venv/lib/python3.11/site-packages/torch/include/ATen/Backtrace.h +2 -0
  3. .venv/lib/python3.11/site-packages/torch/include/ATen/BlasBackend.h +27 -0
  4. .venv/lib/python3.11/site-packages/torch/include/ATen/CPUFunctions_inl.h +540 -0
  5. .venv/lib/python3.11/site-packages/torch/include/ATen/CUDAFunctions.h +29 -0
  6. .venv/lib/python3.11/site-packages/torch/include/ATen/CUDAFunctions_inl.h +623 -0
  7. .venv/lib/python3.11/site-packages/torch/include/ATen/CachedTensorUtils.h +24 -0
  8. .venv/lib/python3.11/site-packages/torch/include/ATen/CollapseDims.h +94 -0
  9. .venv/lib/python3.11/site-packages/torch/include/ATen/CompositeExplicitAutogradFunctions.h +29 -0
  10. .venv/lib/python3.11/site-packages/torch/include/ATen/CompositeExplicitAutogradNonFunctionalFunctions.h +29 -0
  11. .venv/lib/python3.11/site-packages/torch/include/ATen/CompositeExplicitAutogradNonFunctionalFunctions_inl.h +323 -0
  12. .venv/lib/python3.11/site-packages/torch/include/ATen/CompositeImplicitAutogradFunctions_inl.h +502 -0
  13. .venv/lib/python3.11/site-packages/torch/include/ATen/DLConvertor.h +25 -0
  14. .venv/lib/python3.11/site-packages/torch/include/ATen/DeviceGuard.h +41 -0
  15. .venv/lib/python3.11/site-packages/torch/include/ATen/Dispatch_v2.h +186 -0
  16. .venv/lib/python3.11/site-packages/torch/include/ATen/DynamicLibrary.h +34 -0
  17. .venv/lib/python3.11/site-packages/torch/include/ATen/ExpandUtils.h +527 -0
  18. .venv/lib/python3.11/site-packages/torch/include/ATen/Functions.h +1454 -0
  19. .venv/lib/python3.11/site-packages/torch/include/ATen/Generator.h +2 -0
  20. .venv/lib/python3.11/site-packages/torch/include/ATen/LinalgBackend.h +31 -0
  21. .venv/lib/python3.11/site-packages/torch/include/ATen/MemoryOverlap.h +42 -0
  22. .venv/lib/python3.11/site-packages/torch/include/ATen/NativeMetaFunctions.h +1330 -0
  23. .venv/lib/python3.11/site-packages/torch/include/ATen/NumericUtils.h +203 -0
  24. .venv/lib/python3.11/site-packages/torch/include/ATen/OpaqueTensorImpl.h +187 -0
  25. .venv/lib/python3.11/site-packages/torch/include/ATen/Operators.h +1385 -0
  26. .venv/lib/python3.11/site-packages/torch/include/ATen/Parallel-inl.h +93 -0
  27. .venv/lib/python3.11/site-packages/torch/include/ATen/ParallelNative.h +15 -0
  28. .venv/lib/python3.11/site-packages/torch/include/ATen/ParallelOpenMP.h +54 -0
  29. .venv/lib/python3.11/site-packages/torch/include/ATen/PythonTorchFunctionTLS.h +36 -0
  30. .venv/lib/python3.11/site-packages/torch/include/ATen/RedispatchFunctions.h +0 -0
  31. .venv/lib/python3.11/site-packages/torch/include/ATen/SmallVector.h +2 -0
  32. .venv/lib/python3.11/site-packages/torch/include/ATen/SparseTensorImpl.h +421 -0
  33. .venv/lib/python3.11/site-packages/torch/include/ATen/StorageUtils.h +49 -0
  34. .venv/lib/python3.11/site-packages/torch/include/ATen/TensorAccessor.h +2 -0
  35. .venv/lib/python3.11/site-packages/torch/include/ATen/TensorIndexing.h +737 -0
  36. .venv/lib/python3.11/site-packages/torch/include/ATen/TensorIteratorInternal.h +72 -0
  37. .venv/lib/python3.11/site-packages/torch/include/ATen/TensorOptions.h +2 -0
  38. .venv/lib/python3.11/site-packages/torch/include/ATen/TensorSubclassLikeUtils.h +88 -0
  39. .venv/lib/python3.11/site-packages/torch/include/ATen/TensorUtils.h +190 -0
  40. .venv/lib/python3.11/site-packages/torch/include/ATen/TypeDefault.h +30 -0
  41. .venv/lib/python3.11/site-packages/torch/include/ATen/Utils.h +134 -0
  42. .venv/lib/python3.11/site-packages/torch/include/ATen/VmapGeneratedPlumbing.h +0 -0
  43. .venv/lib/python3.11/site-packages/torch/include/ATen/cpp_custom_type_hack.h +110 -0
  44. .venv/lib/python3.11/site-packages/torch/include/ATen/cuda/ATenCUDAGeneral.h +9 -0
  45. .venv/lib/python3.11/site-packages/torch/include/ATen/cuda/ApplyGridUtils.cuh +47 -0
  46. .venv/lib/python3.11/site-packages/torch/include/ATen/cuda/Atomic.cuh +514 -0
  47. .venv/lib/python3.11/site-packages/torch/include/ATen/cuda/CUDAApplyUtils.cuh +537 -0
  48. .venv/lib/python3.11/site-packages/torch/include/ATen/cuda/CUDABlas.h +358 -0
  49. .venv/lib/python3.11/site-packages/torch/include/ATen/cuda/CUDAConfig.h +19 -0
  50. .venv/lib/python3.11/site-packages/torch/include/ATen/cuda/CUDAContext.h +9 -0
.venv/lib/python3.11/site-packages/torch/include/ATen/ATen.h ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #if !defined(_MSC_VER) && __cplusplus < 201703L
4
+ #error C++17 or later compatible compiler is required to use ATen.
5
+ #endif
6
+
7
+ #include <ATen/Context.h>
8
+ #include <ATen/Device.h>
9
+ #include <ATen/DeviceGuard.h>
10
+ #include <ATen/DimVector.h>
11
+ #include <ATen/Dispatch.h>
12
+ #include <ATen/Formatting.h>
13
+ #include <ATen/Functions.h>
14
+ #include <ATen/NamedTensor.h>
15
+ #include <ATen/ScalarOps.h>
16
+ #include <ATen/Tensor.h>
17
+ #include <ATen/TensorGeometry.h>
18
+ #include <ATen/TensorIndexing.h>
19
+ #include <ATen/TensorOperators.h>
20
+ #include <ATen/Version.h>
21
+ #include <ATen/core/ATenGeneral.h>
22
+ #include <ATen/core/Generator.h>
23
+ #include <ATen/core/Reduction.h>
24
+ #include <ATen/core/Scalar.h>
25
+ #include <ATen/core/UnsafeFromTH.h>
26
+ #include <ATen/core/ivalue.h>
27
+ #include <ATen/core/jit_type.h>
28
+ #include <c10/core/Allocator.h>
29
+ #include <c10/core/InferenceMode.h>
30
+ #include <c10/core/Layout.h>
31
+ #include <c10/core/Storage.h>
32
+ #include <c10/core/TensorOptions.h>
33
+ #include <c10/util/Exception.h>
34
+
35
+ // TODO: try to remove this
36
+ // There is some back story, see https://github.com/pytorch/pytorch/issues/48684
37
+ #include <ATen/NativeFunctions.h>
.venv/lib/python3.11/site-packages/torch/include/ATen/Backtrace.h ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ #pragma once
2
+ #include <ATen/core/Backtrace.h>
.venv/lib/python3.11/site-packages/torch/include/ATen/BlasBackend.h ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <c10/util/Exception.h>
4
+
5
+ #include <ostream>
6
+ #include <string>
7
+
8
+ namespace at {
9
+
10
+ enum class BlasBackend : int8_t { Cublas, Cublaslt };
11
+
12
+ inline std::string BlasBackendToString(at::BlasBackend backend) {
13
+ switch (backend) {
14
+ case BlasBackend::Cublas:
15
+ return "at::BlasBackend::Cublas";
16
+ case BlasBackend::Cublaslt:
17
+ return "at::BlasBackend::Cublaslt";
18
+ default:
19
+ TORCH_CHECK(false, "Unknown blas backend");
20
+ }
21
+ }
22
+
23
+ inline std::ostream& operator<<(std::ostream& stream, at::BlasBackend backend) {
24
+ return stream << BlasBackendToString(backend);
25
+ }
26
+
27
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/CPUFunctions_inl.h ADDED
@@ -0,0 +1,540 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+ // @generated by torchgen/gen.py from DispatchKeyFunctions_inl.h
3
+
4
+ // NB: The implementing C++ file is RegisterDispatchKey.cpp
5
+
6
+ // The only #includes we need are for custom classes that have defaults in the C++ API
7
+ #include <c10/core/MemoryFormat.h>
8
+ #include <c10/core/Scalar.h>
9
+ #include <ATen/core/Reduction.h>
10
+
11
+ #if defined(AT_PER_OPERATOR_HEADERS) && defined(TORCH_ASSERT_ONLY_METHOD_OPERATORS)
12
+ #error This change adds a dependency on all pytorch operators, meaning the \
13
+ file will need to be re-compiled every time an operator is changed or added. \
14
+ Consider including a specific operator from \
15
+ <ATen/ops/{my_operator}_cpu_dispatch.h>. \
16
+ See NOTE [TORCH_ASSERT_ONLY_METHOD_OPERATORS].
17
+ #endif
18
+
19
+ #include <ATen/ops/_adaptive_avg_pool2d_cpu_dispatch.h>
20
+ #include <ATen/ops/_adaptive_avg_pool2d_backward_cpu_dispatch.h>
21
+ #include <ATen/ops/_adaptive_avg_pool3d_cpu_dispatch.h>
22
+ #include <ATen/ops/_adaptive_avg_pool3d_backward_cpu_dispatch.h>
23
+ #include <ATen/ops/_add_relu_cpu_dispatch.h>
24
+ #include <ATen/ops/_addmm_activation_cpu_dispatch.h>
25
+ #include <ATen/ops/_aminmax_cpu_dispatch.h>
26
+ #include <ATen/ops/_amp_foreach_non_finite_check_and_unscale_cpu_dispatch.h>
27
+ #include <ATen/ops/_amp_update_scale_cpu_dispatch.h>
28
+ #include <ATen/ops/_assert_async_cpu_dispatch.h>
29
+ #include <ATen/ops/_batch_norm_with_update_cpu_dispatch.h>
30
+ #include <ATen/ops/_cdist_backward_cpu_dispatch.h>
31
+ #include <ATen/ops/_cdist_forward_cpu_dispatch.h>
32
+ #include <ATen/ops/_cholesky_solve_helper_cpu_dispatch.h>
33
+ #include <ATen/ops/_compute_linear_combination_cpu_dispatch.h>
34
+ #include <ATen/ops/_convert_indices_from_coo_to_csr_cpu_dispatch.h>
35
+ #include <ATen/ops/_convert_indices_from_csr_to_coo_cpu_dispatch.h>
36
+ #include <ATen/ops/_convert_weight_to_int4pack_cpu_dispatch.h>
37
+ #include <ATen/ops/_ctc_loss_cpu_dispatch.h>
38
+ #include <ATen/ops/_ctc_loss_backward_cpu_dispatch.h>
39
+ #include <ATen/ops/_cummax_helper_cpu_dispatch.h>
40
+ #include <ATen/ops/_cummin_helper_cpu_dispatch.h>
41
+ #include <ATen/ops/_dirichlet_grad_cpu_dispatch.h>
42
+ #include <ATen/ops/_efficientzerotensor_cpu_dispatch.h>
43
+ #include <ATen/ops/_embedding_bag_cpu_dispatch.h>
44
+ #include <ATen/ops/_embedding_bag_backward_cpu_dispatch.h>
45
+ #include <ATen/ops/_embedding_bag_dense_backward_cpu_dispatch.h>
46
+ #include <ATen/ops/_embedding_bag_forward_only_cpu_dispatch.h>
47
+ #include <ATen/ops/_embedding_bag_per_sample_weights_backward_cpu_dispatch.h>
48
+ #include <ATen/ops/_empty_affine_quantized_cpu_dispatch.h>
49
+ #include <ATen/ops/_empty_per_channel_affine_quantized_cpu_dispatch.h>
50
+ #include <ATen/ops/_fake_quantize_learnable_per_channel_affine_cpu_dispatch.h>
51
+ #include <ATen/ops/_fake_quantize_learnable_per_channel_affine_backward_cpu_dispatch.h>
52
+ #include <ATen/ops/_fake_quantize_learnable_per_tensor_affine_cpu_dispatch.h>
53
+ #include <ATen/ops/_fake_quantize_learnable_per_tensor_affine_backward_cpu_dispatch.h>
54
+ #include <ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams_cpu_dispatch.h>
55
+ #include <ATen/ops/_fft_c2c_cpu_dispatch.h>
56
+ #include <ATen/ops/_fft_c2r_cpu_dispatch.h>
57
+ #include <ATen/ops/_fft_r2c_cpu_dispatch.h>
58
+ #include <ATen/ops/_foobar_cpu_dispatch.h>
59
+ #include <ATen/ops/_functional_assert_async_cpu_dispatch.h>
60
+ #include <ATen/ops/_fused_adagrad_cpu_dispatch.h>
61
+ #include <ATen/ops/_fused_adam_cpu_dispatch.h>
62
+ #include <ATen/ops/_fused_adamw_cpu_dispatch.h>
63
+ #include <ATen/ops/_fused_moving_avg_obs_fq_helper_cpu_dispatch.h>
64
+ #include <ATen/ops/_fused_sdp_choice_cpu_dispatch.h>
65
+ #include <ATen/ops/_fused_sgd_cpu_dispatch.h>
66
+ #include <ATen/ops/_histogramdd_bin_edges_cpu_dispatch.h>
67
+ #include <ATen/ops/_histogramdd_from_bin_cts_cpu_dispatch.h>
68
+ #include <ATen/ops/_histogramdd_from_bin_tensors_cpu_dispatch.h>
69
+ #include <ATen/ops/_index_put_impl_cpu_dispatch.h>
70
+ #include <ATen/ops/_int_mm_cpu_dispatch.h>
71
+ #include <ATen/ops/_jagged_to_padded_dense_forward_cpu_dispatch.h>
72
+ #include <ATen/ops/_linalg_det_cpu_dispatch.h>
73
+ #include <ATen/ops/_linalg_eigh_cpu_dispatch.h>
74
+ #include <ATen/ops/_linalg_eigvals_cpu_dispatch.h>
75
+ #include <ATen/ops/_linalg_slogdet_cpu_dispatch.h>
76
+ #include <ATen/ops/_linalg_solve_ex_cpu_dispatch.h>
77
+ #include <ATen/ops/_linalg_svd_cpu_dispatch.h>
78
+ #include <ATen/ops/_local_scalar_dense_cpu_dispatch.h>
79
+ #include <ATen/ops/_log_softmax_cpu_dispatch.h>
80
+ #include <ATen/ops/_log_softmax_backward_data_cpu_dispatch.h>
81
+ #include <ATen/ops/_logcumsumexp_cpu_dispatch.h>
82
+ #include <ATen/ops/_make_dep_token_cpu_dispatch.h>
83
+ #include <ATen/ops/_make_per_channel_quantized_tensor_cpu_dispatch.h>
84
+ #include <ATen/ops/_make_per_tensor_quantized_tensor_cpu_dispatch.h>
85
+ #include <ATen/ops/_masked_softmax_cpu_dispatch.h>
86
+ #include <ATen/ops/_masked_softmax_backward_cpu_dispatch.h>
87
+ #include <ATen/ops/_native_batch_norm_legit_cpu_dispatch.h>
88
+ #include <ATen/ops/_native_multi_head_attention_cpu_dispatch.h>
89
+ #include <ATen/ops/_nested_compute_contiguous_strides_offsets_cpu_dispatch.h>
90
+ #include <ATen/ops/_nested_from_padded_cpu_dispatch.h>
91
+ #include <ATen/ops/_nested_tensor_from_mask_cpu_dispatch.h>
92
+ #include <ATen/ops/_nested_tensor_from_mask_left_aligned_cpu_dispatch.h>
93
+ #include <ATen/ops/_nested_view_from_buffer_cpu_dispatch.h>
94
+ #include <ATen/ops/_padded_dense_to_jagged_forward_cpu_dispatch.h>
95
+ #include <ATen/ops/_pdist_backward_cpu_dispatch.h>
96
+ #include <ATen/ops/_pdist_forward_cpu_dispatch.h>
97
+ #include <ATen/ops/_prelu_kernel_cpu_dispatch.h>
98
+ #include <ATen/ops/_prelu_kernel_backward_cpu_dispatch.h>
99
+ #include <ATen/ops/_reshape_alias_cpu_dispatch.h>
100
+ #include <ATen/ops/_sample_dirichlet_cpu_dispatch.h>
101
+ #include <ATen/ops/_scaled_dot_product_flash_attention_for_cpu_cpu_dispatch.h>
102
+ #include <ATen/ops/_scaled_dot_product_flash_attention_for_cpu_backward_cpu_dispatch.h>
103
+ #include <ATen/ops/_segment_reduce_backward_cpu_dispatch.h>
104
+ #include <ATen/ops/_slow_conv2d_backward_cpu_dispatch.h>
105
+ #include <ATen/ops/_slow_conv2d_forward_cpu_dispatch.h>
106
+ #include <ATen/ops/_softmax_cpu_dispatch.h>
107
+ #include <ATen/ops/_softmax_backward_data_cpu_dispatch.h>
108
+ #include <ATen/ops/_spdiags_cpu_dispatch.h>
109
+ #include <ATen/ops/_stack_cpu_dispatch.h>
110
+ #include <ATen/ops/_standard_gamma_cpu_dispatch.h>
111
+ #include <ATen/ops/_standard_gamma_grad_cpu_dispatch.h>
112
+ #include <ATen/ops/_test_functorch_fallback_cpu_dispatch.h>
113
+ #include <ATen/ops/_test_optional_filled_intlist_cpu_dispatch.h>
114
+ #include <ATen/ops/_test_optional_floatlist_cpu_dispatch.h>
115
+ #include <ATen/ops/_test_optional_intlist_cpu_dispatch.h>
116
+ #include <ATen/ops/_to_sparse_cpu_dispatch.h>
117
+ #include <ATen/ops/_to_sparse_bsc_cpu_dispatch.h>
118
+ #include <ATen/ops/_to_sparse_bsr_cpu_dispatch.h>
119
+ #include <ATen/ops/_to_sparse_csc_cpu_dispatch.h>
120
+ #include <ATen/ops/_to_sparse_csr_cpu_dispatch.h>
121
+ #include <ATen/ops/_transform_bias_rescale_qkv_cpu_dispatch.h>
122
+ #include <ATen/ops/_transformer_encoder_layer_fwd_cpu_dispatch.h>
123
+ #include <ATen/ops/_unique_cpu_dispatch.h>
124
+ #include <ATen/ops/_unique2_cpu_dispatch.h>
125
+ #include <ATen/ops/_upsample_bicubic2d_aa_cpu_dispatch.h>
126
+ #include <ATen/ops/_upsample_bicubic2d_aa_backward_cpu_dispatch.h>
127
+ #include <ATen/ops/_upsample_bilinear2d_aa_cpu_dispatch.h>
128
+ #include <ATen/ops/_upsample_bilinear2d_aa_backward_cpu_dispatch.h>
129
+ #include <ATen/ops/_upsample_nearest_exact1d_cpu_dispatch.h>
130
+ #include <ATen/ops/_upsample_nearest_exact1d_backward_cpu_dispatch.h>
131
+ #include <ATen/ops/_upsample_nearest_exact2d_cpu_dispatch.h>
132
+ #include <ATen/ops/_upsample_nearest_exact2d_backward_cpu_dispatch.h>
133
+ #include <ATen/ops/_upsample_nearest_exact3d_cpu_dispatch.h>
134
+ #include <ATen/ops/_upsample_nearest_exact3d_backward_cpu_dispatch.h>
135
+ #include <ATen/ops/_validate_compressed_sparse_indices_cpu_dispatch.h>
136
+ #include <ATen/ops/_weight_int4pack_mm_cpu_dispatch.h>
137
+ #include <ATen/ops/_weight_int8pack_mm_cpu_dispatch.h>
138
+ #include <ATen/ops/_weight_norm_interface_cpu_dispatch.h>
139
+ #include <ATen/ops/_weight_norm_interface_backward_cpu_dispatch.h>
140
+ #include <ATen/ops/abs_cpu_dispatch.h>
141
+ #include <ATen/ops/acos_cpu_dispatch.h>
142
+ #include <ATen/ops/acosh_cpu_dispatch.h>
143
+ #include <ATen/ops/adaptive_avg_pool2d_cpu_dispatch.h>
144
+ #include <ATen/ops/adaptive_avg_pool3d_cpu_dispatch.h>
145
+ #include <ATen/ops/adaptive_avg_pool3d_backward_cpu_dispatch.h>
146
+ #include <ATen/ops/adaptive_max_pool2d_cpu_dispatch.h>
147
+ #include <ATen/ops/adaptive_max_pool2d_backward_cpu_dispatch.h>
148
+ #include <ATen/ops/adaptive_max_pool3d_cpu_dispatch.h>
149
+ #include <ATen/ops/adaptive_max_pool3d_backward_cpu_dispatch.h>
150
+ #include <ATen/ops/add_cpu_dispatch.h>
151
+ #include <ATen/ops/addbmm_cpu_dispatch.h>
152
+ #include <ATen/ops/addcdiv_cpu_dispatch.h>
153
+ #include <ATen/ops/addcmul_cpu_dispatch.h>
154
+ #include <ATen/ops/addmm_cpu_dispatch.h>
155
+ #include <ATen/ops/addmv_cpu_dispatch.h>
156
+ #include <ATen/ops/addr_cpu_dispatch.h>
157
+ #include <ATen/ops/all_cpu_dispatch.h>
158
+ #include <ATen/ops/amax_cpu_dispatch.h>
159
+ #include <ATen/ops/amin_cpu_dispatch.h>
160
+ #include <ATen/ops/aminmax_cpu_dispatch.h>
161
+ #include <ATen/ops/angle_cpu_dispatch.h>
162
+ #include <ATen/ops/any_cpu_dispatch.h>
163
+ #include <ATen/ops/arange_cpu_dispatch.h>
164
+ #include <ATen/ops/argmax_cpu_dispatch.h>
165
+ #include <ATen/ops/argmin_cpu_dispatch.h>
166
+ #include <ATen/ops/as_strided_cpu_dispatch.h>
167
+ #include <ATen/ops/asin_cpu_dispatch.h>
168
+ #include <ATen/ops/asinh_cpu_dispatch.h>
169
+ #include <ATen/ops/atan_cpu_dispatch.h>
170
+ #include <ATen/ops/atan2_cpu_dispatch.h>
171
+ #include <ATen/ops/atanh_cpu_dispatch.h>
172
+ #include <ATen/ops/avg_pool2d_cpu_dispatch.h>
173
+ #include <ATen/ops/avg_pool2d_backward_cpu_dispatch.h>
174
+ #include <ATen/ops/avg_pool3d_cpu_dispatch.h>
175
+ #include <ATen/ops/avg_pool3d_backward_cpu_dispatch.h>
176
+ #include <ATen/ops/baddbmm_cpu_dispatch.h>
177
+ #include <ATen/ops/batch_norm_backward_cpu_dispatch.h>
178
+ #include <ATen/ops/batch_norm_update_stats_cpu_dispatch.h>
179
+ #include <ATen/ops/bernoulli_cpu_dispatch.h>
180
+ #include <ATen/ops/binary_cross_entropy_cpu_dispatch.h>
181
+ #include <ATen/ops/binary_cross_entropy_backward_cpu_dispatch.h>
182
+ #include <ATen/ops/bincount_cpu_dispatch.h>
183
+ #include <ATen/ops/binomial_cpu_dispatch.h>
184
+ #include <ATen/ops/bitwise_and_cpu_dispatch.h>
185
+ #include <ATen/ops/bitwise_left_shift_cpu_dispatch.h>
186
+ #include <ATen/ops/bitwise_not_cpu_dispatch.h>
187
+ #include <ATen/ops/bitwise_or_cpu_dispatch.h>
188
+ #include <ATen/ops/bitwise_right_shift_cpu_dispatch.h>
189
+ #include <ATen/ops/bitwise_xor_cpu_dispatch.h>
190
+ #include <ATen/ops/bmm_cpu_dispatch.h>
191
+ #include <ATen/ops/bucketize_cpu_dispatch.h>
192
+ #include <ATen/ops/cat_cpu_dispatch.h>
193
+ #include <ATen/ops/cauchy_cpu_dispatch.h>
194
+ #include <ATen/ops/ceil_cpu_dispatch.h>
195
+ #include <ATen/ops/channel_shuffle_cpu_dispatch.h>
196
+ #include <ATen/ops/cholesky_cpu_dispatch.h>
197
+ #include <ATen/ops/cholesky_inverse_cpu_dispatch.h>
198
+ #include <ATen/ops/clamp_cpu_dispatch.h>
199
+ #include <ATen/ops/clamp_max_cpu_dispatch.h>
200
+ #include <ATen/ops/clamp_min_cpu_dispatch.h>
201
+ #include <ATen/ops/col2im_cpu_dispatch.h>
202
+ #include <ATen/ops/complex_cpu_dispatch.h>
203
+ #include <ATen/ops/conj_physical_cpu_dispatch.h>
204
+ #include <ATen/ops/copysign_cpu_dispatch.h>
205
+ #include <ATen/ops/cos_cpu_dispatch.h>
206
+ #include <ATen/ops/cosh_cpu_dispatch.h>
207
+ #include <ATen/ops/count_nonzero_cpu_dispatch.h>
208
+ #include <ATen/ops/cumprod_cpu_dispatch.h>
209
+ #include <ATen/ops/cumsum_cpu_dispatch.h>
210
+ #include <ATen/ops/dequantize_cpu_dispatch.h>
211
+ #include <ATen/ops/digamma_cpu_dispatch.h>
212
+ #include <ATen/ops/div_cpu_dispatch.h>
213
+ #include <ATen/ops/dot_cpu_dispatch.h>
214
+ #include <ATen/ops/elu_cpu_dispatch.h>
215
+ #include <ATen/ops/elu_backward_cpu_dispatch.h>
216
+ #include <ATen/ops/embedding_dense_backward_cpu_dispatch.h>
217
+ #include <ATen/ops/embedding_renorm_cpu_dispatch.h>
218
+ #include <ATen/ops/empty_cpu_dispatch.h>
219
+ #include <ATen/ops/empty_strided_cpu_dispatch.h>
220
+ #include <ATen/ops/eq_cpu_dispatch.h>
221
+ #include <ATen/ops/equal_cpu_dispatch.h>
222
+ #include <ATen/ops/erf_cpu_dispatch.h>
223
+ #include <ATen/ops/erfc_cpu_dispatch.h>
224
+ #include <ATen/ops/erfinv_cpu_dispatch.h>
225
+ #include <ATen/ops/exp_cpu_dispatch.h>
226
+ #include <ATen/ops/exp2_cpu_dispatch.h>
227
+ #include <ATen/ops/expm1_cpu_dispatch.h>
228
+ #include <ATen/ops/exponential_cpu_dispatch.h>
229
+ #include <ATen/ops/eye_cpu_dispatch.h>
230
+ #include <ATen/ops/fake_quantize_per_channel_affine_cachemask_cpu_dispatch.h>
231
+ #include <ATen/ops/fake_quantize_per_tensor_affine_cachemask_cpu_dispatch.h>
232
+ #include <ATen/ops/fill_cpu_dispatch.h>
233
+ #include <ATen/ops/flip_cpu_dispatch.h>
234
+ #include <ATen/ops/floor_cpu_dispatch.h>
235
+ #include <ATen/ops/floor_divide_cpu_dispatch.h>
236
+ #include <ATen/ops/fmax_cpu_dispatch.h>
237
+ #include <ATen/ops/fmin_cpu_dispatch.h>
238
+ #include <ATen/ops/fmod_cpu_dispatch.h>
239
+ #include <ATen/ops/frac_cpu_dispatch.h>
240
+ #include <ATen/ops/fractional_max_pool2d_cpu_dispatch.h>
241
+ #include <ATen/ops/fractional_max_pool2d_backward_cpu_dispatch.h>
242
+ #include <ATen/ops/fractional_max_pool3d_cpu_dispatch.h>
243
+ #include <ATen/ops/fractional_max_pool3d_backward_cpu_dispatch.h>
244
+ #include <ATen/ops/frexp_cpu_dispatch.h>
245
+ #include <ATen/ops/from_file_cpu_dispatch.h>
246
+ #include <ATen/ops/gather_cpu_dispatch.h>
247
+ #include <ATen/ops/gcd_cpu_dispatch.h>
248
+ #include <ATen/ops/ge_cpu_dispatch.h>
249
+ #include <ATen/ops/gelu_cpu_dispatch.h>
250
+ #include <ATen/ops/gelu_backward_cpu_dispatch.h>
251
+ #include <ATen/ops/geometric_cpu_dispatch.h>
252
+ #include <ATen/ops/geqrf_cpu_dispatch.h>
253
+ #include <ATen/ops/glu_cpu_dispatch.h>
254
+ #include <ATen/ops/glu_backward_cpu_dispatch.h>
255
+ #include <ATen/ops/glu_backward_jvp_cpu_dispatch.h>
256
+ #include <ATen/ops/glu_jvp_cpu_dispatch.h>
257
+ #include <ATen/ops/grid_sampler_2d_cpu_dispatch.h>
258
+ #include <ATen/ops/grid_sampler_2d_backward_cpu_dispatch.h>
259
+ #include <ATen/ops/grid_sampler_3d_cpu_dispatch.h>
260
+ #include <ATen/ops/grid_sampler_3d_backward_cpu_dispatch.h>
261
+ #include <ATen/ops/gt_cpu_dispatch.h>
262
+ #include <ATen/ops/hardshrink_cpu_dispatch.h>
263
+ #include <ATen/ops/hardshrink_backward_cpu_dispatch.h>
264
+ #include <ATen/ops/hardsigmoid_cpu_dispatch.h>
265
+ #include <ATen/ops/hardsigmoid_backward_cpu_dispatch.h>
266
+ #include <ATen/ops/hardswish_cpu_dispatch.h>
267
+ #include <ATen/ops/hardswish_backward_cpu_dispatch.h>
268
+ #include <ATen/ops/hardtanh_cpu_dispatch.h>
269
+ #include <ATen/ops/hardtanh_backward_cpu_dispatch.h>
270
+ #include <ATen/ops/heaviside_cpu_dispatch.h>
271
+ #include <ATen/ops/histc_cpu_dispatch.h>
272
+ #include <ATen/ops/histogram_cpu_dispatch.h>
273
+ #include <ATen/ops/huber_loss_cpu_dispatch.h>
274
+ #include <ATen/ops/huber_loss_backward_cpu_dispatch.h>
275
+ #include <ATen/ops/hypot_cpu_dispatch.h>
276
+ #include <ATen/ops/i0_cpu_dispatch.h>
277
+ #include <ATen/ops/igamma_cpu_dispatch.h>
278
+ #include <ATen/ops/igammac_cpu_dispatch.h>
279
+ #include <ATen/ops/im2col_cpu_dispatch.h>
280
+ #include <ATen/ops/index_cpu_dispatch.h>
281
+ #include <ATen/ops/index_add_cpu_dispatch.h>
282
+ #include <ATen/ops/index_copy_cpu_dispatch.h>
283
+ #include <ATen/ops/index_fill_cpu_dispatch.h>
284
+ #include <ATen/ops/index_reduce_cpu_dispatch.h>
285
+ #include <ATen/ops/index_select_cpu_dispatch.h>
286
+ #include <ATen/ops/is_set_to_cpu_dispatch.h>
287
+ #include <ATen/ops/isin_cpu_dispatch.h>
288
+ #include <ATen/ops/isnan_cpu_dispatch.h>
289
+ #include <ATen/ops/isneginf_cpu_dispatch.h>
290
+ #include <ATen/ops/isposinf_cpu_dispatch.h>
291
+ #include <ATen/ops/kthvalue_cpu_dispatch.h>
292
+ #include <ATen/ops/lcm_cpu_dispatch.h>
293
+ #include <ATen/ops/le_cpu_dispatch.h>
294
+ #include <ATen/ops/leaky_relu_cpu_dispatch.h>
295
+ #include <ATen/ops/leaky_relu_backward_cpu_dispatch.h>
296
+ #include <ATen/ops/lerp_cpu_dispatch.h>
297
+ #include <ATen/ops/lgamma_cpu_dispatch.h>
298
+ #include <ATen/ops/linalg_cholesky_ex_cpu_dispatch.h>
299
+ #include <ATen/ops/linalg_cross_cpu_dispatch.h>
300
+ #include <ATen/ops/linalg_eig_cpu_dispatch.h>
301
+ #include <ATen/ops/linalg_eigvals_cpu_dispatch.h>
302
+ #include <ATen/ops/linalg_householder_product_cpu_dispatch.h>
303
+ #include <ATen/ops/linalg_inv_ex_cpu_dispatch.h>
304
+ #include <ATen/ops/linalg_ldl_factor_ex_cpu_dispatch.h>
305
+ #include <ATen/ops/linalg_ldl_solve_cpu_dispatch.h>
306
+ #include <ATen/ops/linalg_lstsq_cpu_dispatch.h>
307
+ #include <ATen/ops/linalg_lu_cpu_dispatch.h>
308
+ #include <ATen/ops/linalg_lu_factor_ex_cpu_dispatch.h>
309
+ #include <ATen/ops/linalg_lu_solve_cpu_dispatch.h>
310
+ #include <ATen/ops/linalg_matrix_exp_cpu_dispatch.h>
311
+ #include <ATen/ops/linalg_qr_cpu_dispatch.h>
312
+ #include <ATen/ops/linalg_solve_triangular_cpu_dispatch.h>
313
+ #include <ATen/ops/linalg_vector_norm_cpu_dispatch.h>
314
+ #include <ATen/ops/linspace_cpu_dispatch.h>
315
+ #include <ATen/ops/log_cpu_dispatch.h>
316
+ #include <ATen/ops/log10_cpu_dispatch.h>
317
+ #include <ATen/ops/log1p_cpu_dispatch.h>
318
+ #include <ATen/ops/log2_cpu_dispatch.h>
319
+ #include <ATen/ops/log_normal_cpu_dispatch.h>
320
+ #include <ATen/ops/log_sigmoid_backward_cpu_dispatch.h>
321
+ #include <ATen/ops/log_sigmoid_forward_cpu_dispatch.h>
322
+ #include <ATen/ops/logaddexp_cpu_dispatch.h>
323
+ #include <ATen/ops/logaddexp2_cpu_dispatch.h>
324
+ #include <ATen/ops/logical_and_cpu_dispatch.h>
325
+ #include <ATen/ops/logical_not_cpu_dispatch.h>
326
+ #include <ATen/ops/logical_or_cpu_dispatch.h>
327
+ #include <ATen/ops/logical_xor_cpu_dispatch.h>
328
+ #include <ATen/ops/logit_cpu_dispatch.h>
329
+ #include <ATen/ops/logit_backward_cpu_dispatch.h>
330
+ #include <ATen/ops/logspace_cpu_dispatch.h>
331
+ #include <ATen/ops/lshift_cpu_dispatch.h>
332
+ #include <ATen/ops/lt_cpu_dispatch.h>
333
+ #include <ATen/ops/lu_unpack_cpu_dispatch.h>
334
+ #include <ATen/ops/masked_fill_cpu_dispatch.h>
335
+ #include <ATen/ops/masked_scatter_cpu_dispatch.h>
336
+ #include <ATen/ops/masked_select_cpu_dispatch.h>
337
+ #include <ATen/ops/max_cpu_dispatch.h>
338
+ #include <ATen/ops/max_pool2d_with_indices_cpu_dispatch.h>
339
+ #include <ATen/ops/max_pool2d_with_indices_backward_cpu_dispatch.h>
340
+ #include <ATen/ops/max_pool3d_with_indices_cpu_dispatch.h>
341
+ #include <ATen/ops/max_pool3d_with_indices_backward_cpu_dispatch.h>
342
+ #include <ATen/ops/max_unpool2d_cpu_dispatch.h>
343
+ #include <ATen/ops/max_unpool3d_cpu_dispatch.h>
344
+ #include <ATen/ops/maximum_cpu_dispatch.h>
345
+ #include <ATen/ops/mean_cpu_dispatch.h>
346
+ #include <ATen/ops/median_cpu_dispatch.h>
347
+ #include <ATen/ops/min_cpu_dispatch.h>
348
+ #include <ATen/ops/minimum_cpu_dispatch.h>
349
+ #include <ATen/ops/mish_cpu_dispatch.h>
350
+ #include <ATen/ops/mish_backward_cpu_dispatch.h>
351
+ #include <ATen/ops/mkldnn_rnn_layer_cpu_dispatch.h>
352
+ #include <ATen/ops/mkldnn_rnn_layer_backward_cpu_dispatch.h>
353
+ #include <ATen/ops/mm_cpu_dispatch.h>
354
+ #include <ATen/ops/mode_cpu_dispatch.h>
355
+ #include <ATen/ops/mse_loss_cpu_dispatch.h>
356
+ #include <ATen/ops/mse_loss_backward_cpu_dispatch.h>
357
+ #include <ATen/ops/mul_cpu_dispatch.h>
358
+ #include <ATen/ops/multi_margin_loss_cpu_dispatch.h>
359
+ #include <ATen/ops/multi_margin_loss_backward_cpu_dispatch.h>
360
+ #include <ATen/ops/multilabel_margin_loss_backward_cpu_dispatch.h>
361
+ #include <ATen/ops/multilabel_margin_loss_forward_cpu_dispatch.h>
362
+ #include <ATen/ops/multinomial_cpu_dispatch.h>
363
+ #include <ATen/ops/mvlgamma_cpu_dispatch.h>
364
+ #include <ATen/ops/nan_to_num_cpu_dispatch.h>
365
+ #include <ATen/ops/nanmedian_cpu_dispatch.h>
366
+ #include <ATen/ops/nansum_cpu_dispatch.h>
367
+ #include <ATen/ops/narrow_copy_cpu_dispatch.h>
368
+ #include <ATen/ops/native_batch_norm_cpu_dispatch.h>
369
+ #include <ATen/ops/native_batch_norm_backward_cpu_dispatch.h>
370
+ #include <ATen/ops/native_channel_shuffle_cpu_dispatch.h>
371
+ #include <ATen/ops/native_dropout_cpu_dispatch.h>
372
+ #include <ATen/ops/native_dropout_backward_cpu_dispatch.h>
373
+ #include <ATen/ops/native_group_norm_cpu_dispatch.h>
374
+ #include <ATen/ops/native_group_norm_backward_cpu_dispatch.h>
375
+ #include <ATen/ops/native_layer_norm_cpu_dispatch.h>
376
+ #include <ATen/ops/native_layer_norm_backward_cpu_dispatch.h>
377
+ #include <ATen/ops/ne_cpu_dispatch.h>
378
+ #include <ATen/ops/neg_cpu_dispatch.h>
379
+ #include <ATen/ops/nextafter_cpu_dispatch.h>
380
+ #include <ATen/ops/nll_loss2d_backward_cpu_dispatch.h>
381
+ #include <ATen/ops/nll_loss2d_forward_cpu_dispatch.h>
382
+ #include <ATen/ops/nll_loss_backward_cpu_dispatch.h>
383
+ #include <ATen/ops/nll_loss_forward_cpu_dispatch.h>
384
+ #include <ATen/ops/nonzero_cpu_dispatch.h>
385
+ #include <ATen/ops/nonzero_static_cpu_dispatch.h>
386
+ #include <ATen/ops/norm_cpu_dispatch.h>
387
+ #include <ATen/ops/normal_cpu_dispatch.h>
388
+ #include <ATen/ops/ormqr_cpu_dispatch.h>
389
+ #include <ATen/ops/pixel_shuffle_cpu_dispatch.h>
390
+ #include <ATen/ops/pixel_unshuffle_cpu_dispatch.h>
391
+ #include <ATen/ops/poisson_cpu_dispatch.h>
392
+ #include <ATen/ops/polar_cpu_dispatch.h>
393
+ #include <ATen/ops/polygamma_cpu_dispatch.h>
394
+ #include <ATen/ops/pow_cpu_dispatch.h>
395
+ #include <ATen/ops/prod_cpu_dispatch.h>
396
+ #include <ATen/ops/put_cpu_dispatch.h>
397
+ #include <ATen/ops/quantize_per_channel_cpu_dispatch.h>
398
+ #include <ATen/ops/quantize_per_tensor_cpu_dispatch.h>
399
+ #include <ATen/ops/quantize_per_tensor_dynamic_cpu_dispatch.h>
400
+ #include <ATen/ops/random_cpu_dispatch.h>
401
+ #include <ATen/ops/randperm_cpu_dispatch.h>
402
+ #include <ATen/ops/range_cpu_dispatch.h>
403
+ #include <ATen/ops/reciprocal_cpu_dispatch.h>
404
+ #include <ATen/ops/reflection_pad1d_cpu_dispatch.h>
405
+ #include <ATen/ops/reflection_pad1d_backward_cpu_dispatch.h>
406
+ #include <ATen/ops/reflection_pad2d_cpu_dispatch.h>
407
+ #include <ATen/ops/reflection_pad2d_backward_cpu_dispatch.h>
408
+ #include <ATen/ops/reflection_pad3d_cpu_dispatch.h>
409
+ #include <ATen/ops/reflection_pad3d_backward_cpu_dispatch.h>
410
+ #include <ATen/ops/relu_cpu_dispatch.h>
411
+ #include <ATen/ops/remainder_cpu_dispatch.h>
412
+ #include <ATen/ops/renorm_cpu_dispatch.h>
413
+ #include <ATen/ops/repeat_interleave_cpu_dispatch.h>
414
+ #include <ATen/ops/replication_pad1d_cpu_dispatch.h>
415
+ #include <ATen/ops/replication_pad1d_backward_cpu_dispatch.h>
416
+ #include <ATen/ops/replication_pad2d_cpu_dispatch.h>
417
+ #include <ATen/ops/replication_pad2d_backward_cpu_dispatch.h>
418
+ #include <ATen/ops/replication_pad3d_cpu_dispatch.h>
419
+ #include <ATen/ops/replication_pad3d_backward_cpu_dispatch.h>
420
+ #include <ATen/ops/resize_cpu_dispatch.h>
421
+ #include <ATen/ops/roll_cpu_dispatch.h>
422
+ #include <ATen/ops/round_cpu_dispatch.h>
423
+ #include <ATen/ops/rrelu_with_noise_cpu_dispatch.h>
424
+ #include <ATen/ops/rshift_cpu_dispatch.h>
425
+ #include <ATen/ops/rsqrt_cpu_dispatch.h>
426
+ #include <ATen/ops/rsub_cpu_dispatch.h>
427
+ #include <ATen/ops/scatter_cpu_dispatch.h>
428
+ #include <ATen/ops/scatter_add_cpu_dispatch.h>
429
+ #include <ATen/ops/scatter_reduce_cpu_dispatch.h>
430
+ #include <ATen/ops/searchsorted_cpu_dispatch.h>
431
+ #include <ATen/ops/segment_reduce_cpu_dispatch.h>
432
+ #include <ATen/ops/set_cpu_dispatch.h>
433
+ #include <ATen/ops/sgn_cpu_dispatch.h>
434
+ #include <ATen/ops/sigmoid_cpu_dispatch.h>
435
+ #include <ATen/ops/sigmoid_backward_cpu_dispatch.h>
436
+ #include <ATen/ops/sign_cpu_dispatch.h>
437
+ #include <ATen/ops/signbit_cpu_dispatch.h>
438
+ #include <ATen/ops/silu_cpu_dispatch.h>
439
+ #include <ATen/ops/silu_backward_cpu_dispatch.h>
440
+ #include <ATen/ops/sin_cpu_dispatch.h>
441
+ #include <ATen/ops/sinc_cpu_dispatch.h>
442
+ #include <ATen/ops/sinh_cpu_dispatch.h>
443
+ #include <ATen/ops/slow_conv3d_forward_cpu_dispatch.h>
444
+ #include <ATen/ops/slow_conv_dilated2d_cpu_dispatch.h>
445
+ #include <ATen/ops/slow_conv_dilated3d_cpu_dispatch.h>
446
+ #include <ATen/ops/slow_conv_transpose2d_cpu_dispatch.h>
447
+ #include <ATen/ops/slow_conv_transpose3d_cpu_dispatch.h>
448
+ #include <ATen/ops/smooth_l1_loss_cpu_dispatch.h>
449
+ #include <ATen/ops/smooth_l1_loss_backward_cpu_dispatch.h>
450
+ #include <ATen/ops/softplus_cpu_dispatch.h>
451
+ #include <ATen/ops/softplus_backward_cpu_dispatch.h>
452
+ #include <ATen/ops/softshrink_cpu_dispatch.h>
453
+ #include <ATen/ops/softshrink_backward_cpu_dispatch.h>
454
+ #include <ATen/ops/sort_cpu_dispatch.h>
455
+ #include <ATen/ops/special_airy_ai_cpu_dispatch.h>
456
+ #include <ATen/ops/special_bessel_j0_cpu_dispatch.h>
457
+ #include <ATen/ops/special_bessel_j1_cpu_dispatch.h>
458
+ #include <ATen/ops/special_bessel_y0_cpu_dispatch.h>
459
+ #include <ATen/ops/special_bessel_y1_cpu_dispatch.h>
460
+ #include <ATen/ops/special_chebyshev_polynomial_t_cpu_dispatch.h>
461
+ #include <ATen/ops/special_chebyshev_polynomial_u_cpu_dispatch.h>
462
+ #include <ATen/ops/special_chebyshev_polynomial_v_cpu_dispatch.h>
463
+ #include <ATen/ops/special_chebyshev_polynomial_w_cpu_dispatch.h>
464
+ #include <ATen/ops/special_entr_cpu_dispatch.h>
465
+ #include <ATen/ops/special_erfcx_cpu_dispatch.h>
466
+ #include <ATen/ops/special_hermite_polynomial_h_cpu_dispatch.h>
467
+ #include <ATen/ops/special_hermite_polynomial_he_cpu_dispatch.h>
468
+ #include <ATen/ops/special_i0e_cpu_dispatch.h>
469
+ #include <ATen/ops/special_i1_cpu_dispatch.h>
470
+ #include <ATen/ops/special_i1e_cpu_dispatch.h>
471
+ #include <ATen/ops/special_laguerre_polynomial_l_cpu_dispatch.h>
472
+ #include <ATen/ops/special_legendre_polynomial_p_cpu_dispatch.h>
473
+ #include <ATen/ops/special_log_ndtr_cpu_dispatch.h>
474
+ #include <ATen/ops/special_modified_bessel_i0_cpu_dispatch.h>
475
+ #include <ATen/ops/special_modified_bessel_i1_cpu_dispatch.h>
476
+ #include <ATen/ops/special_modified_bessel_k0_cpu_dispatch.h>
477
+ #include <ATen/ops/special_modified_bessel_k1_cpu_dispatch.h>
478
+ #include <ATen/ops/special_ndtri_cpu_dispatch.h>
479
+ #include <ATen/ops/special_scaled_modified_bessel_k0_cpu_dispatch.h>
480
+ #include <ATen/ops/special_scaled_modified_bessel_k1_cpu_dispatch.h>
481
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_t_cpu_dispatch.h>
482
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_u_cpu_dispatch.h>
483
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_v_cpu_dispatch.h>
484
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_w_cpu_dispatch.h>
485
+ #include <ATen/ops/special_spherical_bessel_j0_cpu_dispatch.h>
486
+ #include <ATen/ops/special_xlog1py_cpu_dispatch.h>
487
+ #include <ATen/ops/special_zeta_cpu_dispatch.h>
488
+ #include <ATen/ops/sqrt_cpu_dispatch.h>
489
+ #include <ATen/ops/sspaddmm_cpu_dispatch.h>
490
+ #include <ATen/ops/std_cpu_dispatch.h>
491
+ #include <ATen/ops/std_mean_cpu_dispatch.h>
492
+ #include <ATen/ops/sub_cpu_dispatch.h>
493
+ #include <ATen/ops/sum_cpu_dispatch.h>
494
+ #include <ATen/ops/take_cpu_dispatch.h>
495
+ #include <ATen/ops/tan_cpu_dispatch.h>
496
+ #include <ATen/ops/tanh_cpu_dispatch.h>
497
+ #include <ATen/ops/tanh_backward_cpu_dispatch.h>
498
+ #include <ATen/ops/threshold_cpu_dispatch.h>
499
+ #include <ATen/ops/threshold_backward_cpu_dispatch.h>
500
+ #include <ATen/ops/to_mkldnn_cpu_dispatch.h>
501
+ #include <ATen/ops/topk_cpu_dispatch.h>
502
+ #include <ATen/ops/trace_cpu_dispatch.h>
503
+ #include <ATen/ops/triangular_solve_cpu_dispatch.h>
504
+ #include <ATen/ops/tril_cpu_dispatch.h>
505
+ #include <ATen/ops/tril_indices_cpu_dispatch.h>
506
+ #include <ATen/ops/triu_cpu_dispatch.h>
507
+ #include <ATen/ops/triu_indices_cpu_dispatch.h>
508
+ #include <ATen/ops/trunc_cpu_dispatch.h>
509
+ #include <ATen/ops/unfold_cpu_dispatch.h>
510
+ #include <ATen/ops/unfold_backward_cpu_dispatch.h>
511
+ #include <ATen/ops/uniform_cpu_dispatch.h>
512
+ #include <ATen/ops/unique_consecutive_cpu_dispatch.h>
513
+ #include <ATen/ops/unique_dim_cpu_dispatch.h>
514
+ #include <ATen/ops/unique_dim_consecutive_cpu_dispatch.h>
515
+ #include <ATen/ops/upsample_bicubic2d_cpu_dispatch.h>
516
+ #include <ATen/ops/upsample_bicubic2d_backward_cpu_dispatch.h>
517
+ #include <ATen/ops/upsample_bilinear2d_cpu_dispatch.h>
518
+ #include <ATen/ops/upsample_bilinear2d_backward_cpu_dispatch.h>
519
+ #include <ATen/ops/upsample_linear1d_cpu_dispatch.h>
520
+ #include <ATen/ops/upsample_linear1d_backward_cpu_dispatch.h>
521
+ #include <ATen/ops/upsample_nearest1d_cpu_dispatch.h>
522
+ #include <ATen/ops/upsample_nearest1d_backward_cpu_dispatch.h>
523
+ #include <ATen/ops/upsample_nearest2d_cpu_dispatch.h>
524
+ #include <ATen/ops/upsample_nearest2d_backward_cpu_dispatch.h>
525
+ #include <ATen/ops/upsample_nearest3d_cpu_dispatch.h>
526
+ #include <ATen/ops/upsample_nearest3d_backward_cpu_dispatch.h>
527
+ #include <ATen/ops/upsample_trilinear3d_cpu_dispatch.h>
528
+ #include <ATen/ops/upsample_trilinear3d_backward_cpu_dispatch.h>
529
+ #include <ATen/ops/var_cpu_dispatch.h>
530
+ #include <ATen/ops/var_mean_cpu_dispatch.h>
531
+ #include <ATen/ops/vdot_cpu_dispatch.h>
532
+ #include <ATen/ops/view_cpu_dispatch.h>
533
+ #include <ATen/ops/view_as_complex_cpu_dispatch.h>
534
+ #include <ATen/ops/view_as_real_cpu_dispatch.h>
535
+ #include <ATen/ops/where_cpu_dispatch.h>
536
+ #include <ATen/ops/xlogy_cpu_dispatch.h>
537
+ #include <ATen/ops/zero_cpu_dispatch.h>
538
+
539
+
540
+
.venv/lib/python3.11/site-packages/torch/include/ATen/CUDAFunctions.h ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #include <ATen/core/TensorBody.h>
2
+
3
+ // TODO Undo all logic introduced for Note [Avoiding Include Cycles In Static Dispatch]
4
+ // Code introduced to avoid cyclic dependency in static dispatch is no longer
5
+ // needed as static dispatch logic is moved from TensorBody.h, which caused cycles in the first place,
6
+ // to Operators.cpp for supporting multiple backends with multiple kernels.
7
+ //
8
+ // Note [Avoiding Include Cycles In Static Dispatch]
9
+ // In order to avoid #include cycles in the static dispatch build, we've carefully split out
10
+ // the static function definition files into {DispatchKey}Functions.h and {DispatchKey}Functions_inl.h.
11
+ //
12
+ // Without this split, the include cycle looks like TensorBody.h -> CPUFunctions.h -> TensorBody.h.
13
+ // - TensorBody.h #includes CPUFunctions.h in the static dispatch build, because the tensor methods
14
+ // all need to call into the fastpath C++ API defined in CPUFunctions.h. The methods are also all
15
+ // directly inlined into TensorBody.h.
16
+ // - CPUFunctions.h #includes TensorBody.h because it contains function declarations for the entire C++ API,
17
+ // which include functions that have defaultable std::optional<Tensor> arguments.
18
+ // That requires knowing the full Tensor class definition.
19
+ //
20
+ // We break the cycle by doing the following:
21
+ // - Split out CPUFunction.h into two files: CPUFunctions.h and CPUFunctions_inl.h
22
+ // - CPUFunction.h is a dummy file that just includes the Tensor class and includes CPUFunctions_inl.,
23
+ // - CPUFunctions_inl.h includes everything else
24
+ // - (only in the static dispatch build) TensorBody.h makes sure to finish defining the Tensor class,
25
+ // and then it includes CPUFunctions_inl.h.
26
+ // - All other files that want the cpu fastpath functions can include CPUFunctions.h directly.
27
+ // - This also means that static dispatch build, CPUFunctions.h only needs to
28
+ // #include TensorBody.h, and it will automatically bring in CPUFunctions_inl.h.
29
+ #include <ATen/CUDAFunctions_inl.h>
.venv/lib/python3.11/site-packages/torch/include/ATen/CUDAFunctions_inl.h ADDED
@@ -0,0 +1,623 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+ // @generated by torchgen/gen.py from DispatchKeyFunctions_inl.h
3
+
4
+ // NB: The implementing C++ file is RegisterDispatchKey.cpp
5
+
6
+ // The only #includes we need are for custom classes that have defaults in the C++ API
7
+ #include <c10/core/MemoryFormat.h>
8
+ #include <c10/core/Scalar.h>
9
+ #include <ATen/core/Reduction.h>
10
+
11
+ #if defined(AT_PER_OPERATOR_HEADERS) && defined(TORCH_ASSERT_ONLY_METHOD_OPERATORS)
12
+ #error This change adds a dependency on all pytorch operators, meaning the \
13
+ file will need to be re-compiled every time an operator is changed or added. \
14
+ Consider including a specific operator from \
15
+ <ATen/ops/{my_operator}_cuda_dispatch.h>. \
16
+ See NOTE [TORCH_ASSERT_ONLY_METHOD_OPERATORS].
17
+ #endif
18
+
19
+ #include <ATen/ops/_adaptive_avg_pool2d_cuda_dispatch.h>
20
+ #include <ATen/ops/_adaptive_avg_pool2d_backward_cuda_dispatch.h>
21
+ #include <ATen/ops/_adaptive_avg_pool3d_cuda_dispatch.h>
22
+ #include <ATen/ops/_adaptive_avg_pool3d_backward_cuda_dispatch.h>
23
+ #include <ATen/ops/_addmm_activation_cuda_dispatch.h>
24
+ #include <ATen/ops/_aminmax_cuda_dispatch.h>
25
+ #include <ATen/ops/_amp_foreach_non_finite_check_and_unscale_cuda_dispatch.h>
26
+ #include <ATen/ops/_amp_update_scale_cuda_dispatch.h>
27
+ #include <ATen/ops/_assert_async_cuda_dispatch.h>
28
+ #include <ATen/ops/_batch_norm_with_update_cuda_dispatch.h>
29
+ #include <ATen/ops/_cdist_backward_cuda_dispatch.h>
30
+ #include <ATen/ops/_cdist_forward_cuda_dispatch.h>
31
+ #include <ATen/ops/_cholesky_solve_helper_cuda_dispatch.h>
32
+ #include <ATen/ops/_chunk_cat_cuda_dispatch.h>
33
+ #include <ATen/ops/_compute_linear_combination_cuda_dispatch.h>
34
+ #include <ATen/ops/_conv_depthwise2d_cuda_dispatch.h>
35
+ #include <ATen/ops/_convert_indices_from_coo_to_csr_cuda_dispatch.h>
36
+ #include <ATen/ops/_convert_indices_from_csr_to_coo_cuda_dispatch.h>
37
+ #include <ATen/ops/_convert_weight_to_int4pack_cuda_dispatch.h>
38
+ #include <ATen/ops/_cslt_compress_cuda_dispatch.h>
39
+ #include <ATen/ops/_cslt_sparse_mm_cuda_dispatch.h>
40
+ #include <ATen/ops/_cslt_sparse_mm_search_cuda_dispatch.h>
41
+ #include <ATen/ops/_ctc_loss_cuda_dispatch.h>
42
+ #include <ATen/ops/_ctc_loss_backward_cuda_dispatch.h>
43
+ #include <ATen/ops/_cudnn_ctc_loss_cuda_dispatch.h>
44
+ #include <ATen/ops/_cudnn_init_dropout_state_cuda_dispatch.h>
45
+ #include <ATen/ops/_cudnn_rnn_cuda_dispatch.h>
46
+ #include <ATen/ops/_cudnn_rnn_backward_cuda_dispatch.h>
47
+ #include <ATen/ops/_cudnn_rnn_flatten_weight_cuda_dispatch.h>
48
+ #include <ATen/ops/_cummax_helper_cuda_dispatch.h>
49
+ #include <ATen/ops/_cummin_helper_cuda_dispatch.h>
50
+ #include <ATen/ops/_dirichlet_grad_cuda_dispatch.h>
51
+ #include <ATen/ops/_efficient_attention_backward_cuda_dispatch.h>
52
+ #include <ATen/ops/_efficient_attention_forward_cuda_dispatch.h>
53
+ #include <ATen/ops/_efficientzerotensor_cuda_dispatch.h>
54
+ #include <ATen/ops/_embedding_bag_cuda_dispatch.h>
55
+ #include <ATen/ops/_embedding_bag_backward_cuda_dispatch.h>
56
+ #include <ATen/ops/_embedding_bag_dense_backward_cuda_dispatch.h>
57
+ #include <ATen/ops/_embedding_bag_forward_only_cuda_dispatch.h>
58
+ #include <ATen/ops/_embedding_bag_per_sample_weights_backward_cuda_dispatch.h>
59
+ #include <ATen/ops/_fake_quantize_learnable_per_channel_affine_cuda_dispatch.h>
60
+ #include <ATen/ops/_fake_quantize_learnable_per_channel_affine_backward_cuda_dispatch.h>
61
+ #include <ATen/ops/_fake_quantize_learnable_per_tensor_affine_cuda_dispatch.h>
62
+ #include <ATen/ops/_fake_quantize_learnable_per_tensor_affine_backward_cuda_dispatch.h>
63
+ #include <ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams_cuda_dispatch.h>
64
+ #include <ATen/ops/_fft_c2c_cuda_dispatch.h>
65
+ #include <ATen/ops/_fft_c2r_cuda_dispatch.h>
66
+ #include <ATen/ops/_fft_r2c_cuda_dispatch.h>
67
+ #include <ATen/ops/_fill_mem_eff_dropout_mask_cuda_dispatch.h>
68
+ #include <ATen/ops/_flash_attention_backward_cuda_dispatch.h>
69
+ #include <ATen/ops/_flash_attention_forward_cuda_dispatch.h>
70
+ #include <ATen/ops/_foreach_abs_cuda_dispatch.h>
71
+ #include <ATen/ops/_foreach_acos_cuda_dispatch.h>
72
+ #include <ATen/ops/_foreach_add_cuda_dispatch.h>
73
+ #include <ATen/ops/_foreach_addcdiv_cuda_dispatch.h>
74
+ #include <ATen/ops/_foreach_addcmul_cuda_dispatch.h>
75
+ #include <ATen/ops/_foreach_asin_cuda_dispatch.h>
76
+ #include <ATen/ops/_foreach_atan_cuda_dispatch.h>
77
+ #include <ATen/ops/_foreach_ceil_cuda_dispatch.h>
78
+ #include <ATen/ops/_foreach_clamp_max_cuda_dispatch.h>
79
+ #include <ATen/ops/_foreach_clamp_min_cuda_dispatch.h>
80
+ #include <ATen/ops/_foreach_copy_cuda_dispatch.h>
81
+ #include <ATen/ops/_foreach_cos_cuda_dispatch.h>
82
+ #include <ATen/ops/_foreach_cosh_cuda_dispatch.h>
83
+ #include <ATen/ops/_foreach_div_cuda_dispatch.h>
84
+ #include <ATen/ops/_foreach_erf_cuda_dispatch.h>
85
+ #include <ATen/ops/_foreach_erfc_cuda_dispatch.h>
86
+ #include <ATen/ops/_foreach_exp_cuda_dispatch.h>
87
+ #include <ATen/ops/_foreach_expm1_cuda_dispatch.h>
88
+ #include <ATen/ops/_foreach_floor_cuda_dispatch.h>
89
+ #include <ATen/ops/_foreach_frac_cuda_dispatch.h>
90
+ #include <ATen/ops/_foreach_lerp_cuda_dispatch.h>
91
+ #include <ATen/ops/_foreach_lgamma_cuda_dispatch.h>
92
+ #include <ATen/ops/_foreach_log_cuda_dispatch.h>
93
+ #include <ATen/ops/_foreach_log10_cuda_dispatch.h>
94
+ #include <ATen/ops/_foreach_log1p_cuda_dispatch.h>
95
+ #include <ATen/ops/_foreach_log2_cuda_dispatch.h>
96
+ #include <ATen/ops/_foreach_max_cuda_dispatch.h>
97
+ #include <ATen/ops/_foreach_maximum_cuda_dispatch.h>
98
+ #include <ATen/ops/_foreach_minimum_cuda_dispatch.h>
99
+ #include <ATen/ops/_foreach_mul_cuda_dispatch.h>
100
+ #include <ATen/ops/_foreach_neg_cuda_dispatch.h>
101
+ #include <ATen/ops/_foreach_norm_cuda_dispatch.h>
102
+ #include <ATen/ops/_foreach_pow_cuda_dispatch.h>
103
+ #include <ATen/ops/_foreach_reciprocal_cuda_dispatch.h>
104
+ #include <ATen/ops/_foreach_round_cuda_dispatch.h>
105
+ #include <ATen/ops/_foreach_sigmoid_cuda_dispatch.h>
106
+ #include <ATen/ops/_foreach_sign_cuda_dispatch.h>
107
+ #include <ATen/ops/_foreach_sin_cuda_dispatch.h>
108
+ #include <ATen/ops/_foreach_sinh_cuda_dispatch.h>
109
+ #include <ATen/ops/_foreach_sqrt_cuda_dispatch.h>
110
+ #include <ATen/ops/_foreach_sub_cuda_dispatch.h>
111
+ #include <ATen/ops/_foreach_tan_cuda_dispatch.h>
112
+ #include <ATen/ops/_foreach_tanh_cuda_dispatch.h>
113
+ #include <ATen/ops/_foreach_trunc_cuda_dispatch.h>
114
+ #include <ATen/ops/_foreach_zero_cuda_dispatch.h>
115
+ #include <ATen/ops/_fused_adam_cuda_dispatch.h>
116
+ #include <ATen/ops/_fused_adamw_cuda_dispatch.h>
117
+ #include <ATen/ops/_fused_dropout_cuda_dispatch.h>
118
+ #include <ATen/ops/_fused_moving_avg_obs_fq_helper_cuda_dispatch.h>
119
+ #include <ATen/ops/_fused_sdp_choice_cuda_dispatch.h>
120
+ #include <ATen/ops/_fused_sgd_cuda_dispatch.h>
121
+ #include <ATen/ops/_index_put_impl_cuda_dispatch.h>
122
+ #include <ATen/ops/_int_mm_cuda_dispatch.h>
123
+ #include <ATen/ops/_jagged_to_padded_dense_forward_cuda_dispatch.h>
124
+ #include <ATen/ops/_linalg_det_cuda_dispatch.h>
125
+ #include <ATen/ops/_linalg_eigh_cuda_dispatch.h>
126
+ #include <ATen/ops/_linalg_eigvals_cuda_dispatch.h>
127
+ #include <ATen/ops/_linalg_slogdet_cuda_dispatch.h>
128
+ #include <ATen/ops/_linalg_solve_ex_cuda_dispatch.h>
129
+ #include <ATen/ops/_linalg_svd_cuda_dispatch.h>
130
+ #include <ATen/ops/_local_scalar_dense_cuda_dispatch.h>
131
+ #include <ATen/ops/_log_softmax_cuda_dispatch.h>
132
+ #include <ATen/ops/_log_softmax_backward_data_cuda_dispatch.h>
133
+ #include <ATen/ops/_logcumsumexp_cuda_dispatch.h>
134
+ #include <ATen/ops/_make_per_channel_quantized_tensor_cuda_dispatch.h>
135
+ #include <ATen/ops/_make_per_tensor_quantized_tensor_cuda_dispatch.h>
136
+ #include <ATen/ops/_masked_scale_cuda_dispatch.h>
137
+ #include <ATen/ops/_masked_softmax_cuda_dispatch.h>
138
+ #include <ATen/ops/_masked_softmax_backward_cuda_dispatch.h>
139
+ #include <ATen/ops/_mixed_dtypes_linear_cuda_dispatch.h>
140
+ #include <ATen/ops/_native_batch_norm_legit_cuda_dispatch.h>
141
+ #include <ATen/ops/_native_multi_head_attention_cuda_dispatch.h>
142
+ #include <ATen/ops/_nested_compute_contiguous_strides_offsets_cuda_dispatch.h>
143
+ #include <ATen/ops/_nested_from_padded_cuda_dispatch.h>
144
+ #include <ATen/ops/_nested_tensor_from_mask_cuda_dispatch.h>
145
+ #include <ATen/ops/_nested_tensor_from_mask_left_aligned_cuda_dispatch.h>
146
+ #include <ATen/ops/_nested_view_from_buffer_cuda_dispatch.h>
147
+ #include <ATen/ops/_padded_dense_to_jagged_forward_cuda_dispatch.h>
148
+ #include <ATen/ops/_pdist_backward_cuda_dispatch.h>
149
+ #include <ATen/ops/_pdist_forward_cuda_dispatch.h>
150
+ #include <ATen/ops/_prelu_kernel_cuda_dispatch.h>
151
+ #include <ATen/ops/_prelu_kernel_backward_cuda_dispatch.h>
152
+ #include <ATen/ops/_reshape_alias_cuda_dispatch.h>
153
+ #include <ATen/ops/_sample_dirichlet_cuda_dispatch.h>
154
+ #include <ATen/ops/_scaled_dot_product_cudnn_attention_cuda_dispatch.h>
155
+ #include <ATen/ops/_scaled_dot_product_cudnn_attention_backward_cuda_dispatch.h>
156
+ #include <ATen/ops/_scaled_dot_product_efficient_attention_cuda_dispatch.h>
157
+ #include <ATen/ops/_scaled_dot_product_efficient_attention_backward_cuda_dispatch.h>
158
+ #include <ATen/ops/_scaled_dot_product_flash_attention_cuda_dispatch.h>
159
+ #include <ATen/ops/_scaled_dot_product_flash_attention_backward_cuda_dispatch.h>
160
+ #include <ATen/ops/_scaled_mm_cuda_dispatch.h>
161
+ #include <ATen/ops/_segment_reduce_backward_cuda_dispatch.h>
162
+ #include <ATen/ops/_slow_conv2d_backward_cuda_dispatch.h>
163
+ #include <ATen/ops/_slow_conv2d_forward_cuda_dispatch.h>
164
+ #include <ATen/ops/_softmax_cuda_dispatch.h>
165
+ #include <ATen/ops/_softmax_backward_data_cuda_dispatch.h>
166
+ #include <ATen/ops/_sparse_semi_structured_addmm_cuda_dispatch.h>
167
+ #include <ATen/ops/_sparse_semi_structured_apply_cuda_dispatch.h>
168
+ #include <ATen/ops/_sparse_semi_structured_apply_dense_cuda_dispatch.h>
169
+ #include <ATen/ops/_sparse_semi_structured_linear_cuda_dispatch.h>
170
+ #include <ATen/ops/_sparse_semi_structured_mm_cuda_dispatch.h>
171
+ #include <ATen/ops/_sparse_semi_structured_tile_cuda_dispatch.h>
172
+ #include <ATen/ops/_standard_gamma_cuda_dispatch.h>
173
+ #include <ATen/ops/_standard_gamma_grad_cuda_dispatch.h>
174
+ #include <ATen/ops/_thnn_fused_gru_cell_cuda_dispatch.h>
175
+ #include <ATen/ops/_thnn_fused_gru_cell_backward_cuda_dispatch.h>
176
+ #include <ATen/ops/_thnn_fused_lstm_cell_cuda_dispatch.h>
177
+ #include <ATen/ops/_thnn_fused_lstm_cell_backward_impl_cuda_dispatch.h>
178
+ #include <ATen/ops/_to_sparse_cuda_dispatch.h>
179
+ #include <ATen/ops/_to_sparse_bsc_cuda_dispatch.h>
180
+ #include <ATen/ops/_to_sparse_bsr_cuda_dispatch.h>
181
+ #include <ATen/ops/_to_sparse_csc_cuda_dispatch.h>
182
+ #include <ATen/ops/_to_sparse_csr_cuda_dispatch.h>
183
+ #include <ATen/ops/_to_sparse_semi_structured_cuda_dispatch.h>
184
+ #include <ATen/ops/_transform_bias_rescale_qkv_cuda_dispatch.h>
185
+ #include <ATen/ops/_transformer_encoder_layer_fwd_cuda_dispatch.h>
186
+ #include <ATen/ops/_triton_multi_head_attention_cuda_dispatch.h>
187
+ #include <ATen/ops/_triton_scaled_dot_attention_cuda_dispatch.h>
188
+ #include <ATen/ops/_unique_cuda_dispatch.h>
189
+ #include <ATen/ops/_unique2_cuda_dispatch.h>
190
+ #include <ATen/ops/_upsample_bicubic2d_aa_cuda_dispatch.h>
191
+ #include <ATen/ops/_upsample_bicubic2d_aa_backward_cuda_dispatch.h>
192
+ #include <ATen/ops/_upsample_bilinear2d_aa_cuda_dispatch.h>
193
+ #include <ATen/ops/_upsample_bilinear2d_aa_backward_cuda_dispatch.h>
194
+ #include <ATen/ops/_upsample_nearest_exact1d_cuda_dispatch.h>
195
+ #include <ATen/ops/_upsample_nearest_exact1d_backward_cuda_dispatch.h>
196
+ #include <ATen/ops/_upsample_nearest_exact2d_cuda_dispatch.h>
197
+ #include <ATen/ops/_upsample_nearest_exact2d_backward_cuda_dispatch.h>
198
+ #include <ATen/ops/_upsample_nearest_exact3d_cuda_dispatch.h>
199
+ #include <ATen/ops/_upsample_nearest_exact3d_backward_cuda_dispatch.h>
200
+ #include <ATen/ops/_use_cudnn_ctc_loss_cuda_dispatch.h>
201
+ #include <ATen/ops/_validate_compressed_sparse_indices_cuda_dispatch.h>
202
+ #include <ATen/ops/_weight_int4pack_mm_cuda_dispatch.h>
203
+ #include <ATen/ops/_weight_norm_interface_cuda_dispatch.h>
204
+ #include <ATen/ops/_weight_norm_interface_backward_cuda_dispatch.h>
205
+ #include <ATen/ops/abs_cuda_dispatch.h>
206
+ #include <ATen/ops/acos_cuda_dispatch.h>
207
+ #include <ATen/ops/acosh_cuda_dispatch.h>
208
+ #include <ATen/ops/adaptive_avg_pool2d_cuda_dispatch.h>
209
+ #include <ATen/ops/adaptive_avg_pool3d_cuda_dispatch.h>
210
+ #include <ATen/ops/adaptive_avg_pool3d_backward_cuda_dispatch.h>
211
+ #include <ATen/ops/adaptive_max_pool2d_cuda_dispatch.h>
212
+ #include <ATen/ops/adaptive_max_pool2d_backward_cuda_dispatch.h>
213
+ #include <ATen/ops/adaptive_max_pool3d_cuda_dispatch.h>
214
+ #include <ATen/ops/adaptive_max_pool3d_backward_cuda_dispatch.h>
215
+ #include <ATen/ops/add_cuda_dispatch.h>
216
+ #include <ATen/ops/addbmm_cuda_dispatch.h>
217
+ #include <ATen/ops/addcdiv_cuda_dispatch.h>
218
+ #include <ATen/ops/addcmul_cuda_dispatch.h>
219
+ #include <ATen/ops/addmm_cuda_dispatch.h>
220
+ #include <ATen/ops/addmv_cuda_dispatch.h>
221
+ #include <ATen/ops/addr_cuda_dispatch.h>
222
+ #include <ATen/ops/all_cuda_dispatch.h>
223
+ #include <ATen/ops/amax_cuda_dispatch.h>
224
+ #include <ATen/ops/amin_cuda_dispatch.h>
225
+ #include <ATen/ops/aminmax_cuda_dispatch.h>
226
+ #include <ATen/ops/angle_cuda_dispatch.h>
227
+ #include <ATen/ops/any_cuda_dispatch.h>
228
+ #include <ATen/ops/arange_cuda_dispatch.h>
229
+ #include <ATen/ops/argmax_cuda_dispatch.h>
230
+ #include <ATen/ops/argmin_cuda_dispatch.h>
231
+ #include <ATen/ops/as_strided_cuda_dispatch.h>
232
+ #include <ATen/ops/asin_cuda_dispatch.h>
233
+ #include <ATen/ops/asinh_cuda_dispatch.h>
234
+ #include <ATen/ops/atan_cuda_dispatch.h>
235
+ #include <ATen/ops/atan2_cuda_dispatch.h>
236
+ #include <ATen/ops/atanh_cuda_dispatch.h>
237
+ #include <ATen/ops/avg_pool2d_cuda_dispatch.h>
238
+ #include <ATen/ops/avg_pool2d_backward_cuda_dispatch.h>
239
+ #include <ATen/ops/avg_pool3d_cuda_dispatch.h>
240
+ #include <ATen/ops/avg_pool3d_backward_cuda_dispatch.h>
241
+ #include <ATen/ops/baddbmm_cuda_dispatch.h>
242
+ #include <ATen/ops/batch_norm_backward_cuda_dispatch.h>
243
+ #include <ATen/ops/batch_norm_backward_elemt_cuda_dispatch.h>
244
+ #include <ATen/ops/batch_norm_backward_reduce_cuda_dispatch.h>
245
+ #include <ATen/ops/batch_norm_elemt_cuda_dispatch.h>
246
+ #include <ATen/ops/batch_norm_gather_stats_cuda_dispatch.h>
247
+ #include <ATen/ops/batch_norm_gather_stats_with_counts_cuda_dispatch.h>
248
+ #include <ATen/ops/batch_norm_stats_cuda_dispatch.h>
249
+ #include <ATen/ops/batch_norm_update_stats_cuda_dispatch.h>
250
+ #include <ATen/ops/bernoulli_cuda_dispatch.h>
251
+ #include <ATen/ops/binary_cross_entropy_cuda_dispatch.h>
252
+ #include <ATen/ops/binary_cross_entropy_backward_cuda_dispatch.h>
253
+ #include <ATen/ops/bincount_cuda_dispatch.h>
254
+ #include <ATen/ops/binomial_cuda_dispatch.h>
255
+ #include <ATen/ops/bitwise_and_cuda_dispatch.h>
256
+ #include <ATen/ops/bitwise_left_shift_cuda_dispatch.h>
257
+ #include <ATen/ops/bitwise_not_cuda_dispatch.h>
258
+ #include <ATen/ops/bitwise_or_cuda_dispatch.h>
259
+ #include <ATen/ops/bitwise_right_shift_cuda_dispatch.h>
260
+ #include <ATen/ops/bitwise_xor_cuda_dispatch.h>
261
+ #include <ATen/ops/bmm_cuda_dispatch.h>
262
+ #include <ATen/ops/bucketize_cuda_dispatch.h>
263
+ #include <ATen/ops/cat_cuda_dispatch.h>
264
+ #include <ATen/ops/cauchy_cuda_dispatch.h>
265
+ #include <ATen/ops/ceil_cuda_dispatch.h>
266
+ #include <ATen/ops/channel_shuffle_cuda_dispatch.h>
267
+ #include <ATen/ops/cholesky_cuda_dispatch.h>
268
+ #include <ATen/ops/cholesky_inverse_cuda_dispatch.h>
269
+ #include <ATen/ops/clamp_cuda_dispatch.h>
270
+ #include <ATen/ops/clamp_max_cuda_dispatch.h>
271
+ #include <ATen/ops/clamp_min_cuda_dispatch.h>
272
+ #include <ATen/ops/col2im_cuda_dispatch.h>
273
+ #include <ATen/ops/complex_cuda_dispatch.h>
274
+ #include <ATen/ops/conj_physical_cuda_dispatch.h>
275
+ #include <ATen/ops/conv_depthwise3d_cuda_dispatch.h>
276
+ #include <ATen/ops/convolution_backward_cuda_dispatch.h>
277
+ #include <ATen/ops/copysign_cuda_dispatch.h>
278
+ #include <ATen/ops/cos_cuda_dispatch.h>
279
+ #include <ATen/ops/cosh_cuda_dispatch.h>
280
+ #include <ATen/ops/count_nonzero_cuda_dispatch.h>
281
+ #include <ATen/ops/cudnn_affine_grid_generator_cuda_dispatch.h>
282
+ #include <ATen/ops/cudnn_affine_grid_generator_backward_cuda_dispatch.h>
283
+ #include <ATen/ops/cudnn_batch_norm_cuda_dispatch.h>
284
+ #include <ATen/ops/cudnn_batch_norm_backward_cuda_dispatch.h>
285
+ #include <ATen/ops/cudnn_convolution_cuda_dispatch.h>
286
+ #include <ATen/ops/cudnn_convolution_add_relu_cuda_dispatch.h>
287
+ #include <ATen/ops/cudnn_convolution_relu_cuda_dispatch.h>
288
+ #include <ATen/ops/cudnn_convolution_transpose_cuda_dispatch.h>
289
+ #include <ATen/ops/cudnn_grid_sampler_cuda_dispatch.h>
290
+ #include <ATen/ops/cudnn_grid_sampler_backward_cuda_dispatch.h>
291
+ #include <ATen/ops/cumprod_cuda_dispatch.h>
292
+ #include <ATen/ops/cumsum_cuda_dispatch.h>
293
+ #include <ATen/ops/dequantize_cuda_dispatch.h>
294
+ #include <ATen/ops/digamma_cuda_dispatch.h>
295
+ #include <ATen/ops/div_cuda_dispatch.h>
296
+ #include <ATen/ops/dot_cuda_dispatch.h>
297
+ #include <ATen/ops/elu_cuda_dispatch.h>
298
+ #include <ATen/ops/elu_backward_cuda_dispatch.h>
299
+ #include <ATen/ops/embedding_dense_backward_cuda_dispatch.h>
300
+ #include <ATen/ops/embedding_renorm_cuda_dispatch.h>
301
+ #include <ATen/ops/empty_cuda_dispatch.h>
302
+ #include <ATen/ops/empty_strided_cuda_dispatch.h>
303
+ #include <ATen/ops/eq_cuda_dispatch.h>
304
+ #include <ATen/ops/equal_cuda_dispatch.h>
305
+ #include <ATen/ops/erf_cuda_dispatch.h>
306
+ #include <ATen/ops/erfc_cuda_dispatch.h>
307
+ #include <ATen/ops/erfinv_cuda_dispatch.h>
308
+ #include <ATen/ops/exp_cuda_dispatch.h>
309
+ #include <ATen/ops/exp2_cuda_dispatch.h>
310
+ #include <ATen/ops/expm1_cuda_dispatch.h>
311
+ #include <ATen/ops/exponential_cuda_dispatch.h>
312
+ #include <ATen/ops/eye_cuda_dispatch.h>
313
+ #include <ATen/ops/fake_quantize_per_channel_affine_cachemask_cuda_dispatch.h>
314
+ #include <ATen/ops/fake_quantize_per_tensor_affine_cachemask_cuda_dispatch.h>
315
+ #include <ATen/ops/fill_cuda_dispatch.h>
316
+ #include <ATen/ops/flip_cuda_dispatch.h>
317
+ #include <ATen/ops/floor_cuda_dispatch.h>
318
+ #include <ATen/ops/floor_divide_cuda_dispatch.h>
319
+ #include <ATen/ops/fmax_cuda_dispatch.h>
320
+ #include <ATen/ops/fmin_cuda_dispatch.h>
321
+ #include <ATen/ops/fmod_cuda_dispatch.h>
322
+ #include <ATen/ops/frac_cuda_dispatch.h>
323
+ #include <ATen/ops/fractional_max_pool2d_cuda_dispatch.h>
324
+ #include <ATen/ops/fractional_max_pool2d_backward_cuda_dispatch.h>
325
+ #include <ATen/ops/fractional_max_pool3d_cuda_dispatch.h>
326
+ #include <ATen/ops/fractional_max_pool3d_backward_cuda_dispatch.h>
327
+ #include <ATen/ops/frexp_cuda_dispatch.h>
328
+ #include <ATen/ops/gather_cuda_dispatch.h>
329
+ #include <ATen/ops/gcd_cuda_dispatch.h>
330
+ #include <ATen/ops/ge_cuda_dispatch.h>
331
+ #include <ATen/ops/gelu_cuda_dispatch.h>
332
+ #include <ATen/ops/gelu_backward_cuda_dispatch.h>
333
+ #include <ATen/ops/geometric_cuda_dispatch.h>
334
+ #include <ATen/ops/geqrf_cuda_dispatch.h>
335
+ #include <ATen/ops/glu_cuda_dispatch.h>
336
+ #include <ATen/ops/glu_backward_cuda_dispatch.h>
337
+ #include <ATen/ops/glu_backward_jvp_cuda_dispatch.h>
338
+ #include <ATen/ops/glu_jvp_cuda_dispatch.h>
339
+ #include <ATen/ops/grid_sampler_2d_cuda_dispatch.h>
340
+ #include <ATen/ops/grid_sampler_2d_backward_cuda_dispatch.h>
341
+ #include <ATen/ops/grid_sampler_3d_cuda_dispatch.h>
342
+ #include <ATen/ops/grid_sampler_3d_backward_cuda_dispatch.h>
343
+ #include <ATen/ops/gt_cuda_dispatch.h>
344
+ #include <ATen/ops/hardshrink_cuda_dispatch.h>
345
+ #include <ATen/ops/hardshrink_backward_cuda_dispatch.h>
346
+ #include <ATen/ops/hardsigmoid_cuda_dispatch.h>
347
+ #include <ATen/ops/hardsigmoid_backward_cuda_dispatch.h>
348
+ #include <ATen/ops/hardswish_cuda_dispatch.h>
349
+ #include <ATen/ops/hardswish_backward_cuda_dispatch.h>
350
+ #include <ATen/ops/hardtanh_cuda_dispatch.h>
351
+ #include <ATen/ops/hardtanh_backward_cuda_dispatch.h>
352
+ #include <ATen/ops/heaviside_cuda_dispatch.h>
353
+ #include <ATen/ops/histc_cuda_dispatch.h>
354
+ #include <ATen/ops/huber_loss_cuda_dispatch.h>
355
+ #include <ATen/ops/huber_loss_backward_cuda_dispatch.h>
356
+ #include <ATen/ops/hypot_cuda_dispatch.h>
357
+ #include <ATen/ops/i0_cuda_dispatch.h>
358
+ #include <ATen/ops/igamma_cuda_dispatch.h>
359
+ #include <ATen/ops/igammac_cuda_dispatch.h>
360
+ #include <ATen/ops/im2col_cuda_dispatch.h>
361
+ #include <ATen/ops/index_cuda_dispatch.h>
362
+ #include <ATen/ops/index_add_cuda_dispatch.h>
363
+ #include <ATen/ops/index_copy_cuda_dispatch.h>
364
+ #include <ATen/ops/index_fill_cuda_dispatch.h>
365
+ #include <ATen/ops/index_reduce_cuda_dispatch.h>
366
+ #include <ATen/ops/index_select_cuda_dispatch.h>
367
+ #include <ATen/ops/is_set_to_cuda_dispatch.h>
368
+ #include <ATen/ops/isin_cuda_dispatch.h>
369
+ #include <ATen/ops/isnan_cuda_dispatch.h>
370
+ #include <ATen/ops/isneginf_cuda_dispatch.h>
371
+ #include <ATen/ops/isposinf_cuda_dispatch.h>
372
+ #include <ATen/ops/kthvalue_cuda_dispatch.h>
373
+ #include <ATen/ops/lcm_cuda_dispatch.h>
374
+ #include <ATen/ops/le_cuda_dispatch.h>
375
+ #include <ATen/ops/leaky_relu_cuda_dispatch.h>
376
+ #include <ATen/ops/leaky_relu_backward_cuda_dispatch.h>
377
+ #include <ATen/ops/lerp_cuda_dispatch.h>
378
+ #include <ATen/ops/lgamma_cuda_dispatch.h>
379
+ #include <ATen/ops/linalg_cholesky_ex_cuda_dispatch.h>
380
+ #include <ATen/ops/linalg_cross_cuda_dispatch.h>
381
+ #include <ATen/ops/linalg_eig_cuda_dispatch.h>
382
+ #include <ATen/ops/linalg_eigvals_cuda_dispatch.h>
383
+ #include <ATen/ops/linalg_householder_product_cuda_dispatch.h>
384
+ #include <ATen/ops/linalg_inv_ex_cuda_dispatch.h>
385
+ #include <ATen/ops/linalg_ldl_factor_ex_cuda_dispatch.h>
386
+ #include <ATen/ops/linalg_ldl_solve_cuda_dispatch.h>
387
+ #include <ATen/ops/linalg_lstsq_cuda_dispatch.h>
388
+ #include <ATen/ops/linalg_lu_cuda_dispatch.h>
389
+ #include <ATen/ops/linalg_lu_factor_ex_cuda_dispatch.h>
390
+ #include <ATen/ops/linalg_lu_solve_cuda_dispatch.h>
391
+ #include <ATen/ops/linalg_matrix_exp_cuda_dispatch.h>
392
+ #include <ATen/ops/linalg_qr_cuda_dispatch.h>
393
+ #include <ATen/ops/linalg_solve_triangular_cuda_dispatch.h>
394
+ #include <ATen/ops/linalg_vector_norm_cuda_dispatch.h>
395
+ #include <ATen/ops/linspace_cuda_dispatch.h>
396
+ #include <ATen/ops/log_cuda_dispatch.h>
397
+ #include <ATen/ops/log10_cuda_dispatch.h>
398
+ #include <ATen/ops/log1p_cuda_dispatch.h>
399
+ #include <ATen/ops/log2_cuda_dispatch.h>
400
+ #include <ATen/ops/log_normal_cuda_dispatch.h>
401
+ #include <ATen/ops/log_sigmoid_backward_cuda_dispatch.h>
402
+ #include <ATen/ops/log_sigmoid_forward_cuda_dispatch.h>
403
+ #include <ATen/ops/logaddexp_cuda_dispatch.h>
404
+ #include <ATen/ops/logaddexp2_cuda_dispatch.h>
405
+ #include <ATen/ops/logical_and_cuda_dispatch.h>
406
+ #include <ATen/ops/logical_not_cuda_dispatch.h>
407
+ #include <ATen/ops/logical_or_cuda_dispatch.h>
408
+ #include <ATen/ops/logical_xor_cuda_dispatch.h>
409
+ #include <ATen/ops/logit_cuda_dispatch.h>
410
+ #include <ATen/ops/logit_backward_cuda_dispatch.h>
411
+ #include <ATen/ops/logspace_cuda_dispatch.h>
412
+ #include <ATen/ops/lshift_cuda_dispatch.h>
413
+ #include <ATen/ops/lt_cuda_dispatch.h>
414
+ #include <ATen/ops/lu_unpack_cuda_dispatch.h>
415
+ #include <ATen/ops/masked_fill_cuda_dispatch.h>
416
+ #include <ATen/ops/masked_scatter_cuda_dispatch.h>
417
+ #include <ATen/ops/masked_select_cuda_dispatch.h>
418
+ #include <ATen/ops/max_cuda_dispatch.h>
419
+ #include <ATen/ops/max_pool2d_with_indices_cuda_dispatch.h>
420
+ #include <ATen/ops/max_pool2d_with_indices_backward_cuda_dispatch.h>
421
+ #include <ATen/ops/max_pool3d_with_indices_cuda_dispatch.h>
422
+ #include <ATen/ops/max_pool3d_with_indices_backward_cuda_dispatch.h>
423
+ #include <ATen/ops/max_unpool2d_cuda_dispatch.h>
424
+ #include <ATen/ops/max_unpool3d_cuda_dispatch.h>
425
+ #include <ATen/ops/maximum_cuda_dispatch.h>
426
+ #include <ATen/ops/mean_cuda_dispatch.h>
427
+ #include <ATen/ops/median_cuda_dispatch.h>
428
+ #include <ATen/ops/min_cuda_dispatch.h>
429
+ #include <ATen/ops/minimum_cuda_dispatch.h>
430
+ #include <ATen/ops/miopen_batch_norm_cuda_dispatch.h>
431
+ #include <ATen/ops/miopen_batch_norm_backward_cuda_dispatch.h>
432
+ #include <ATen/ops/miopen_convolution_cuda_dispatch.h>
433
+ #include <ATen/ops/miopen_convolution_add_relu_cuda_dispatch.h>
434
+ #include <ATen/ops/miopen_convolution_relu_cuda_dispatch.h>
435
+ #include <ATen/ops/miopen_convolution_transpose_cuda_dispatch.h>
436
+ #include <ATen/ops/miopen_depthwise_convolution_cuda_dispatch.h>
437
+ #include <ATen/ops/miopen_rnn_cuda_dispatch.h>
438
+ #include <ATen/ops/miopen_rnn_backward_cuda_dispatch.h>
439
+ #include <ATen/ops/mish_cuda_dispatch.h>
440
+ #include <ATen/ops/mish_backward_cuda_dispatch.h>
441
+ #include <ATen/ops/mm_cuda_dispatch.h>
442
+ #include <ATen/ops/mode_cuda_dispatch.h>
443
+ #include <ATen/ops/mse_loss_cuda_dispatch.h>
444
+ #include <ATen/ops/mse_loss_backward_cuda_dispatch.h>
445
+ #include <ATen/ops/mul_cuda_dispatch.h>
446
+ #include <ATen/ops/multi_margin_loss_cuda_dispatch.h>
447
+ #include <ATen/ops/multi_margin_loss_backward_cuda_dispatch.h>
448
+ #include <ATen/ops/multilabel_margin_loss_backward_cuda_dispatch.h>
449
+ #include <ATen/ops/multilabel_margin_loss_forward_cuda_dispatch.h>
450
+ #include <ATen/ops/multinomial_cuda_dispatch.h>
451
+ #include <ATen/ops/mvlgamma_cuda_dispatch.h>
452
+ #include <ATen/ops/nan_to_num_cuda_dispatch.h>
453
+ #include <ATen/ops/nanmedian_cuda_dispatch.h>
454
+ #include <ATen/ops/nansum_cuda_dispatch.h>
455
+ #include <ATen/ops/native_batch_norm_cuda_dispatch.h>
456
+ #include <ATen/ops/native_batch_norm_backward_cuda_dispatch.h>
457
+ #include <ATen/ops/native_dropout_cuda_dispatch.h>
458
+ #include <ATen/ops/native_dropout_backward_cuda_dispatch.h>
459
+ #include <ATen/ops/native_group_norm_cuda_dispatch.h>
460
+ #include <ATen/ops/native_group_norm_backward_cuda_dispatch.h>
461
+ #include <ATen/ops/native_layer_norm_cuda_dispatch.h>
462
+ #include <ATen/ops/native_layer_norm_backward_cuda_dispatch.h>
463
+ #include <ATen/ops/ne_cuda_dispatch.h>
464
+ #include <ATen/ops/neg_cuda_dispatch.h>
465
+ #include <ATen/ops/nextafter_cuda_dispatch.h>
466
+ #include <ATen/ops/nll_loss2d_backward_cuda_dispatch.h>
467
+ #include <ATen/ops/nll_loss2d_forward_cuda_dispatch.h>
468
+ #include <ATen/ops/nll_loss_backward_cuda_dispatch.h>
469
+ #include <ATen/ops/nll_loss_forward_cuda_dispatch.h>
470
+ #include <ATen/ops/nonzero_cuda_dispatch.h>
471
+ #include <ATen/ops/norm_cuda_dispatch.h>
472
+ #include <ATen/ops/normal_cuda_dispatch.h>
473
+ #include <ATen/ops/ormqr_cuda_dispatch.h>
474
+ #include <ATen/ops/poisson_cuda_dispatch.h>
475
+ #include <ATen/ops/polar_cuda_dispatch.h>
476
+ #include <ATen/ops/polygamma_cuda_dispatch.h>
477
+ #include <ATen/ops/pow_cuda_dispatch.h>
478
+ #include <ATen/ops/prod_cuda_dispatch.h>
479
+ #include <ATen/ops/put_cuda_dispatch.h>
480
+ #include <ATen/ops/quantize_per_channel_cuda_dispatch.h>
481
+ #include <ATen/ops/quantize_per_tensor_cuda_dispatch.h>
482
+ #include <ATen/ops/quantize_per_tensor_dynamic_cuda_dispatch.h>
483
+ #include <ATen/ops/random_cuda_dispatch.h>
484
+ #include <ATen/ops/randperm_cuda_dispatch.h>
485
+ #include <ATen/ops/range_cuda_dispatch.h>
486
+ #include <ATen/ops/reciprocal_cuda_dispatch.h>
487
+ #include <ATen/ops/record_stream_cuda_dispatch.h>
488
+ #include <ATen/ops/reflection_pad1d_cuda_dispatch.h>
489
+ #include <ATen/ops/reflection_pad1d_backward_cuda_dispatch.h>
490
+ #include <ATen/ops/reflection_pad2d_cuda_dispatch.h>
491
+ #include <ATen/ops/reflection_pad2d_backward_cuda_dispatch.h>
492
+ #include <ATen/ops/reflection_pad3d_cuda_dispatch.h>
493
+ #include <ATen/ops/reflection_pad3d_backward_cuda_dispatch.h>
494
+ #include <ATen/ops/relu_cuda_dispatch.h>
495
+ #include <ATen/ops/remainder_cuda_dispatch.h>
496
+ #include <ATen/ops/renorm_cuda_dispatch.h>
497
+ #include <ATen/ops/repeat_interleave_cuda_dispatch.h>
498
+ #include <ATen/ops/replication_pad1d_cuda_dispatch.h>
499
+ #include <ATen/ops/replication_pad1d_backward_cuda_dispatch.h>
500
+ #include <ATen/ops/replication_pad2d_cuda_dispatch.h>
501
+ #include <ATen/ops/replication_pad2d_backward_cuda_dispatch.h>
502
+ #include <ATen/ops/replication_pad3d_cuda_dispatch.h>
503
+ #include <ATen/ops/replication_pad3d_backward_cuda_dispatch.h>
504
+ #include <ATen/ops/resize_cuda_dispatch.h>
505
+ #include <ATen/ops/roll_cuda_dispatch.h>
506
+ #include <ATen/ops/round_cuda_dispatch.h>
507
+ #include <ATen/ops/rrelu_with_noise_cuda_dispatch.h>
508
+ #include <ATen/ops/rshift_cuda_dispatch.h>
509
+ #include <ATen/ops/rsqrt_cuda_dispatch.h>
510
+ #include <ATen/ops/rsub_cuda_dispatch.h>
511
+ #include <ATen/ops/scatter_cuda_dispatch.h>
512
+ #include <ATen/ops/scatter_add_cuda_dispatch.h>
513
+ #include <ATen/ops/scatter_reduce_cuda_dispatch.h>
514
+ #include <ATen/ops/searchsorted_cuda_dispatch.h>
515
+ #include <ATen/ops/segment_reduce_cuda_dispatch.h>
516
+ #include <ATen/ops/set_cuda_dispatch.h>
517
+ #include <ATen/ops/sgn_cuda_dispatch.h>
518
+ #include <ATen/ops/sigmoid_cuda_dispatch.h>
519
+ #include <ATen/ops/sigmoid_backward_cuda_dispatch.h>
520
+ #include <ATen/ops/sign_cuda_dispatch.h>
521
+ #include <ATen/ops/signbit_cuda_dispatch.h>
522
+ #include <ATen/ops/silu_cuda_dispatch.h>
523
+ #include <ATen/ops/silu_backward_cuda_dispatch.h>
524
+ #include <ATen/ops/sin_cuda_dispatch.h>
525
+ #include <ATen/ops/sinc_cuda_dispatch.h>
526
+ #include <ATen/ops/sinh_cuda_dispatch.h>
527
+ #include <ATen/ops/slow_conv_dilated2d_cuda_dispatch.h>
528
+ #include <ATen/ops/slow_conv_dilated3d_cuda_dispatch.h>
529
+ #include <ATen/ops/slow_conv_transpose2d_cuda_dispatch.h>
530
+ #include <ATen/ops/slow_conv_transpose3d_cuda_dispatch.h>
531
+ #include <ATen/ops/smooth_l1_loss_cuda_dispatch.h>
532
+ #include <ATen/ops/smooth_l1_loss_backward_cuda_dispatch.h>
533
+ #include <ATen/ops/softplus_cuda_dispatch.h>
534
+ #include <ATen/ops/softplus_backward_cuda_dispatch.h>
535
+ #include <ATen/ops/softshrink_cuda_dispatch.h>
536
+ #include <ATen/ops/softshrink_backward_cuda_dispatch.h>
537
+ #include <ATen/ops/sort_cuda_dispatch.h>
538
+ #include <ATen/ops/special_airy_ai_cuda_dispatch.h>
539
+ #include <ATen/ops/special_bessel_j0_cuda_dispatch.h>
540
+ #include <ATen/ops/special_bessel_j1_cuda_dispatch.h>
541
+ #include <ATen/ops/special_bessel_y0_cuda_dispatch.h>
542
+ #include <ATen/ops/special_bessel_y1_cuda_dispatch.h>
543
+ #include <ATen/ops/special_chebyshev_polynomial_t_cuda_dispatch.h>
544
+ #include <ATen/ops/special_chebyshev_polynomial_u_cuda_dispatch.h>
545
+ #include <ATen/ops/special_chebyshev_polynomial_v_cuda_dispatch.h>
546
+ #include <ATen/ops/special_chebyshev_polynomial_w_cuda_dispatch.h>
547
+ #include <ATen/ops/special_entr_cuda_dispatch.h>
548
+ #include <ATen/ops/special_erfcx_cuda_dispatch.h>
549
+ #include <ATen/ops/special_hermite_polynomial_h_cuda_dispatch.h>
550
+ #include <ATen/ops/special_hermite_polynomial_he_cuda_dispatch.h>
551
+ #include <ATen/ops/special_i0e_cuda_dispatch.h>
552
+ #include <ATen/ops/special_i1_cuda_dispatch.h>
553
+ #include <ATen/ops/special_i1e_cuda_dispatch.h>
554
+ #include <ATen/ops/special_laguerre_polynomial_l_cuda_dispatch.h>
555
+ #include <ATen/ops/special_legendre_polynomial_p_cuda_dispatch.h>
556
+ #include <ATen/ops/special_log_ndtr_cuda_dispatch.h>
557
+ #include <ATen/ops/special_modified_bessel_i0_cuda_dispatch.h>
558
+ #include <ATen/ops/special_modified_bessel_i1_cuda_dispatch.h>
559
+ #include <ATen/ops/special_modified_bessel_k0_cuda_dispatch.h>
560
+ #include <ATen/ops/special_modified_bessel_k1_cuda_dispatch.h>
561
+ #include <ATen/ops/special_ndtri_cuda_dispatch.h>
562
+ #include <ATen/ops/special_scaled_modified_bessel_k0_cuda_dispatch.h>
563
+ #include <ATen/ops/special_scaled_modified_bessel_k1_cuda_dispatch.h>
564
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_t_cuda_dispatch.h>
565
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_u_cuda_dispatch.h>
566
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_v_cuda_dispatch.h>
567
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_w_cuda_dispatch.h>
568
+ #include <ATen/ops/special_spherical_bessel_j0_cuda_dispatch.h>
569
+ #include <ATen/ops/special_xlog1py_cuda_dispatch.h>
570
+ #include <ATen/ops/special_zeta_cuda_dispatch.h>
571
+ #include <ATen/ops/split_with_sizes_copy_cuda_dispatch.h>
572
+ #include <ATen/ops/sqrt_cuda_dispatch.h>
573
+ #include <ATen/ops/sspaddmm_cuda_dispatch.h>
574
+ #include <ATen/ops/std_cuda_dispatch.h>
575
+ #include <ATen/ops/std_mean_cuda_dispatch.h>
576
+ #include <ATen/ops/sub_cuda_dispatch.h>
577
+ #include <ATen/ops/sum_cuda_dispatch.h>
578
+ #include <ATen/ops/take_cuda_dispatch.h>
579
+ #include <ATen/ops/tan_cuda_dispatch.h>
580
+ #include <ATen/ops/tanh_cuda_dispatch.h>
581
+ #include <ATen/ops/tanh_backward_cuda_dispatch.h>
582
+ #include <ATen/ops/threshold_cuda_dispatch.h>
583
+ #include <ATen/ops/threshold_backward_cuda_dispatch.h>
584
+ #include <ATen/ops/topk_cuda_dispatch.h>
585
+ #include <ATen/ops/trace_cuda_dispatch.h>
586
+ #include <ATen/ops/triangular_solve_cuda_dispatch.h>
587
+ #include <ATen/ops/tril_cuda_dispatch.h>
588
+ #include <ATen/ops/tril_indices_cuda_dispatch.h>
589
+ #include <ATen/ops/triu_cuda_dispatch.h>
590
+ #include <ATen/ops/triu_indices_cuda_dispatch.h>
591
+ #include <ATen/ops/trunc_cuda_dispatch.h>
592
+ #include <ATen/ops/unfold_cuda_dispatch.h>
593
+ #include <ATen/ops/unfold_backward_cuda_dispatch.h>
594
+ #include <ATen/ops/uniform_cuda_dispatch.h>
595
+ #include <ATen/ops/unique_consecutive_cuda_dispatch.h>
596
+ #include <ATen/ops/unique_dim_cuda_dispatch.h>
597
+ #include <ATen/ops/unique_dim_consecutive_cuda_dispatch.h>
598
+ #include <ATen/ops/upsample_bicubic2d_cuda_dispatch.h>
599
+ #include <ATen/ops/upsample_bicubic2d_backward_cuda_dispatch.h>
600
+ #include <ATen/ops/upsample_bilinear2d_cuda_dispatch.h>
601
+ #include <ATen/ops/upsample_bilinear2d_backward_cuda_dispatch.h>
602
+ #include <ATen/ops/upsample_linear1d_cuda_dispatch.h>
603
+ #include <ATen/ops/upsample_linear1d_backward_cuda_dispatch.h>
604
+ #include <ATen/ops/upsample_nearest1d_cuda_dispatch.h>
605
+ #include <ATen/ops/upsample_nearest1d_backward_cuda_dispatch.h>
606
+ #include <ATen/ops/upsample_nearest2d_cuda_dispatch.h>
607
+ #include <ATen/ops/upsample_nearest2d_backward_cuda_dispatch.h>
608
+ #include <ATen/ops/upsample_nearest3d_cuda_dispatch.h>
609
+ #include <ATen/ops/upsample_nearest3d_backward_cuda_dispatch.h>
610
+ #include <ATen/ops/upsample_trilinear3d_cuda_dispatch.h>
611
+ #include <ATen/ops/upsample_trilinear3d_backward_cuda_dispatch.h>
612
+ #include <ATen/ops/var_cuda_dispatch.h>
613
+ #include <ATen/ops/var_mean_cuda_dispatch.h>
614
+ #include <ATen/ops/vdot_cuda_dispatch.h>
615
+ #include <ATen/ops/view_cuda_dispatch.h>
616
+ #include <ATen/ops/view_as_complex_cuda_dispatch.h>
617
+ #include <ATen/ops/view_as_real_cuda_dispatch.h>
618
+ #include <ATen/ops/where_cuda_dispatch.h>
619
+ #include <ATen/ops/xlogy_cuda_dispatch.h>
620
+ #include <ATen/ops/zero_cuda_dispatch.h>
621
+
622
+
623
+
.venv/lib/python3.11/site-packages/torch/include/ATen/CachedTensorUtils.h ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <ATen/ATen.h>
4
+
5
+ namespace at::caching {
6
+
7
+ // Some systems (just cudagraphs currently) will persist a static tensor output
8
+ // whose TensorImpl does not change across iterations. For these tensors caching
9
+ // dtype conversions is invalid. Additionally, there will be an extra reference
10
+ // count to these cached tensors that would prevent buffer inplacing and other
11
+ // checks on tensor uniqueness. If we are not using these systems the enabled
12
+ // flag will be false and we will avoid the hash lookup.
13
+
14
+ TORCH_API bool is_cached_tensor(const at::Tensor& t);
15
+ TORCH_API void add_cached_tensor(const at::Tensor& t);
16
+ TORCH_API void remove_cached_tensor(const at::Tensor& t);
17
+ TORCH_API void set_cached_tensors_enabled(bool enable);
18
+
19
+ // For gradient buffer stealing we will adjust the use count of tensors
20
+ // which are persisted by cudagraphs, just as we need to adjust reference
21
+ // count of tensors with hooks.
22
+ TORCH_API size_t adjusted_use_count(const at::Tensor& t);
23
+
24
+ } // namespace at::caching
.venv/lib/python3.11/site-packages/torch/include/ATen/CollapseDims.h ADDED
@@ -0,0 +1,94 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #include <c10/util/Exception.h>
2
+ #include <utility>
3
+
4
+ namespace at {
5
+
6
+ /*
7
+ [collapse dims] Updates sizes, and strides to reflect a "collapse" of
8
+ the info, possibly excluding the optional excludeDim. A "collapsed" version
9
+ of the info is the fewest dims that order the tensor's elements in the same
10
+ way as the original info. If excludeDim is specified, the collapse is the
11
+ fewest dims that order the tensor's elements as the original and preserve the
12
+ excluded dimension, unless the tensor collapses to a point.
13
+
14
+ This function returns a pair of values.
15
+
16
+ 1) The (new) index of the preserved dimension if excludeDim is
17
+ specified. 0 if the tensor is collapsed to a point. -1
18
+ otherwise.
19
+
20
+ 2) The new number of dimensions.
21
+ */
22
+ template <typename T>
23
+ inline std::pair<int64_t, int64_t> collapse_dims(
24
+ T* sizes,
25
+ T* strides,
26
+ int64_t dims,
27
+ const int excludeDim = -1) {
28
+ TORCH_CHECK(
29
+ excludeDim >= -1 && excludeDim < dims,
30
+ "expected excluded dim between -1 and dims - 1");
31
+
32
+ int64_t stopDim = (excludeDim == -1) ? dims : excludeDim;
33
+ int64_t newIndex = -1;
34
+ int64_t oldIndex = 0;
35
+ int64_t remappedExcludedDim = -1;
36
+
37
+ while (oldIndex < dims) {
38
+ // Finds a dimension to collapse into
39
+ for (; oldIndex < stopDim; ++oldIndex) {
40
+ if (sizes[oldIndex] == 1) {
41
+ continue;
42
+ }
43
+
44
+ ++newIndex;
45
+ sizes[newIndex] = sizes[oldIndex];
46
+ strides[newIndex] = strides[oldIndex];
47
+ ++oldIndex;
48
+ break;
49
+ }
50
+
51
+ // Collapses dims
52
+ for (; oldIndex < stopDim; ++oldIndex) {
53
+ if (sizes[oldIndex] == 1) {
54
+ continue;
55
+ }
56
+
57
+ if (strides[newIndex] == sizes[oldIndex] * strides[oldIndex]) {
58
+ sizes[newIndex] *= sizes[oldIndex];
59
+ strides[newIndex] = strides[oldIndex];
60
+ } else {
61
+ ++newIndex;
62
+ sizes[newIndex] = sizes[oldIndex];
63
+ strides[newIndex] = strides[oldIndex];
64
+ }
65
+ }
66
+
67
+ // Handles excludeDim being set (oldIndex == excludeDim)
68
+ if (oldIndex != dims) {
69
+ // Preserves excluded dimension
70
+ ++newIndex;
71
+ sizes[newIndex] = sizes[oldIndex];
72
+ strides[newIndex] = strides[oldIndex];
73
+ remappedExcludedDim = newIndex;
74
+
75
+ // Restarts iteration after excludeDim
76
+ ++oldIndex;
77
+ stopDim = dims;
78
+ }
79
+ }
80
+
81
+ // Handles special case of all dims size 1
82
+ if (newIndex == -1 || (newIndex == 0 && sizes[0] == 1)) {
83
+ dims = 1;
84
+ sizes[0] = 1;
85
+ strides[0] = 1;
86
+
87
+ return std::pair<int64_t, int64_t>(0, 1);
88
+ }
89
+
90
+ dims = newIndex + 1;
91
+ return std::pair<int64_t, int64_t>(remappedExcludedDim, dims);
92
+ }
93
+
94
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/CompositeExplicitAutogradFunctions.h ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #include <ATen/core/TensorBody.h>
2
+
3
+ // TODO Undo all logic introduced for Note [Avoiding Include Cycles In Static Dispatch]
4
+ // Code introduced to avoid cyclic dependency in static dispatch is no longer
5
+ // needed as static dispatch logic is moved from TensorBody.h, which caused cycles in the first place,
6
+ // to Operators.cpp for supporting multiple backends with multiple kernels.
7
+ //
8
+ // Note [Avoiding Include Cycles In Static Dispatch]
9
+ // In order to avoid #include cycles in the static dispatch build, we've carefully split out
10
+ // the static function definition files into {DispatchKey}Functions.h and {DispatchKey}Functions_inl.h.
11
+ //
12
+ // Without this split, the include cycle looks like TensorBody.h -> CPUFunctions.h -> TensorBody.h.
13
+ // - TensorBody.h #includes CPUFunctions.h in the static dispatch build, because the tensor methods
14
+ // all need to call into the fastpath C++ API defined in CPUFunctions.h. The methods are also all
15
+ // directly inlined into TensorBody.h.
16
+ // - CPUFunctions.h #includes TensorBody.h because it contains function declarations for the entire C++ API,
17
+ // which include functions that have defaultable std::optional<Tensor> arguments.
18
+ // That requires knowing the full Tensor class definition.
19
+ //
20
+ // We break the cycle by doing the following:
21
+ // - Split out CPUFunction.h into two files: CPUFunctions.h and CPUFunctions_inl.h
22
+ // - CPUFunction.h is a dummy file that just includes the Tensor class and includes CPUFunctions_inl.,
23
+ // - CPUFunctions_inl.h includes everything else
24
+ // - (only in the static dispatch build) TensorBody.h makes sure to finish defining the Tensor class,
25
+ // and then it includes CPUFunctions_inl.h.
26
+ // - All other files that want the cpu fastpath functions can include CPUFunctions.h directly.
27
+ // - This also means that static dispatch build, CPUFunctions.h only needs to
28
+ // #include TensorBody.h, and it will automatically bring in CPUFunctions_inl.h.
29
+ #include <ATen/CompositeExplicitAutogradFunctions_inl.h>
.venv/lib/python3.11/site-packages/torch/include/ATen/CompositeExplicitAutogradNonFunctionalFunctions.h ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #include <ATen/core/TensorBody.h>
2
+
3
+ // TODO Undo all logic introduced for Note [Avoiding Include Cycles In Static Dispatch]
4
+ // Code introduced to avoid cyclic dependency in static dispatch is no longer
5
+ // needed as static dispatch logic is moved from TensorBody.h, which caused cycles in the first place,
6
+ // to Operators.cpp for supporting multiple backends with multiple kernels.
7
+ //
8
+ // Note [Avoiding Include Cycles In Static Dispatch]
9
+ // In order to avoid #include cycles in the static dispatch build, we've carefully split out
10
+ // the static function definition files into {DispatchKey}Functions.h and {DispatchKey}Functions_inl.h.
11
+ //
12
+ // Without this split, the include cycle looks like TensorBody.h -> CPUFunctions.h -> TensorBody.h.
13
+ // - TensorBody.h #includes CPUFunctions.h in the static dispatch build, because the tensor methods
14
+ // all need to call into the fastpath C++ API defined in CPUFunctions.h. The methods are also all
15
+ // directly inlined into TensorBody.h.
16
+ // - CPUFunctions.h #includes TensorBody.h because it contains function declarations for the entire C++ API,
17
+ // which include functions that have defaultable std::optional<Tensor> arguments.
18
+ // That requires knowing the full Tensor class definition.
19
+ //
20
+ // We break the cycle by doing the following:
21
+ // - Split out CPUFunction.h into two files: CPUFunctions.h and CPUFunctions_inl.h
22
+ // - CPUFunction.h is a dummy file that just includes the Tensor class and includes CPUFunctions_inl.,
23
+ // - CPUFunctions_inl.h includes everything else
24
+ // - (only in the static dispatch build) TensorBody.h makes sure to finish defining the Tensor class,
25
+ // and then it includes CPUFunctions_inl.h.
26
+ // - All other files that want the cpu fastpath functions can include CPUFunctions.h directly.
27
+ // - This also means that static dispatch build, CPUFunctions.h only needs to
28
+ // #include TensorBody.h, and it will automatically bring in CPUFunctions_inl.h.
29
+ #include <ATen/CompositeExplicitAutogradNonFunctionalFunctions_inl.h>
.venv/lib/python3.11/site-packages/torch/include/ATen/CompositeExplicitAutogradNonFunctionalFunctions_inl.h ADDED
@@ -0,0 +1,323 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+ // @generated by torchgen/gen.py from DispatchKeyFunctions_inl.h
3
+
4
+ // NB: The implementing C++ file is RegisterDispatchKey.cpp
5
+
6
+ // The only #includes we need are for custom classes that have defaults in the C++ API
7
+ #include <c10/core/MemoryFormat.h>
8
+ #include <c10/core/Scalar.h>
9
+ #include <ATen/core/Reduction.h>
10
+
11
+ #if defined(AT_PER_OPERATOR_HEADERS) && defined(TORCH_ASSERT_ONLY_METHOD_OPERATORS)
12
+ #error This change adds a dependency on all pytorch operators, meaning the \
13
+ file will need to be re-compiled every time an operator is changed or added. \
14
+ Consider including a specific operator from \
15
+ <ATen/ops/{my_operator}_compositeexplicitautogradnonfunctional_dispatch.h>. \
16
+ See NOTE [TORCH_ASSERT_ONLY_METHOD_OPERATORS].
17
+ #endif
18
+
19
+ #include <ATen/ops/_addmm_activation_compositeexplicitautogradnonfunctional_dispatch.h>
20
+ #include <ATen/ops/_conj_copy_compositeexplicitautogradnonfunctional_dispatch.h>
21
+ #include <ATen/ops/_convert_indices_from_coo_to_csr_compositeexplicitautogradnonfunctional_dispatch.h>
22
+ #include <ATen/ops/_convert_indices_from_csr_to_coo_compositeexplicitautogradnonfunctional_dispatch.h>
23
+ #include <ATen/ops/_fw_primal_copy_compositeexplicitautogradnonfunctional_dispatch.h>
24
+ #include <ATen/ops/_indices_copy_compositeexplicitautogradnonfunctional_dispatch.h>
25
+ #include <ATen/ops/_linalg_det_compositeexplicitautogradnonfunctional_dispatch.h>
26
+ #include <ATen/ops/_linalg_eigh_compositeexplicitautogradnonfunctional_dispatch.h>
27
+ #include <ATen/ops/_linalg_slogdet_compositeexplicitautogradnonfunctional_dispatch.h>
28
+ #include <ATen/ops/_linalg_solve_ex_compositeexplicitautogradnonfunctional_dispatch.h>
29
+ #include <ATen/ops/_linalg_svd_compositeexplicitautogradnonfunctional_dispatch.h>
30
+ #include <ATen/ops/_log_softmax_compositeexplicitautogradnonfunctional_dispatch.h>
31
+ #include <ATen/ops/_log_softmax_backward_data_compositeexplicitautogradnonfunctional_dispatch.h>
32
+ #include <ATen/ops/_make_dual_copy_compositeexplicitautogradnonfunctional_dispatch.h>
33
+ #include <ATen/ops/_neg_view_copy_compositeexplicitautogradnonfunctional_dispatch.h>
34
+ #include <ATen/ops/_nested_get_values_copy_compositeexplicitautogradnonfunctional_dispatch.h>
35
+ #include <ATen/ops/_nested_view_from_buffer_copy_compositeexplicitautogradnonfunctional_dispatch.h>
36
+ #include <ATen/ops/_nested_view_from_jagged_copy_compositeexplicitautogradnonfunctional_dispatch.h>
37
+ #include <ATen/ops/_reshape_alias_copy_compositeexplicitautogradnonfunctional_dispatch.h>
38
+ #include <ATen/ops/_softmax_compositeexplicitautogradnonfunctional_dispatch.h>
39
+ #include <ATen/ops/_softmax_backward_data_compositeexplicitautogradnonfunctional_dispatch.h>
40
+ #include <ATen/ops/_sparse_broadcast_to_copy_compositeexplicitautogradnonfunctional_dispatch.h>
41
+ #include <ATen/ops/_test_autograd_multiple_dispatch_view_copy_compositeexplicitautogradnonfunctional_dispatch.h>
42
+ #include <ATen/ops/_trilinear_compositeexplicitautogradnonfunctional_dispatch.h>
43
+ #include <ATen/ops/_upsample_bicubic2d_aa_compositeexplicitautogradnonfunctional_dispatch.h>
44
+ #include <ATen/ops/_upsample_bicubic2d_aa_backward_compositeexplicitautogradnonfunctional_dispatch.h>
45
+ #include <ATen/ops/_upsample_bilinear2d_aa_compositeexplicitautogradnonfunctional_dispatch.h>
46
+ #include <ATen/ops/_upsample_bilinear2d_aa_backward_compositeexplicitautogradnonfunctional_dispatch.h>
47
+ #include <ATen/ops/_upsample_nearest_exact1d_compositeexplicitautogradnonfunctional_dispatch.h>
48
+ #include <ATen/ops/_upsample_nearest_exact1d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
49
+ #include <ATen/ops/_upsample_nearest_exact2d_compositeexplicitautogradnonfunctional_dispatch.h>
50
+ #include <ATen/ops/_upsample_nearest_exact2d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
51
+ #include <ATen/ops/_upsample_nearest_exact3d_compositeexplicitautogradnonfunctional_dispatch.h>
52
+ #include <ATen/ops/_upsample_nearest_exact3d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
53
+ #include <ATen/ops/_values_copy_compositeexplicitautogradnonfunctional_dispatch.h>
54
+ #include <ATen/ops/acos_compositeexplicitautogradnonfunctional_dispatch.h>
55
+ #include <ATen/ops/acosh_compositeexplicitautogradnonfunctional_dispatch.h>
56
+ #include <ATen/ops/adaptive_max_pool2d_compositeexplicitautogradnonfunctional_dispatch.h>
57
+ #include <ATen/ops/adaptive_max_pool2d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
58
+ #include <ATen/ops/adaptive_max_pool3d_compositeexplicitautogradnonfunctional_dispatch.h>
59
+ #include <ATen/ops/adaptive_max_pool3d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
60
+ #include <ATen/ops/add_compositeexplicitautogradnonfunctional_dispatch.h>
61
+ #include <ATen/ops/addcdiv_compositeexplicitautogradnonfunctional_dispatch.h>
62
+ #include <ATen/ops/addcmul_compositeexplicitautogradnonfunctional_dispatch.h>
63
+ #include <ATen/ops/addmm_compositeexplicitautogradnonfunctional_dispatch.h>
64
+ #include <ATen/ops/addmv_compositeexplicitautogradnonfunctional_dispatch.h>
65
+ #include <ATen/ops/alias_copy_compositeexplicitautogradnonfunctional_dispatch.h>
66
+ #include <ATen/ops/all_compositeexplicitautogradnonfunctional_dispatch.h>
67
+ #include <ATen/ops/amax_compositeexplicitautogradnonfunctional_dispatch.h>
68
+ #include <ATen/ops/amin_compositeexplicitautogradnonfunctional_dispatch.h>
69
+ #include <ATen/ops/aminmax_compositeexplicitautogradnonfunctional_dispatch.h>
70
+ #include <ATen/ops/any_compositeexplicitautogradnonfunctional_dispatch.h>
71
+ #include <ATen/ops/argmax_compositeexplicitautogradnonfunctional_dispatch.h>
72
+ #include <ATen/ops/argmin_compositeexplicitautogradnonfunctional_dispatch.h>
73
+ #include <ATen/ops/as_strided_compositeexplicitautogradnonfunctional_dispatch.h>
74
+ #include <ATen/ops/as_strided_copy_compositeexplicitautogradnonfunctional_dispatch.h>
75
+ #include <ATen/ops/as_strided_scatter_compositeexplicitautogradnonfunctional_dispatch.h>
76
+ #include <ATen/ops/asin_compositeexplicitautogradnonfunctional_dispatch.h>
77
+ #include <ATen/ops/asinh_compositeexplicitautogradnonfunctional_dispatch.h>
78
+ #include <ATen/ops/atan_compositeexplicitautogradnonfunctional_dispatch.h>
79
+ #include <ATen/ops/atan2_compositeexplicitautogradnonfunctional_dispatch.h>
80
+ #include <ATen/ops/atanh_compositeexplicitautogradnonfunctional_dispatch.h>
81
+ #include <ATen/ops/avg_pool2d_compositeexplicitautogradnonfunctional_dispatch.h>
82
+ #include <ATen/ops/avg_pool2d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
83
+ #include <ATen/ops/avg_pool3d_compositeexplicitautogradnonfunctional_dispatch.h>
84
+ #include <ATen/ops/avg_pool3d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
85
+ #include <ATen/ops/baddbmm_compositeexplicitautogradnonfunctional_dispatch.h>
86
+ #include <ATen/ops/bernoulli_compositeexplicitautogradnonfunctional_dispatch.h>
87
+ #include <ATen/ops/bitwise_and_compositeexplicitautogradnonfunctional_dispatch.h>
88
+ #include <ATen/ops/bitwise_left_shift_compositeexplicitautogradnonfunctional_dispatch.h>
89
+ #include <ATen/ops/bitwise_not_compositeexplicitautogradnonfunctional_dispatch.h>
90
+ #include <ATen/ops/bitwise_or_compositeexplicitautogradnonfunctional_dispatch.h>
91
+ #include <ATen/ops/bitwise_right_shift_compositeexplicitautogradnonfunctional_dispatch.h>
92
+ #include <ATen/ops/bitwise_xor_compositeexplicitautogradnonfunctional_dispatch.h>
93
+ #include <ATen/ops/bmm_compositeexplicitautogradnonfunctional_dispatch.h>
94
+ #include <ATen/ops/cat_compositeexplicitautogradnonfunctional_dispatch.h>
95
+ #include <ATen/ops/ccol_indices_copy_compositeexplicitautogradnonfunctional_dispatch.h>
96
+ #include <ATen/ops/ceil_compositeexplicitautogradnonfunctional_dispatch.h>
97
+ #include <ATen/ops/clamp_compositeexplicitautogradnonfunctional_dispatch.h>
98
+ #include <ATen/ops/clamp_max_compositeexplicitautogradnonfunctional_dispatch.h>
99
+ #include <ATen/ops/clamp_min_compositeexplicitautogradnonfunctional_dispatch.h>
100
+ #include <ATen/ops/col_indices_copy_compositeexplicitautogradnonfunctional_dispatch.h>
101
+ #include <ATen/ops/copy_compositeexplicitautogradnonfunctional_dispatch.h>
102
+ #include <ATen/ops/copysign_compositeexplicitautogradnonfunctional_dispatch.h>
103
+ #include <ATen/ops/cos_compositeexplicitautogradnonfunctional_dispatch.h>
104
+ #include <ATen/ops/cosh_compositeexplicitautogradnonfunctional_dispatch.h>
105
+ #include <ATen/ops/crow_indices_copy_compositeexplicitautogradnonfunctional_dispatch.h>
106
+ #include <ATen/ops/cumprod_compositeexplicitautogradnonfunctional_dispatch.h>
107
+ #include <ATen/ops/cumsum_compositeexplicitautogradnonfunctional_dispatch.h>
108
+ #include <ATen/ops/detach_copy_compositeexplicitautogradnonfunctional_dispatch.h>
109
+ #include <ATen/ops/diag_embed_compositeexplicitautogradnonfunctional_dispatch.h>
110
+ #include <ATen/ops/diagonal_copy_compositeexplicitautogradnonfunctional_dispatch.h>
111
+ #include <ATen/ops/diagonal_scatter_compositeexplicitautogradnonfunctional_dispatch.h>
112
+ #include <ATen/ops/digamma_compositeexplicitautogradnonfunctional_dispatch.h>
113
+ #include <ATen/ops/div_compositeexplicitautogradnonfunctional_dispatch.h>
114
+ #include <ATen/ops/elu_compositeexplicitautogradnonfunctional_dispatch.h>
115
+ #include <ATen/ops/elu_backward_compositeexplicitautogradnonfunctional_dispatch.h>
116
+ #include <ATen/ops/eq_compositeexplicitautogradnonfunctional_dispatch.h>
117
+ #include <ATen/ops/erf_compositeexplicitautogradnonfunctional_dispatch.h>
118
+ #include <ATen/ops/erfc_compositeexplicitautogradnonfunctional_dispatch.h>
119
+ #include <ATen/ops/erfinv_compositeexplicitautogradnonfunctional_dispatch.h>
120
+ #include <ATen/ops/exp_compositeexplicitautogradnonfunctional_dispatch.h>
121
+ #include <ATen/ops/exp2_compositeexplicitautogradnonfunctional_dispatch.h>
122
+ #include <ATen/ops/expand_copy_compositeexplicitautogradnonfunctional_dispatch.h>
123
+ #include <ATen/ops/expm1_compositeexplicitautogradnonfunctional_dispatch.h>
124
+ #include <ATen/ops/floor_compositeexplicitautogradnonfunctional_dispatch.h>
125
+ #include <ATen/ops/fmax_compositeexplicitautogradnonfunctional_dispatch.h>
126
+ #include <ATen/ops/fmin_compositeexplicitautogradnonfunctional_dispatch.h>
127
+ #include <ATen/ops/fmod_compositeexplicitautogradnonfunctional_dispatch.h>
128
+ #include <ATen/ops/frac_compositeexplicitautogradnonfunctional_dispatch.h>
129
+ #include <ATen/ops/fractional_max_pool2d_compositeexplicitautogradnonfunctional_dispatch.h>
130
+ #include <ATen/ops/fractional_max_pool2d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
131
+ #include <ATen/ops/fractional_max_pool3d_compositeexplicitautogradnonfunctional_dispatch.h>
132
+ #include <ATen/ops/gather_compositeexplicitautogradnonfunctional_dispatch.h>
133
+ #include <ATen/ops/gcd_compositeexplicitautogradnonfunctional_dispatch.h>
134
+ #include <ATen/ops/ge_compositeexplicitautogradnonfunctional_dispatch.h>
135
+ #include <ATen/ops/gelu_compositeexplicitautogradnonfunctional_dispatch.h>
136
+ #include <ATen/ops/gelu_backward_compositeexplicitautogradnonfunctional_dispatch.h>
137
+ #include <ATen/ops/glu_compositeexplicitautogradnonfunctional_dispatch.h>
138
+ #include <ATen/ops/gt_compositeexplicitautogradnonfunctional_dispatch.h>
139
+ #include <ATen/ops/hardshrink_compositeexplicitautogradnonfunctional_dispatch.h>
140
+ #include <ATen/ops/hardshrink_backward_compositeexplicitautogradnonfunctional_dispatch.h>
141
+ #include <ATen/ops/hardsigmoid_compositeexplicitautogradnonfunctional_dispatch.h>
142
+ #include <ATen/ops/hardsigmoid_backward_compositeexplicitautogradnonfunctional_dispatch.h>
143
+ #include <ATen/ops/heaviside_compositeexplicitautogradnonfunctional_dispatch.h>
144
+ #include <ATen/ops/hypot_compositeexplicitautogradnonfunctional_dispatch.h>
145
+ #include <ATen/ops/i0_compositeexplicitautogradnonfunctional_dispatch.h>
146
+ #include <ATen/ops/igamma_compositeexplicitautogradnonfunctional_dispatch.h>
147
+ #include <ATen/ops/igammac_compositeexplicitautogradnonfunctional_dispatch.h>
148
+ #include <ATen/ops/index_compositeexplicitautogradnonfunctional_dispatch.h>
149
+ #include <ATen/ops/index_add_compositeexplicitautogradnonfunctional_dispatch.h>
150
+ #include <ATen/ops/index_copy_compositeexplicitautogradnonfunctional_dispatch.h>
151
+ #include <ATen/ops/index_reduce_compositeexplicitautogradnonfunctional_dispatch.h>
152
+ #include <ATen/ops/indices_copy_compositeexplicitautogradnonfunctional_dispatch.h>
153
+ #include <ATen/ops/isin_compositeexplicitautogradnonfunctional_dispatch.h>
154
+ #include <ATen/ops/isneginf_compositeexplicitautogradnonfunctional_dispatch.h>
155
+ #include <ATen/ops/isposinf_compositeexplicitautogradnonfunctional_dispatch.h>
156
+ #include <ATen/ops/lcm_compositeexplicitautogradnonfunctional_dispatch.h>
157
+ #include <ATen/ops/le_compositeexplicitautogradnonfunctional_dispatch.h>
158
+ #include <ATen/ops/leaky_relu_compositeexplicitautogradnonfunctional_dispatch.h>
159
+ #include <ATen/ops/leaky_relu_backward_compositeexplicitautogradnonfunctional_dispatch.h>
160
+ #include <ATen/ops/lerp_compositeexplicitautogradnonfunctional_dispatch.h>
161
+ #include <ATen/ops/lgamma_compositeexplicitautogradnonfunctional_dispatch.h>
162
+ #include <ATen/ops/lift_fresh_copy_compositeexplicitautogradnonfunctional_dispatch.h>
163
+ #include <ATen/ops/linalg_cholesky_ex_compositeexplicitautogradnonfunctional_dispatch.h>
164
+ #include <ATen/ops/linalg_cross_compositeexplicitautogradnonfunctional_dispatch.h>
165
+ #include <ATen/ops/linalg_inv_ex_compositeexplicitautogradnonfunctional_dispatch.h>
166
+ #include <ATen/ops/linalg_ldl_factor_ex_compositeexplicitautogradnonfunctional_dispatch.h>
167
+ #include <ATen/ops/linalg_ldl_solve_compositeexplicitautogradnonfunctional_dispatch.h>
168
+ #include <ATen/ops/linalg_lu_compositeexplicitautogradnonfunctional_dispatch.h>
169
+ #include <ATen/ops/linalg_lu_factor_ex_compositeexplicitautogradnonfunctional_dispatch.h>
170
+ #include <ATen/ops/linalg_lu_solve_compositeexplicitautogradnonfunctional_dispatch.h>
171
+ #include <ATen/ops/linalg_pinv_compositeexplicitautogradnonfunctional_dispatch.h>
172
+ #include <ATen/ops/linalg_qr_compositeexplicitautogradnonfunctional_dispatch.h>
173
+ #include <ATen/ops/linalg_vector_norm_compositeexplicitautogradnonfunctional_dispatch.h>
174
+ #include <ATen/ops/log_compositeexplicitautogradnonfunctional_dispatch.h>
175
+ #include <ATen/ops/log10_compositeexplicitautogradnonfunctional_dispatch.h>
176
+ #include <ATen/ops/log1p_compositeexplicitautogradnonfunctional_dispatch.h>
177
+ #include <ATen/ops/log2_compositeexplicitautogradnonfunctional_dispatch.h>
178
+ #include <ATen/ops/logaddexp_compositeexplicitautogradnonfunctional_dispatch.h>
179
+ #include <ATen/ops/logaddexp2_compositeexplicitautogradnonfunctional_dispatch.h>
180
+ #include <ATen/ops/logit_backward_compositeexplicitautogradnonfunctional_dispatch.h>
181
+ #include <ATen/ops/logsumexp_compositeexplicitautogradnonfunctional_dispatch.h>
182
+ #include <ATen/ops/lt_compositeexplicitautogradnonfunctional_dispatch.h>
183
+ #include <ATen/ops/lu_unpack_compositeexplicitautogradnonfunctional_dispatch.h>
184
+ #include <ATen/ops/max_compositeexplicitautogradnonfunctional_dispatch.h>
185
+ #include <ATen/ops/max_pool2d_with_indices_compositeexplicitautogradnonfunctional_dispatch.h>
186
+ #include <ATen/ops/max_pool2d_with_indices_backward_compositeexplicitautogradnonfunctional_dispatch.h>
187
+ #include <ATen/ops/maximum_compositeexplicitautogradnonfunctional_dispatch.h>
188
+ #include <ATen/ops/mean_compositeexplicitautogradnonfunctional_dispatch.h>
189
+ #include <ATen/ops/min_compositeexplicitautogradnonfunctional_dispatch.h>
190
+ #include <ATen/ops/minimum_compositeexplicitautogradnonfunctional_dispatch.h>
191
+ #include <ATen/ops/mish_compositeexplicitautogradnonfunctional_dispatch.h>
192
+ #include <ATen/ops/mm_compositeexplicitautogradnonfunctional_dispatch.h>
193
+ #include <ATen/ops/mse_loss_compositeexplicitautogradnonfunctional_dispatch.h>
194
+ #include <ATen/ops/mul_compositeexplicitautogradnonfunctional_dispatch.h>
195
+ #include <ATen/ops/narrow_copy_compositeexplicitautogradnonfunctional_dispatch.h>
196
+ #include <ATen/ops/ne_compositeexplicitautogradnonfunctional_dispatch.h>
197
+ #include <ATen/ops/neg_compositeexplicitautogradnonfunctional_dispatch.h>
198
+ #include <ATen/ops/new_empty_strided_compositeexplicitautogradnonfunctional_dispatch.h>
199
+ #include <ATen/ops/nextafter_compositeexplicitautogradnonfunctional_dispatch.h>
200
+ #include <ATen/ops/nll_loss_backward_compositeexplicitautogradnonfunctional_dispatch.h>
201
+ #include <ATen/ops/nll_loss_forward_compositeexplicitautogradnonfunctional_dispatch.h>
202
+ #include <ATen/ops/norm_compositeexplicitautogradnonfunctional_dispatch.h>
203
+ #include <ATen/ops/permute_copy_compositeexplicitautogradnonfunctional_dispatch.h>
204
+ #include <ATen/ops/pixel_shuffle_compositeexplicitautogradnonfunctional_dispatch.h>
205
+ #include <ATen/ops/pixel_unshuffle_compositeexplicitautogradnonfunctional_dispatch.h>
206
+ #include <ATen/ops/polygamma_compositeexplicitautogradnonfunctional_dispatch.h>
207
+ #include <ATen/ops/pow_compositeexplicitautogradnonfunctional_dispatch.h>
208
+ #include <ATen/ops/prod_compositeexplicitautogradnonfunctional_dispatch.h>
209
+ #include <ATen/ops/reciprocal_compositeexplicitautogradnonfunctional_dispatch.h>
210
+ #include <ATen/ops/reflection_pad1d_compositeexplicitautogradnonfunctional_dispatch.h>
211
+ #include <ATen/ops/reflection_pad1d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
212
+ #include <ATen/ops/reflection_pad3d_compositeexplicitautogradnonfunctional_dispatch.h>
213
+ #include <ATen/ops/reflection_pad3d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
214
+ #include <ATen/ops/remainder_compositeexplicitautogradnonfunctional_dispatch.h>
215
+ #include <ATen/ops/renorm_compositeexplicitautogradnonfunctional_dispatch.h>
216
+ #include <ATen/ops/replication_pad1d_compositeexplicitautogradnonfunctional_dispatch.h>
217
+ #include <ATen/ops/replication_pad1d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
218
+ #include <ATen/ops/replication_pad2d_compositeexplicitautogradnonfunctional_dispatch.h>
219
+ #include <ATen/ops/replication_pad3d_compositeexplicitautogradnonfunctional_dispatch.h>
220
+ #include <ATen/ops/round_compositeexplicitautogradnonfunctional_dispatch.h>
221
+ #include <ATen/ops/row_indices_copy_compositeexplicitautogradnonfunctional_dispatch.h>
222
+ #include <ATen/ops/rsqrt_compositeexplicitautogradnonfunctional_dispatch.h>
223
+ #include <ATen/ops/scatter_compositeexplicitautogradnonfunctional_dispatch.h>
224
+ #include <ATen/ops/scatter_add_compositeexplicitautogradnonfunctional_dispatch.h>
225
+ #include <ATen/ops/scatter_reduce_compositeexplicitautogradnonfunctional_dispatch.h>
226
+ #include <ATen/ops/select_backward_compositeexplicitautogradnonfunctional_dispatch.h>
227
+ #include <ATen/ops/select_copy_compositeexplicitautogradnonfunctional_dispatch.h>
228
+ #include <ATen/ops/select_scatter_compositeexplicitautogradnonfunctional_dispatch.h>
229
+ #include <ATen/ops/sgn_compositeexplicitautogradnonfunctional_dispatch.h>
230
+ #include <ATen/ops/sigmoid_compositeexplicitautogradnonfunctional_dispatch.h>
231
+ #include <ATen/ops/sigmoid_backward_compositeexplicitautogradnonfunctional_dispatch.h>
232
+ #include <ATen/ops/sign_compositeexplicitautogradnonfunctional_dispatch.h>
233
+ #include <ATen/ops/signbit_compositeexplicitautogradnonfunctional_dispatch.h>
234
+ #include <ATen/ops/silu_compositeexplicitautogradnonfunctional_dispatch.h>
235
+ #include <ATen/ops/silu_backward_compositeexplicitautogradnonfunctional_dispatch.h>
236
+ #include <ATen/ops/sin_compositeexplicitautogradnonfunctional_dispatch.h>
237
+ #include <ATen/ops/sinc_compositeexplicitautogradnonfunctional_dispatch.h>
238
+ #include <ATen/ops/sinh_compositeexplicitautogradnonfunctional_dispatch.h>
239
+ #include <ATen/ops/slice_copy_compositeexplicitautogradnonfunctional_dispatch.h>
240
+ #include <ATen/ops/slice_scatter_compositeexplicitautogradnonfunctional_dispatch.h>
241
+ #include <ATen/ops/slow_conv_transpose2d_compositeexplicitautogradnonfunctional_dispatch.h>
242
+ #include <ATen/ops/smooth_l1_loss_compositeexplicitautogradnonfunctional_dispatch.h>
243
+ #include <ATen/ops/softplus_compositeexplicitautogradnonfunctional_dispatch.h>
244
+ #include <ATen/ops/softplus_backward_compositeexplicitautogradnonfunctional_dispatch.h>
245
+ #include <ATen/ops/softshrink_compositeexplicitautogradnonfunctional_dispatch.h>
246
+ #include <ATen/ops/softshrink_backward_compositeexplicitautogradnonfunctional_dispatch.h>
247
+ #include <ATen/ops/sort_compositeexplicitautogradnonfunctional_dispatch.h>
248
+ #include <ATen/ops/special_airy_ai_compositeexplicitautogradnonfunctional_dispatch.h>
249
+ #include <ATen/ops/special_bessel_j0_compositeexplicitautogradnonfunctional_dispatch.h>
250
+ #include <ATen/ops/special_bessel_j1_compositeexplicitautogradnonfunctional_dispatch.h>
251
+ #include <ATen/ops/special_bessel_y0_compositeexplicitautogradnonfunctional_dispatch.h>
252
+ #include <ATen/ops/special_bessel_y1_compositeexplicitautogradnonfunctional_dispatch.h>
253
+ #include <ATen/ops/special_chebyshev_polynomial_t_compositeexplicitautogradnonfunctional_dispatch.h>
254
+ #include <ATen/ops/special_chebyshev_polynomial_u_compositeexplicitautogradnonfunctional_dispatch.h>
255
+ #include <ATen/ops/special_chebyshev_polynomial_v_compositeexplicitautogradnonfunctional_dispatch.h>
256
+ #include <ATen/ops/special_chebyshev_polynomial_w_compositeexplicitautogradnonfunctional_dispatch.h>
257
+ #include <ATen/ops/special_entr_compositeexplicitautogradnonfunctional_dispatch.h>
258
+ #include <ATen/ops/special_erfcx_compositeexplicitautogradnonfunctional_dispatch.h>
259
+ #include <ATen/ops/special_hermite_polynomial_h_compositeexplicitautogradnonfunctional_dispatch.h>
260
+ #include <ATen/ops/special_hermite_polynomial_he_compositeexplicitautogradnonfunctional_dispatch.h>
261
+ #include <ATen/ops/special_i0e_compositeexplicitautogradnonfunctional_dispatch.h>
262
+ #include <ATen/ops/special_i1_compositeexplicitautogradnonfunctional_dispatch.h>
263
+ #include <ATen/ops/special_i1e_compositeexplicitautogradnonfunctional_dispatch.h>
264
+ #include <ATen/ops/special_laguerre_polynomial_l_compositeexplicitautogradnonfunctional_dispatch.h>
265
+ #include <ATen/ops/special_legendre_polynomial_p_compositeexplicitautogradnonfunctional_dispatch.h>
266
+ #include <ATen/ops/special_log_ndtr_compositeexplicitautogradnonfunctional_dispatch.h>
267
+ #include <ATen/ops/special_modified_bessel_i0_compositeexplicitautogradnonfunctional_dispatch.h>
268
+ #include <ATen/ops/special_modified_bessel_i1_compositeexplicitautogradnonfunctional_dispatch.h>
269
+ #include <ATen/ops/special_modified_bessel_k0_compositeexplicitautogradnonfunctional_dispatch.h>
270
+ #include <ATen/ops/special_modified_bessel_k1_compositeexplicitautogradnonfunctional_dispatch.h>
271
+ #include <ATen/ops/special_ndtri_compositeexplicitautogradnonfunctional_dispatch.h>
272
+ #include <ATen/ops/special_scaled_modified_bessel_k0_compositeexplicitautogradnonfunctional_dispatch.h>
273
+ #include <ATen/ops/special_scaled_modified_bessel_k1_compositeexplicitautogradnonfunctional_dispatch.h>
274
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_t_compositeexplicitautogradnonfunctional_dispatch.h>
275
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_u_compositeexplicitautogradnonfunctional_dispatch.h>
276
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_v_compositeexplicitautogradnonfunctional_dispatch.h>
277
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_w_compositeexplicitautogradnonfunctional_dispatch.h>
278
+ #include <ATen/ops/special_spherical_bessel_j0_compositeexplicitautogradnonfunctional_dispatch.h>
279
+ #include <ATen/ops/special_xlog1py_compositeexplicitautogradnonfunctional_dispatch.h>
280
+ #include <ATen/ops/special_zeta_compositeexplicitautogradnonfunctional_dispatch.h>
281
+ #include <ATen/ops/split_copy_compositeexplicitautogradnonfunctional_dispatch.h>
282
+ #include <ATen/ops/split_with_sizes_copy_compositeexplicitautogradnonfunctional_dispatch.h>
283
+ #include <ATen/ops/sqrt_compositeexplicitautogradnonfunctional_dispatch.h>
284
+ #include <ATen/ops/squeeze_copy_compositeexplicitautogradnonfunctional_dispatch.h>
285
+ #include <ATen/ops/sub_compositeexplicitautogradnonfunctional_dispatch.h>
286
+ #include <ATen/ops/sum_compositeexplicitautogradnonfunctional_dispatch.h>
287
+ #include <ATen/ops/t_copy_compositeexplicitautogradnonfunctional_dispatch.h>
288
+ #include <ATen/ops/tan_compositeexplicitautogradnonfunctional_dispatch.h>
289
+ #include <ATen/ops/tanh_compositeexplicitautogradnonfunctional_dispatch.h>
290
+ #include <ATen/ops/tanh_backward_compositeexplicitautogradnonfunctional_dispatch.h>
291
+ #include <ATen/ops/threshold_compositeexplicitautogradnonfunctional_dispatch.h>
292
+ #include <ATen/ops/threshold_backward_compositeexplicitautogradnonfunctional_dispatch.h>
293
+ #include <ATen/ops/topk_compositeexplicitautogradnonfunctional_dispatch.h>
294
+ #include <ATen/ops/transpose_copy_compositeexplicitautogradnonfunctional_dispatch.h>
295
+ #include <ATen/ops/triangular_solve_compositeexplicitautogradnonfunctional_dispatch.h>
296
+ #include <ATen/ops/tril_compositeexplicitautogradnonfunctional_dispatch.h>
297
+ #include <ATen/ops/triu_compositeexplicitautogradnonfunctional_dispatch.h>
298
+ #include <ATen/ops/trunc_compositeexplicitautogradnonfunctional_dispatch.h>
299
+ #include <ATen/ops/unbind_copy_compositeexplicitautogradnonfunctional_dispatch.h>
300
+ #include <ATen/ops/unfold_copy_compositeexplicitautogradnonfunctional_dispatch.h>
301
+ #include <ATen/ops/unsqueeze_copy_compositeexplicitautogradnonfunctional_dispatch.h>
302
+ #include <ATen/ops/upsample_bicubic2d_compositeexplicitautogradnonfunctional_dispatch.h>
303
+ #include <ATen/ops/upsample_bicubic2d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
304
+ #include <ATen/ops/upsample_bilinear2d_compositeexplicitautogradnonfunctional_dispatch.h>
305
+ #include <ATen/ops/upsample_bilinear2d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
306
+ #include <ATen/ops/upsample_linear1d_compositeexplicitautogradnonfunctional_dispatch.h>
307
+ #include <ATen/ops/upsample_linear1d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
308
+ #include <ATen/ops/upsample_nearest1d_compositeexplicitautogradnonfunctional_dispatch.h>
309
+ #include <ATen/ops/upsample_nearest1d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
310
+ #include <ATen/ops/upsample_nearest2d_compositeexplicitautogradnonfunctional_dispatch.h>
311
+ #include <ATen/ops/upsample_nearest2d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
312
+ #include <ATen/ops/upsample_nearest3d_compositeexplicitautogradnonfunctional_dispatch.h>
313
+ #include <ATen/ops/upsample_nearest3d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
314
+ #include <ATen/ops/upsample_trilinear3d_compositeexplicitautogradnonfunctional_dispatch.h>
315
+ #include <ATen/ops/upsample_trilinear3d_backward_compositeexplicitautogradnonfunctional_dispatch.h>
316
+ #include <ATen/ops/values_copy_compositeexplicitautogradnonfunctional_dispatch.h>
317
+ #include <ATen/ops/view_as_complex_copy_compositeexplicitautogradnonfunctional_dispatch.h>
318
+ #include <ATen/ops/view_as_real_copy_compositeexplicitautogradnonfunctional_dispatch.h>
319
+ #include <ATen/ops/view_copy_compositeexplicitautogradnonfunctional_dispatch.h>
320
+ #include <ATen/ops/xlogy_compositeexplicitautogradnonfunctional_dispatch.h>
321
+
322
+
323
+
.venv/lib/python3.11/site-packages/torch/include/ATen/CompositeImplicitAutogradFunctions_inl.h ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+ // @generated by torchgen/gen.py from DispatchKeyFunctions_inl.h
3
+
4
+ // NB: The implementing C++ file is RegisterDispatchKey.cpp
5
+
6
+ // The only #includes we need are for custom classes that have defaults in the C++ API
7
+ #include <c10/core/MemoryFormat.h>
8
+ #include <c10/core/Scalar.h>
9
+ #include <ATen/core/Reduction.h>
10
+
11
+ #if defined(AT_PER_OPERATOR_HEADERS) && defined(TORCH_ASSERT_ONLY_METHOD_OPERATORS)
12
+ #error This change adds a dependency on all pytorch operators, meaning the \
13
+ file will need to be re-compiled every time an operator is changed or added. \
14
+ Consider including a specific operator from \
15
+ <ATen/ops/{my_operator}_compositeimplicitautograd_dispatch.h>. \
16
+ See NOTE [TORCH_ASSERT_ONLY_METHOD_OPERATORS].
17
+ #endif
18
+
19
+ #include <ATen/ops/_add_batch_dim_compositeimplicitautograd_dispatch.h>
20
+ #include <ATen/ops/_assert_tensor_metadata_compositeimplicitautograd_dispatch.h>
21
+ #include <ATen/ops/_autocast_to_full_precision_compositeimplicitautograd_dispatch.h>
22
+ #include <ATen/ops/_autocast_to_reduced_precision_compositeimplicitautograd_dispatch.h>
23
+ #include <ATen/ops/_backward_compositeimplicitautograd_dispatch.h>
24
+ #include <ATen/ops/_batch_norm_impl_index_compositeimplicitautograd_dispatch.h>
25
+ #include <ATen/ops/_batch_norm_impl_index_backward_compositeimplicitautograd_dispatch.h>
26
+ #include <ATen/ops/_cast_Byte_compositeimplicitautograd_dispatch.h>
27
+ #include <ATen/ops/_cast_Char_compositeimplicitautograd_dispatch.h>
28
+ #include <ATen/ops/_cast_Double_compositeimplicitautograd_dispatch.h>
29
+ #include <ATen/ops/_cast_Float_compositeimplicitautograd_dispatch.h>
30
+ #include <ATen/ops/_cast_Half_compositeimplicitautograd_dispatch.h>
31
+ #include <ATen/ops/_cast_Int_compositeimplicitautograd_dispatch.h>
32
+ #include <ATen/ops/_cast_Long_compositeimplicitautograd_dispatch.h>
33
+ #include <ATen/ops/_cast_Short_compositeimplicitautograd_dispatch.h>
34
+ #include <ATen/ops/_choose_qparams_per_tensor_compositeimplicitautograd_dispatch.h>
35
+ #include <ATen/ops/_convolution_compositeimplicitautograd_dispatch.h>
36
+ #include <ATen/ops/_convolution_double_backward_compositeimplicitautograd_dispatch.h>
37
+ #include <ATen/ops/_convolution_mode_compositeimplicitautograd_dispatch.h>
38
+ #include <ATen/ops/_cufft_clear_plan_cache_compositeimplicitautograd_dispatch.h>
39
+ #include <ATen/ops/_cufft_get_plan_cache_max_size_compositeimplicitautograd_dispatch.h>
40
+ #include <ATen/ops/_cufft_get_plan_cache_size_compositeimplicitautograd_dispatch.h>
41
+ #include <ATen/ops/_cufft_set_plan_cache_max_size_compositeimplicitautograd_dispatch.h>
42
+ #include <ATen/ops/_debug_has_internal_overlap_compositeimplicitautograd_dispatch.h>
43
+ #include <ATen/ops/_dim_arange_compositeimplicitautograd_dispatch.h>
44
+ #include <ATen/ops/_embedding_bag_sparse_backward_compositeimplicitautograd_dispatch.h>
45
+ #include <ATen/ops/_gather_sparse_backward_compositeimplicitautograd_dispatch.h>
46
+ #include <ATen/ops/_grid_sampler_2d_cpu_fallback_backward_compositeimplicitautograd_dispatch.h>
47
+ #include <ATen/ops/_has_compatible_shallow_copy_type_compositeimplicitautograd_dispatch.h>
48
+ #include <ATen/ops/_is_zerotensor_compositeimplicitautograd_dispatch.h>
49
+ #include <ATen/ops/_lu_with_info_compositeimplicitautograd_dispatch.h>
50
+ #include <ATen/ops/_nnpack_available_compositeimplicitautograd_dispatch.h>
51
+ #include <ATen/ops/_pack_padded_sequence_backward_compositeimplicitautograd_dispatch.h>
52
+ #include <ATen/ops/_pad_circular_compositeimplicitautograd_dispatch.h>
53
+ #include <ATen/ops/_pad_enum_compositeimplicitautograd_dispatch.h>
54
+ #include <ATen/ops/_pad_packed_sequence_compositeimplicitautograd_dispatch.h>
55
+ #include <ATen/ops/_propagate_xla_data_compositeimplicitautograd_dispatch.h>
56
+ #include <ATen/ops/_remove_batch_dim_compositeimplicitautograd_dispatch.h>
57
+ #include <ATen/ops/_reshape_from_tensor_compositeimplicitautograd_dispatch.h>
58
+ #include <ATen/ops/_rowwise_prune_compositeimplicitautograd_dispatch.h>
59
+ #include <ATen/ops/_saturate_weight_to_fp16_compositeimplicitautograd_dispatch.h>
60
+ #include <ATen/ops/_scaled_dot_product_attention_math_compositeimplicitautograd_dispatch.h>
61
+ #include <ATen/ops/_shape_as_tensor_compositeimplicitautograd_dispatch.h>
62
+ #include <ATen/ops/_sobol_engine_draw_compositeimplicitautograd_dispatch.h>
63
+ #include <ATen/ops/_sobol_engine_ff_compositeimplicitautograd_dispatch.h>
64
+ #include <ATen/ops/_sobol_engine_initialize_state_compositeimplicitautograd_dispatch.h>
65
+ #include <ATen/ops/_sobol_engine_scramble_compositeimplicitautograd_dispatch.h>
66
+ #include <ATen/ops/_sparse_bsc_tensor_unsafe_compositeimplicitautograd_dispatch.h>
67
+ #include <ATen/ops/_sparse_bsr_tensor_unsafe_compositeimplicitautograd_dispatch.h>
68
+ #include <ATen/ops/_sparse_compressed_tensor_unsafe_compositeimplicitautograd_dispatch.h>
69
+ #include <ATen/ops/_sparse_coo_tensor_unsafe_compositeimplicitautograd_dispatch.h>
70
+ #include <ATen/ops/_sparse_csc_tensor_unsafe_compositeimplicitautograd_dispatch.h>
71
+ #include <ATen/ops/_sparse_csr_tensor_unsafe_compositeimplicitautograd_dispatch.h>
72
+ #include <ATen/ops/_sparse_log_softmax_compositeimplicitautograd_dispatch.h>
73
+ #include <ATen/ops/_sparse_mm_compositeimplicitautograd_dispatch.h>
74
+ #include <ATen/ops/_sparse_softmax_compositeimplicitautograd_dispatch.h>
75
+ #include <ATen/ops/_sparse_sum_compositeimplicitautograd_dispatch.h>
76
+ #include <ATen/ops/_test_ambiguous_defaults_compositeimplicitautograd_dispatch.h>
77
+ #include <ATen/ops/_test_autograd_multiple_dispatch_compositeimplicitautograd_dispatch.h>
78
+ #include <ATen/ops/_test_check_tensor_compositeimplicitautograd_dispatch.h>
79
+ #include <ATen/ops/_test_serialization_subcmul_compositeimplicitautograd_dispatch.h>
80
+ #include <ATen/ops/_test_string_default_compositeimplicitautograd_dispatch.h>
81
+ #include <ATen/ops/_thnn_differentiable_gru_cell_backward_compositeimplicitautograd_dispatch.h>
82
+ #include <ATen/ops/_thnn_differentiable_lstm_cell_backward_compositeimplicitautograd_dispatch.h>
83
+ #include <ATen/ops/_thnn_fused_lstm_cell_backward_compositeimplicitautograd_dispatch.h>
84
+ #include <ATen/ops/_to_cpu_compositeimplicitautograd_dispatch.h>
85
+ #include <ATen/ops/_unpack_dual_compositeimplicitautograd_dispatch.h>
86
+ #include <ATen/ops/_upsample_bicubic2d_aa_compositeimplicitautograd_dispatch.h>
87
+ #include <ATen/ops/_upsample_bilinear2d_aa_compositeimplicitautograd_dispatch.h>
88
+ #include <ATen/ops/_upsample_nearest_exact1d_compositeimplicitautograd_dispatch.h>
89
+ #include <ATen/ops/_upsample_nearest_exact2d_compositeimplicitautograd_dispatch.h>
90
+ #include <ATen/ops/_upsample_nearest_exact3d_compositeimplicitautograd_dispatch.h>
91
+ #include <ATen/ops/_use_cudnn_rnn_flatten_weight_compositeimplicitautograd_dispatch.h>
92
+ #include <ATen/ops/_validate_sparse_bsc_tensor_args_compositeimplicitautograd_dispatch.h>
93
+ #include <ATen/ops/_validate_sparse_bsr_tensor_args_compositeimplicitautograd_dispatch.h>
94
+ #include <ATen/ops/_validate_sparse_compressed_tensor_args_compositeimplicitautograd_dispatch.h>
95
+ #include <ATen/ops/_validate_sparse_coo_tensor_args_compositeimplicitautograd_dispatch.h>
96
+ #include <ATen/ops/_validate_sparse_csc_tensor_args_compositeimplicitautograd_dispatch.h>
97
+ #include <ATen/ops/_validate_sparse_csr_tensor_args_compositeimplicitautograd_dispatch.h>
98
+ #include <ATen/ops/_version_compositeimplicitautograd_dispatch.h>
99
+ #include <ATen/ops/_weight_norm_compositeimplicitautograd_dispatch.h>
100
+ #include <ATen/ops/_weight_norm_differentiable_backward_compositeimplicitautograd_dispatch.h>
101
+ #include <ATen/ops/_wrapped_linear_prepack_compositeimplicitautograd_dispatch.h>
102
+ #include <ATen/ops/_wrapped_quantized_linear_prepacked_compositeimplicitautograd_dispatch.h>
103
+ #include <ATen/ops/absolute_compositeimplicitautograd_dispatch.h>
104
+ #include <ATen/ops/adaptive_avg_pool1d_compositeimplicitautograd_dispatch.h>
105
+ #include <ATen/ops/adaptive_avg_pool2d_compositeimplicitautograd_dispatch.h>
106
+ #include <ATen/ops/adaptive_avg_pool3d_compositeimplicitautograd_dispatch.h>
107
+ #include <ATen/ops/adaptive_max_pool1d_compositeimplicitautograd_dispatch.h>
108
+ #include <ATen/ops/adjoint_compositeimplicitautograd_dispatch.h>
109
+ #include <ATen/ops/affine_grid_generator_backward_compositeimplicitautograd_dispatch.h>
110
+ #include <ATen/ops/align_as_compositeimplicitautograd_dispatch.h>
111
+ #include <ATen/ops/align_tensors_compositeimplicitautograd_dispatch.h>
112
+ #include <ATen/ops/align_to_compositeimplicitautograd_dispatch.h>
113
+ #include <ATen/ops/all_compositeimplicitautograd_dispatch.h>
114
+ #include <ATen/ops/alpha_dropout_compositeimplicitautograd_dispatch.h>
115
+ #include <ATen/ops/and_compositeimplicitautograd_dispatch.h>
116
+ #include <ATen/ops/any_compositeimplicitautograd_dispatch.h>
117
+ #include <ATen/ops/arccos_compositeimplicitautograd_dispatch.h>
118
+ #include <ATen/ops/arccosh_compositeimplicitautograd_dispatch.h>
119
+ #include <ATen/ops/arcsin_compositeimplicitautograd_dispatch.h>
120
+ #include <ATen/ops/arcsinh_compositeimplicitautograd_dispatch.h>
121
+ #include <ATen/ops/arctan_compositeimplicitautograd_dispatch.h>
122
+ #include <ATen/ops/arctan2_compositeimplicitautograd_dispatch.h>
123
+ #include <ATen/ops/arctanh_compositeimplicitautograd_dispatch.h>
124
+ #include <ATen/ops/argsort_compositeimplicitautograd_dispatch.h>
125
+ #include <ATen/ops/argwhere_compositeimplicitautograd_dispatch.h>
126
+ #include <ATen/ops/atleast_1d_compositeimplicitautograd_dispatch.h>
127
+ #include <ATen/ops/atleast_2d_compositeimplicitautograd_dispatch.h>
128
+ #include <ATen/ops/atleast_3d_compositeimplicitautograd_dispatch.h>
129
+ #include <ATen/ops/avg_pool1d_compositeimplicitautograd_dispatch.h>
130
+ #include <ATen/ops/batch_norm_compositeimplicitautograd_dispatch.h>
131
+ #include <ATen/ops/bilinear_compositeimplicitautograd_dispatch.h>
132
+ #include <ATen/ops/broadcast_tensors_compositeimplicitautograd_dispatch.h>
133
+ #include <ATen/ops/broadcast_to_compositeimplicitautograd_dispatch.h>
134
+ #include <ATen/ops/can_cast_compositeimplicitautograd_dispatch.h>
135
+ #include <ATen/ops/cartesian_prod_compositeimplicitautograd_dispatch.h>
136
+ #include <ATen/ops/cat_compositeimplicitautograd_dispatch.h>
137
+ #include <ATen/ops/cdist_compositeimplicitautograd_dispatch.h>
138
+ #include <ATen/ops/chain_matmul_compositeimplicitautograd_dispatch.h>
139
+ #include <ATen/ops/chalf_compositeimplicitautograd_dispatch.h>
140
+ #include <ATen/ops/choose_qparams_optimized_compositeimplicitautograd_dispatch.h>
141
+ #include <ATen/ops/chunk_compositeimplicitautograd_dispatch.h>
142
+ #include <ATen/ops/clip_compositeimplicitautograd_dispatch.h>
143
+ #include <ATen/ops/coalesce_compositeimplicitautograd_dispatch.h>
144
+ #include <ATen/ops/column_stack_compositeimplicitautograd_dispatch.h>
145
+ #include <ATen/ops/combinations_compositeimplicitautograd_dispatch.h>
146
+ #include <ATen/ops/concat_compositeimplicitautograd_dispatch.h>
147
+ #include <ATen/ops/concatenate_compositeimplicitautograd_dispatch.h>
148
+ #include <ATen/ops/conj_compositeimplicitautograd_dispatch.h>
149
+ #include <ATen/ops/conj_physical_compositeimplicitautograd_dispatch.h>
150
+ #include <ATen/ops/contiguous_compositeimplicitautograd_dispatch.h>
151
+ #include <ATen/ops/conv1d_compositeimplicitautograd_dispatch.h>
152
+ #include <ATen/ops/conv2d_compositeimplicitautograd_dispatch.h>
153
+ #include <ATen/ops/conv3d_compositeimplicitautograd_dispatch.h>
154
+ #include <ATen/ops/conv_tbc_backward_compositeimplicitautograd_dispatch.h>
155
+ #include <ATen/ops/conv_transpose1d_compositeimplicitautograd_dispatch.h>
156
+ #include <ATen/ops/conv_transpose2d_compositeimplicitautograd_dispatch.h>
157
+ #include <ATen/ops/conv_transpose3d_compositeimplicitautograd_dispatch.h>
158
+ #include <ATen/ops/corrcoef_compositeimplicitautograd_dispatch.h>
159
+ #include <ATen/ops/cosine_embedding_loss_compositeimplicitautograd_dispatch.h>
160
+ #include <ATen/ops/cosine_similarity_compositeimplicitautograd_dispatch.h>
161
+ #include <ATen/ops/cov_compositeimplicitautograd_dispatch.h>
162
+ #include <ATen/ops/cross_compositeimplicitautograd_dispatch.h>
163
+ #include <ATen/ops/cross_entropy_loss_compositeimplicitautograd_dispatch.h>
164
+ #include <ATen/ops/ctc_loss_compositeimplicitautograd_dispatch.h>
165
+ #include <ATen/ops/cudnn_is_acceptable_compositeimplicitautograd_dispatch.h>
166
+ #include <ATen/ops/cummax_compositeimplicitautograd_dispatch.h>
167
+ #include <ATen/ops/cummaxmin_backward_compositeimplicitautograd_dispatch.h>
168
+ #include <ATen/ops/cummin_compositeimplicitautograd_dispatch.h>
169
+ #include <ATen/ops/cumprod_compositeimplicitautograd_dispatch.h>
170
+ #include <ATen/ops/cumprod_backward_compositeimplicitautograd_dispatch.h>
171
+ #include <ATen/ops/cumsum_compositeimplicitautograd_dispatch.h>
172
+ #include <ATen/ops/cumulative_trapezoid_compositeimplicitautograd_dispatch.h>
173
+ #include <ATen/ops/data_compositeimplicitautograd_dispatch.h>
174
+ #include <ATen/ops/det_compositeimplicitautograd_dispatch.h>
175
+ #include <ATen/ops/diag_compositeimplicitautograd_dispatch.h>
176
+ #include <ATen/ops/diagflat_compositeimplicitautograd_dispatch.h>
177
+ #include <ATen/ops/diagonal_compositeimplicitautograd_dispatch.h>
178
+ #include <ATen/ops/diff_compositeimplicitautograd_dispatch.h>
179
+ #include <ATen/ops/divide_compositeimplicitautograd_dispatch.h>
180
+ #include <ATen/ops/dropout_compositeimplicitautograd_dispatch.h>
181
+ #include <ATen/ops/dsplit_compositeimplicitautograd_dispatch.h>
182
+ #include <ATen/ops/dstack_compositeimplicitautograd_dispatch.h>
183
+ #include <ATen/ops/einsum_compositeimplicitautograd_dispatch.h>
184
+ #include <ATen/ops/embedding_backward_compositeimplicitautograd_dispatch.h>
185
+ #include <ATen/ops/embedding_bag_compositeimplicitautograd_dispatch.h>
186
+ #include <ATen/ops/embedding_sparse_backward_compositeimplicitautograd_dispatch.h>
187
+ #include <ATen/ops/empty_compositeimplicitautograd_dispatch.h>
188
+ #include <ATen/ops/expand_as_compositeimplicitautograd_dispatch.h>
189
+ #include <ATen/ops/fake_quantize_per_channel_affine_compositeimplicitautograd_dispatch.h>
190
+ #include <ATen/ops/fake_quantize_per_channel_affine_cachemask_backward_compositeimplicitautograd_dispatch.h>
191
+ #include <ATen/ops/fake_quantize_per_tensor_affine_compositeimplicitautograd_dispatch.h>
192
+ #include <ATen/ops/fake_quantize_per_tensor_affine_cachemask_backward_compositeimplicitautograd_dispatch.h>
193
+ #include <ATen/ops/fbgemm_linear_fp16_weight_compositeimplicitautograd_dispatch.h>
194
+ #include <ATen/ops/fbgemm_linear_fp16_weight_fp32_activation_compositeimplicitautograd_dispatch.h>
195
+ #include <ATen/ops/fbgemm_linear_int8_weight_compositeimplicitautograd_dispatch.h>
196
+ #include <ATen/ops/fbgemm_linear_int8_weight_fp32_activation_compositeimplicitautograd_dispatch.h>
197
+ #include <ATen/ops/fbgemm_linear_quantize_weight_compositeimplicitautograd_dispatch.h>
198
+ #include <ATen/ops/fbgemm_pack_gemm_matrix_fp16_compositeimplicitautograd_dispatch.h>
199
+ #include <ATen/ops/fbgemm_pack_quantized_matrix_compositeimplicitautograd_dispatch.h>
200
+ #include <ATen/ops/feature_alpha_dropout_compositeimplicitautograd_dispatch.h>
201
+ #include <ATen/ops/feature_dropout_compositeimplicitautograd_dispatch.h>
202
+ #include <ATen/ops/fft_fft_compositeimplicitautograd_dispatch.h>
203
+ #include <ATen/ops/fft_fft2_compositeimplicitautograd_dispatch.h>
204
+ #include <ATen/ops/fft_fftn_compositeimplicitautograd_dispatch.h>
205
+ #include <ATen/ops/fft_fftshift_compositeimplicitautograd_dispatch.h>
206
+ #include <ATen/ops/fft_hfft_compositeimplicitautograd_dispatch.h>
207
+ #include <ATen/ops/fft_hfft2_compositeimplicitautograd_dispatch.h>
208
+ #include <ATen/ops/fft_hfftn_compositeimplicitautograd_dispatch.h>
209
+ #include <ATen/ops/fft_ifft_compositeimplicitautograd_dispatch.h>
210
+ #include <ATen/ops/fft_ifft2_compositeimplicitautograd_dispatch.h>
211
+ #include <ATen/ops/fft_ifftn_compositeimplicitautograd_dispatch.h>
212
+ #include <ATen/ops/fft_ifftshift_compositeimplicitautograd_dispatch.h>
213
+ #include <ATen/ops/fft_ihfft_compositeimplicitautograd_dispatch.h>
214
+ #include <ATen/ops/fft_ihfft2_compositeimplicitautograd_dispatch.h>
215
+ #include <ATen/ops/fft_ihfftn_compositeimplicitautograd_dispatch.h>
216
+ #include <ATen/ops/fft_irfft_compositeimplicitautograd_dispatch.h>
217
+ #include <ATen/ops/fft_irfft2_compositeimplicitautograd_dispatch.h>
218
+ #include <ATen/ops/fft_irfftn_compositeimplicitautograd_dispatch.h>
219
+ #include <ATen/ops/fft_rfft_compositeimplicitautograd_dispatch.h>
220
+ #include <ATen/ops/fft_rfft2_compositeimplicitautograd_dispatch.h>
221
+ #include <ATen/ops/fft_rfftn_compositeimplicitautograd_dispatch.h>
222
+ #include <ATen/ops/fill_diagonal_compositeimplicitautograd_dispatch.h>
223
+ #include <ATen/ops/fix_compositeimplicitautograd_dispatch.h>
224
+ #include <ATen/ops/flatten_compositeimplicitautograd_dispatch.h>
225
+ #include <ATen/ops/flatten_dense_tensors_compositeimplicitautograd_dispatch.h>
226
+ #include <ATen/ops/fliplr_compositeimplicitautograd_dispatch.h>
227
+ #include <ATen/ops/flipud_compositeimplicitautograd_dispatch.h>
228
+ #include <ATen/ops/float_power_compositeimplicitautograd_dispatch.h>
229
+ #include <ATen/ops/frobenius_norm_compositeimplicitautograd_dispatch.h>
230
+ #include <ATen/ops/fused_moving_avg_obs_fake_quant_compositeimplicitautograd_dispatch.h>
231
+ #include <ATen/ops/gather_compositeimplicitautograd_dispatch.h>
232
+ #include <ATen/ops/gather_backward_compositeimplicitautograd_dispatch.h>
233
+ #include <ATen/ops/ger_compositeimplicitautograd_dispatch.h>
234
+ #include <ATen/ops/gradient_compositeimplicitautograd_dispatch.h>
235
+ #include <ATen/ops/greater_compositeimplicitautograd_dispatch.h>
236
+ #include <ATen/ops/greater_equal_compositeimplicitautograd_dispatch.h>
237
+ #include <ATen/ops/grid_sampler_compositeimplicitautograd_dispatch.h>
238
+ #include <ATen/ops/group_norm_compositeimplicitautograd_dispatch.h>
239
+ #include <ATen/ops/gru_compositeimplicitautograd_dispatch.h>
240
+ #include <ATen/ops/gru_cell_compositeimplicitautograd_dispatch.h>
241
+ #include <ATen/ops/hinge_embedding_loss_compositeimplicitautograd_dispatch.h>
242
+ #include <ATen/ops/histogramdd_compositeimplicitautograd_dispatch.h>
243
+ #include <ATen/ops/hsplit_compositeimplicitautograd_dispatch.h>
244
+ #include <ATen/ops/hstack_compositeimplicitautograd_dispatch.h>
245
+ #include <ATen/ops/imag_compositeimplicitautograd_dispatch.h>
246
+ #include <ATen/ops/index_add_compositeimplicitautograd_dispatch.h>
247
+ #include <ATen/ops/index_copy_compositeimplicitautograd_dispatch.h>
248
+ #include <ATen/ops/index_fill_compositeimplicitautograd_dispatch.h>
249
+ #include <ATen/ops/index_select_compositeimplicitautograd_dispatch.h>
250
+ #include <ATen/ops/index_select_backward_compositeimplicitautograd_dispatch.h>
251
+ #include <ATen/ops/infinitely_differentiable_gelu_backward_compositeimplicitautograd_dispatch.h>
252
+ #include <ATen/ops/inner_compositeimplicitautograd_dispatch.h>
253
+ #include <ATen/ops/instance_norm_compositeimplicitautograd_dispatch.h>
254
+ #include <ATen/ops/inverse_compositeimplicitautograd_dispatch.h>
255
+ #include <ATen/ops/is_complex_compositeimplicitautograd_dispatch.h>
256
+ #include <ATen/ops/is_conj_compositeimplicitautograd_dispatch.h>
257
+ #include <ATen/ops/is_distributed_compositeimplicitautograd_dispatch.h>
258
+ #include <ATen/ops/is_floating_point_compositeimplicitautograd_dispatch.h>
259
+ #include <ATen/ops/is_inference_compositeimplicitautograd_dispatch.h>
260
+ #include <ATen/ops/is_leaf_compositeimplicitautograd_dispatch.h>
261
+ #include <ATen/ops/is_neg_compositeimplicitautograd_dispatch.h>
262
+ #include <ATen/ops/is_nonzero_compositeimplicitautograd_dispatch.h>
263
+ #include <ATen/ops/is_signed_compositeimplicitautograd_dispatch.h>
264
+ #include <ATen/ops/is_vulkan_available_compositeimplicitautograd_dispatch.h>
265
+ #include <ATen/ops/isclose_compositeimplicitautograd_dispatch.h>
266
+ #include <ATen/ops/isfinite_compositeimplicitautograd_dispatch.h>
267
+ #include <ATen/ops/isreal_compositeimplicitautograd_dispatch.h>
268
+ #include <ATen/ops/istft_compositeimplicitautograd_dispatch.h>
269
+ #include <ATen/ops/item_compositeimplicitautograd_dispatch.h>
270
+ #include <ATen/ops/kl_div_compositeimplicitautograd_dispatch.h>
271
+ #include <ATen/ops/kron_compositeimplicitautograd_dispatch.h>
272
+ #include <ATen/ops/kthvalue_compositeimplicitautograd_dispatch.h>
273
+ #include <ATen/ops/l1_loss_compositeimplicitautograd_dispatch.h>
274
+ #include <ATen/ops/layer_norm_compositeimplicitautograd_dispatch.h>
275
+ #include <ATen/ops/ldexp_compositeimplicitautograd_dispatch.h>
276
+ #include <ATen/ops/less_compositeimplicitautograd_dispatch.h>
277
+ #include <ATen/ops/less_equal_compositeimplicitautograd_dispatch.h>
278
+ #include <ATen/ops/linalg_cholesky_compositeimplicitautograd_dispatch.h>
279
+ #include <ATen/ops/linalg_cond_compositeimplicitautograd_dispatch.h>
280
+ #include <ATen/ops/linalg_det_compositeimplicitautograd_dispatch.h>
281
+ #include <ATen/ops/linalg_diagonal_compositeimplicitautograd_dispatch.h>
282
+ #include <ATen/ops/linalg_eigh_compositeimplicitautograd_dispatch.h>
283
+ #include <ATen/ops/linalg_eigvals_compositeimplicitautograd_dispatch.h>
284
+ #include <ATen/ops/linalg_eigvalsh_compositeimplicitautograd_dispatch.h>
285
+ #include <ATen/ops/linalg_inv_compositeimplicitautograd_dispatch.h>
286
+ #include <ATen/ops/linalg_ldl_factor_compositeimplicitautograd_dispatch.h>
287
+ #include <ATen/ops/linalg_lu_factor_compositeimplicitautograd_dispatch.h>
288
+ #include <ATen/ops/linalg_matmul_compositeimplicitautograd_dispatch.h>
289
+ #include <ATen/ops/linalg_matrix_norm_compositeimplicitautograd_dispatch.h>
290
+ #include <ATen/ops/linalg_matrix_power_compositeimplicitautograd_dispatch.h>
291
+ #include <ATen/ops/linalg_matrix_rank_compositeimplicitautograd_dispatch.h>
292
+ #include <ATen/ops/linalg_multi_dot_compositeimplicitautograd_dispatch.h>
293
+ #include <ATen/ops/linalg_norm_compositeimplicitautograd_dispatch.h>
294
+ #include <ATen/ops/linalg_pinv_compositeimplicitautograd_dispatch.h>
295
+ #include <ATen/ops/linalg_slogdet_compositeimplicitautograd_dispatch.h>
296
+ #include <ATen/ops/linalg_solve_compositeimplicitautograd_dispatch.h>
297
+ #include <ATen/ops/linalg_solve_ex_compositeimplicitautograd_dispatch.h>
298
+ #include <ATen/ops/linalg_svd_compositeimplicitautograd_dispatch.h>
299
+ #include <ATen/ops/linalg_svdvals_compositeimplicitautograd_dispatch.h>
300
+ #include <ATen/ops/linalg_tensorinv_compositeimplicitautograd_dispatch.h>
301
+ #include <ATen/ops/linalg_tensorsolve_compositeimplicitautograd_dispatch.h>
302
+ #include <ATen/ops/linalg_vander_compositeimplicitautograd_dispatch.h>
303
+ #include <ATen/ops/linalg_vecdot_compositeimplicitautograd_dispatch.h>
304
+ #include <ATen/ops/linear_compositeimplicitautograd_dispatch.h>
305
+ #include <ATen/ops/log_sigmoid_compositeimplicitautograd_dispatch.h>
306
+ #include <ATen/ops/log_softmax_compositeimplicitautograd_dispatch.h>
307
+ #include <ATen/ops/logcumsumexp_compositeimplicitautograd_dispatch.h>
308
+ #include <ATen/ops/logdet_compositeimplicitautograd_dispatch.h>
309
+ #include <ATen/ops/logsumexp_compositeimplicitautograd_dispatch.h>
310
+ #include <ATen/ops/lstm_compositeimplicitautograd_dispatch.h>
311
+ #include <ATen/ops/lstm_cell_compositeimplicitautograd_dispatch.h>
312
+ #include <ATen/ops/lu_solve_compositeimplicitautograd_dispatch.h>
313
+ #include <ATen/ops/mH_compositeimplicitautograd_dispatch.h>
314
+ #include <ATen/ops/mT_compositeimplicitautograd_dispatch.h>
315
+ #include <ATen/ops/margin_ranking_loss_compositeimplicitautograd_dispatch.h>
316
+ #include <ATen/ops/masked_select_backward_compositeimplicitautograd_dispatch.h>
317
+ #include <ATen/ops/matmul_compositeimplicitautograd_dispatch.h>
318
+ #include <ATen/ops/matrix_H_compositeimplicitautograd_dispatch.h>
319
+ #include <ATen/ops/matrix_exp_compositeimplicitautograd_dispatch.h>
320
+ #include <ATen/ops/matrix_exp_backward_compositeimplicitautograd_dispatch.h>
321
+ #include <ATen/ops/matrix_power_compositeimplicitautograd_dispatch.h>
322
+ #include <ATen/ops/max_compositeimplicitautograd_dispatch.h>
323
+ #include <ATen/ops/max_pool1d_compositeimplicitautograd_dispatch.h>
324
+ #include <ATen/ops/max_pool1d_with_indices_compositeimplicitautograd_dispatch.h>
325
+ #include <ATen/ops/max_pool2d_compositeimplicitautograd_dispatch.h>
326
+ #include <ATen/ops/max_pool3d_compositeimplicitautograd_dispatch.h>
327
+ #include <ATen/ops/mean_compositeimplicitautograd_dispatch.h>
328
+ #include <ATen/ops/median_compositeimplicitautograd_dispatch.h>
329
+ #include <ATen/ops/meshgrid_compositeimplicitautograd_dispatch.h>
330
+ #include <ATen/ops/min_compositeimplicitautograd_dispatch.h>
331
+ #include <ATen/ops/mish_backward_compositeimplicitautograd_dispatch.h>
332
+ #include <ATen/ops/mode_compositeimplicitautograd_dispatch.h>
333
+ #include <ATen/ops/moveaxis_compositeimplicitautograd_dispatch.h>
334
+ #include <ATen/ops/movedim_compositeimplicitautograd_dispatch.h>
335
+ #include <ATen/ops/msort_compositeimplicitautograd_dispatch.h>
336
+ #include <ATen/ops/multilabel_margin_loss_compositeimplicitautograd_dispatch.h>
337
+ #include <ATen/ops/multiply_compositeimplicitautograd_dispatch.h>
338
+ #include <ATen/ops/nanmean_compositeimplicitautograd_dispatch.h>
339
+ #include <ATen/ops/nanmedian_compositeimplicitautograd_dispatch.h>
340
+ #include <ATen/ops/nanquantile_compositeimplicitautograd_dispatch.h>
341
+ #include <ATen/ops/narrow_compositeimplicitautograd_dispatch.h>
342
+ #include <ATen/ops/native_channel_shuffle_compositeimplicitautograd_dispatch.h>
343
+ #include <ATen/ops/negative_compositeimplicitautograd_dispatch.h>
344
+ #include <ATen/ops/nested_to_padded_tensor_compositeimplicitautograd_dispatch.h>
345
+ #include <ATen/ops/nll_loss_compositeimplicitautograd_dispatch.h>
346
+ #include <ATen/ops/nll_loss2d_compositeimplicitautograd_dispatch.h>
347
+ #include <ATen/ops/nll_loss_nd_compositeimplicitautograd_dispatch.h>
348
+ #include <ATen/ops/nonzero_numpy_compositeimplicitautograd_dispatch.h>
349
+ #include <ATen/ops/norm_compositeimplicitautograd_dispatch.h>
350
+ #include <ATen/ops/norm_except_dim_compositeimplicitautograd_dispatch.h>
351
+ #include <ATen/ops/not_equal_compositeimplicitautograd_dispatch.h>
352
+ #include <ATen/ops/nuclear_norm_compositeimplicitautograd_dispatch.h>
353
+ #include <ATen/ops/numpy_T_compositeimplicitautograd_dispatch.h>
354
+ #include <ATen/ops/one_hot_compositeimplicitautograd_dispatch.h>
355
+ #include <ATen/ops/or_compositeimplicitautograd_dispatch.h>
356
+ #include <ATen/ops/orgqr_compositeimplicitautograd_dispatch.h>
357
+ #include <ATen/ops/outer_compositeimplicitautograd_dispatch.h>
358
+ #include <ATen/ops/output_nr_compositeimplicitautograd_dispatch.h>
359
+ #include <ATen/ops/pad_compositeimplicitautograd_dispatch.h>
360
+ #include <ATen/ops/pad_sequence_compositeimplicitautograd_dispatch.h>
361
+ #include <ATen/ops/pairwise_distance_compositeimplicitautograd_dispatch.h>
362
+ #include <ATen/ops/pdist_compositeimplicitautograd_dispatch.h>
363
+ #include <ATen/ops/pin_memory_compositeimplicitautograd_dispatch.h>
364
+ #include <ATen/ops/pinverse_compositeimplicitautograd_dispatch.h>
365
+ #include <ATen/ops/poisson_nll_loss_compositeimplicitautograd_dispatch.h>
366
+ #include <ATen/ops/positive_compositeimplicitautograd_dispatch.h>
367
+ #include <ATen/ops/prelu_compositeimplicitautograd_dispatch.h>
368
+ #include <ATen/ops/prod_compositeimplicitautograd_dispatch.h>
369
+ #include <ATen/ops/promote_types_compositeimplicitautograd_dispatch.h>
370
+ #include <ATen/ops/qr_compositeimplicitautograd_dispatch.h>
371
+ #include <ATen/ops/quantile_compositeimplicitautograd_dispatch.h>
372
+ #include <ATen/ops/quantized_gru_cell_compositeimplicitautograd_dispatch.h>
373
+ #include <ATen/ops/quantized_lstm_cell_compositeimplicitautograd_dispatch.h>
374
+ #include <ATen/ops/quantized_rnn_relu_cell_compositeimplicitautograd_dispatch.h>
375
+ #include <ATen/ops/quantized_rnn_tanh_cell_compositeimplicitautograd_dispatch.h>
376
+ #include <ATen/ops/rand_compositeimplicitautograd_dispatch.h>
377
+ #include <ATen/ops/randn_compositeimplicitautograd_dispatch.h>
378
+ #include <ATen/ops/ravel_compositeimplicitautograd_dispatch.h>
379
+ #include <ATen/ops/real_compositeimplicitautograd_dispatch.h>
380
+ #include <ATen/ops/refine_names_compositeimplicitautograd_dispatch.h>
381
+ #include <ATen/ops/relu6_compositeimplicitautograd_dispatch.h>
382
+ #include <ATen/ops/rename_compositeimplicitautograd_dispatch.h>
383
+ #include <ATen/ops/repeat_interleave_compositeimplicitautograd_dispatch.h>
384
+ #include <ATen/ops/requires_grad_compositeimplicitautograd_dispatch.h>
385
+ #include <ATen/ops/reshape_compositeimplicitautograd_dispatch.h>
386
+ #include <ATen/ops/reshape_as_compositeimplicitautograd_dispatch.h>
387
+ #include <ATen/ops/resolve_conj_compositeimplicitautograd_dispatch.h>
388
+ #include <ATen/ops/resolve_neg_compositeimplicitautograd_dispatch.h>
389
+ #include <ATen/ops/result_type_compositeimplicitautograd_dispatch.h>
390
+ #include <ATen/ops/retain_grad_compositeimplicitautograd_dispatch.h>
391
+ #include <ATen/ops/retains_grad_compositeimplicitautograd_dispatch.h>
392
+ #include <ATen/ops/rms_norm_compositeimplicitautograd_dispatch.h>
393
+ #include <ATen/ops/rnn_relu_compositeimplicitautograd_dispatch.h>
394
+ #include <ATen/ops/rnn_relu_cell_compositeimplicitautograd_dispatch.h>
395
+ #include <ATen/ops/rnn_tanh_compositeimplicitautograd_dispatch.h>
396
+ #include <ATen/ops/rnn_tanh_cell_compositeimplicitautograd_dispatch.h>
397
+ #include <ATen/ops/row_stack_compositeimplicitautograd_dispatch.h>
398
+ #include <ATen/ops/rrelu_compositeimplicitautograd_dispatch.h>
399
+ #include <ATen/ops/scaled_dot_product_attention_compositeimplicitautograd_dispatch.h>
400
+ #include <ATen/ops/scatter_compositeimplicitautograd_dispatch.h>
401
+ #include <ATen/ops/scatter_add_compositeimplicitautograd_dispatch.h>
402
+ #include <ATen/ops/select_compositeimplicitautograd_dispatch.h>
403
+ #include <ATen/ops/selu_compositeimplicitautograd_dispatch.h>
404
+ #include <ATen/ops/set_compositeimplicitautograd_dispatch.h>
405
+ #include <ATen/ops/set_data_compositeimplicitautograd_dispatch.h>
406
+ #include <ATen/ops/silu_backward_compositeimplicitautograd_dispatch.h>
407
+ #include <ATen/ops/size_compositeimplicitautograd_dispatch.h>
408
+ #include <ATen/ops/slogdet_compositeimplicitautograd_dispatch.h>
409
+ #include <ATen/ops/slow_conv3d_compositeimplicitautograd_dispatch.h>
410
+ #include <ATen/ops/smm_compositeimplicitautograd_dispatch.h>
411
+ #include <ATen/ops/softmax_compositeimplicitautograd_dispatch.h>
412
+ #include <ATen/ops/sort_compositeimplicitautograd_dispatch.h>
413
+ #include <ATen/ops/sparse_bsc_tensor_compositeimplicitautograd_dispatch.h>
414
+ #include <ATen/ops/sparse_bsr_tensor_compositeimplicitautograd_dispatch.h>
415
+ #include <ATen/ops/sparse_coo_tensor_compositeimplicitautograd_dispatch.h>
416
+ #include <ATen/ops/sparse_csc_tensor_compositeimplicitautograd_dispatch.h>
417
+ #include <ATen/ops/sparse_csr_tensor_compositeimplicitautograd_dispatch.h>
418
+ #include <ATen/ops/special_digamma_compositeimplicitautograd_dispatch.h>
419
+ #include <ATen/ops/special_erf_compositeimplicitautograd_dispatch.h>
420
+ #include <ATen/ops/special_erfc_compositeimplicitautograd_dispatch.h>
421
+ #include <ATen/ops/special_erfinv_compositeimplicitautograd_dispatch.h>
422
+ #include <ATen/ops/special_exp2_compositeimplicitautograd_dispatch.h>
423
+ #include <ATen/ops/special_expit_compositeimplicitautograd_dispatch.h>
424
+ #include <ATen/ops/special_expm1_compositeimplicitautograd_dispatch.h>
425
+ #include <ATen/ops/special_gammainc_compositeimplicitautograd_dispatch.h>
426
+ #include <ATen/ops/special_gammaincc_compositeimplicitautograd_dispatch.h>
427
+ #include <ATen/ops/special_gammaln_compositeimplicitautograd_dispatch.h>
428
+ #include <ATen/ops/special_i0_compositeimplicitautograd_dispatch.h>
429
+ #include <ATen/ops/special_log1p_compositeimplicitautograd_dispatch.h>
430
+ #include <ATen/ops/special_log_softmax_compositeimplicitautograd_dispatch.h>
431
+ #include <ATen/ops/special_logit_compositeimplicitautograd_dispatch.h>
432
+ #include <ATen/ops/special_logsumexp_compositeimplicitautograd_dispatch.h>
433
+ #include <ATen/ops/special_multigammaln_compositeimplicitautograd_dispatch.h>
434
+ #include <ATen/ops/special_ndtr_compositeimplicitautograd_dispatch.h>
435
+ #include <ATen/ops/special_polygamma_compositeimplicitautograd_dispatch.h>
436
+ #include <ATen/ops/special_psi_compositeimplicitautograd_dispatch.h>
437
+ #include <ATen/ops/special_round_compositeimplicitautograd_dispatch.h>
438
+ #include <ATen/ops/special_sinc_compositeimplicitautograd_dispatch.h>
439
+ #include <ATen/ops/special_softmax_compositeimplicitautograd_dispatch.h>
440
+ #include <ATen/ops/special_xlogy_compositeimplicitautograd_dispatch.h>
441
+ #include <ATen/ops/split_compositeimplicitautograd_dispatch.h>
442
+ #include <ATen/ops/square_compositeimplicitautograd_dispatch.h>
443
+ #include <ATen/ops/squeeze_compositeimplicitautograd_dispatch.h>
444
+ #include <ATen/ops/sspaddmm_compositeimplicitautograd_dispatch.h>
445
+ #include <ATen/ops/std_compositeimplicitautograd_dispatch.h>
446
+ #include <ATen/ops/std_mean_compositeimplicitautograd_dispatch.h>
447
+ #include <ATen/ops/stft_compositeimplicitautograd_dispatch.h>
448
+ #include <ATen/ops/stride_compositeimplicitautograd_dispatch.h>
449
+ #include <ATen/ops/subtract_compositeimplicitautograd_dispatch.h>
450
+ #include <ATen/ops/sum_compositeimplicitautograd_dispatch.h>
451
+ #include <ATen/ops/sum_to_size_compositeimplicitautograd_dispatch.h>
452
+ #include <ATen/ops/svd_compositeimplicitautograd_dispatch.h>
453
+ #include <ATen/ops/swapaxes_compositeimplicitautograd_dispatch.h>
454
+ #include <ATen/ops/swapdims_compositeimplicitautograd_dispatch.h>
455
+ #include <ATen/ops/sym_numel_compositeimplicitautograd_dispatch.h>
456
+ #include <ATen/ops/sym_size_compositeimplicitautograd_dispatch.h>
457
+ #include <ATen/ops/sym_storage_offset_compositeimplicitautograd_dispatch.h>
458
+ #include <ATen/ops/sym_stride_compositeimplicitautograd_dispatch.h>
459
+ #include <ATen/ops/take_along_dim_compositeimplicitautograd_dispatch.h>
460
+ #include <ATen/ops/tensor_split_compositeimplicitautograd_dispatch.h>
461
+ #include <ATen/ops/tensordot_compositeimplicitautograd_dispatch.h>
462
+ #include <ATen/ops/thnn_conv2d_compositeimplicitautograd_dispatch.h>
463
+ #include <ATen/ops/tile_compositeimplicitautograd_dispatch.h>
464
+ #include <ATen/ops/to_compositeimplicitautograd_dispatch.h>
465
+ #include <ATen/ops/to_dense_compositeimplicitautograd_dispatch.h>
466
+ #include <ATen/ops/to_dense_backward_compositeimplicitautograd_dispatch.h>
467
+ #include <ATen/ops/to_mkldnn_backward_compositeimplicitautograd_dispatch.h>
468
+ #include <ATen/ops/to_sparse_compositeimplicitautograd_dispatch.h>
469
+ #include <ATen/ops/to_sparse_bsc_compositeimplicitautograd_dispatch.h>
470
+ #include <ATen/ops/to_sparse_bsr_compositeimplicitautograd_dispatch.h>
471
+ #include <ATen/ops/to_sparse_csc_compositeimplicitautograd_dispatch.h>
472
+ #include <ATen/ops/to_sparse_csr_compositeimplicitautograd_dispatch.h>
473
+ #include <ATen/ops/trace_backward_compositeimplicitautograd_dispatch.h>
474
+ #include <ATen/ops/transpose_compositeimplicitautograd_dispatch.h>
475
+ #include <ATen/ops/trapezoid_compositeimplicitautograd_dispatch.h>
476
+ #include <ATen/ops/trapz_compositeimplicitautograd_dispatch.h>
477
+ #include <ATen/ops/triplet_margin_loss_compositeimplicitautograd_dispatch.h>
478
+ #include <ATen/ops/true_divide_compositeimplicitautograd_dispatch.h>
479
+ #include <ATen/ops/type_as_compositeimplicitautograd_dispatch.h>
480
+ #include <ATen/ops/unbind_compositeimplicitautograd_dispatch.h>
481
+ #include <ATen/ops/unflatten_compositeimplicitautograd_dispatch.h>
482
+ #include <ATen/ops/unflatten_dense_tensors_compositeimplicitautograd_dispatch.h>
483
+ #include <ATen/ops/unsafe_chunk_compositeimplicitautograd_dispatch.h>
484
+ #include <ATen/ops/upsample_bicubic2d_compositeimplicitautograd_dispatch.h>
485
+ #include <ATen/ops/upsample_bilinear2d_compositeimplicitautograd_dispatch.h>
486
+ #include <ATen/ops/upsample_linear1d_compositeimplicitautograd_dispatch.h>
487
+ #include <ATen/ops/upsample_nearest1d_compositeimplicitautograd_dispatch.h>
488
+ #include <ATen/ops/upsample_nearest2d_compositeimplicitautograd_dispatch.h>
489
+ #include <ATen/ops/upsample_nearest3d_compositeimplicitautograd_dispatch.h>
490
+ #include <ATen/ops/upsample_trilinear3d_compositeimplicitautograd_dispatch.h>
491
+ #include <ATen/ops/value_selecting_reduction_backward_compositeimplicitautograd_dispatch.h>
492
+ #include <ATen/ops/vander_compositeimplicitautograd_dispatch.h>
493
+ #include <ATen/ops/var_compositeimplicitautograd_dispatch.h>
494
+ #include <ATen/ops/var_mean_compositeimplicitautograd_dispatch.h>
495
+ #include <ATen/ops/view_as_compositeimplicitautograd_dispatch.h>
496
+ #include <ATen/ops/vsplit_compositeimplicitautograd_dispatch.h>
497
+ #include <ATen/ops/vstack_compositeimplicitautograd_dispatch.h>
498
+ #include <ATen/ops/where_compositeimplicitautograd_dispatch.h>
499
+ #include <ATen/ops/xor_compositeimplicitautograd_dispatch.h>
500
+
501
+
502
+
.venv/lib/python3.11/site-packages/torch/include/ATen/DLConvertor.h ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <ATen/ATen.h>
4
+ #include <ATen/Tensor.h>
5
+ #include <ATen/dlpack.h>
6
+
7
+ // this convertor will:
8
+ // 1) take a Tensor object and wrap it in the DLPack tensor
9
+ // 2) take a dlpack tensor and convert it to the ATen Tensor
10
+
11
+ namespace at {
12
+
13
+ TORCH_API ScalarType toScalarType(const DLDataType& dtype);
14
+ TORCH_API DLManagedTensor* toDLPack(const Tensor& src);
15
+ TORCH_API Tensor fromDLPack(DLManagedTensor* src);
16
+ C10_DEPRECATED_MESSAGE("Please migrate to a non-const variant")
17
+ inline Tensor fromDLPack(const DLManagedTensor* src) {
18
+ return fromDLPack(const_cast<DLManagedTensor*>(src));
19
+ }
20
+ TORCH_API Tensor
21
+ fromDLPack(DLManagedTensor* src, std::function<void(void*)> deleter);
22
+ TORCH_API DLDataType getDLDataType(const Tensor& t);
23
+ TORCH_API DLDevice getDLContext(const Tensor& tensor, const int64_t& device_id);
24
+
25
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/DeviceGuard.h ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <ATen/core/IListRef.h>
4
+ #include <ATen/core/Tensor.h>
5
+ #include <c10/core/DeviceGuard.h>
6
+ #include <c10/core/ScalarType.h> // TensorList whyyyyy
7
+
8
+ namespace at {
9
+
10
+ // Are you here because you're wondering why DeviceGuard(tensor) no
11
+ // longer works? For code organization reasons, we have temporarily(?)
12
+ // removed this constructor from DeviceGuard. The new way to
13
+ // spell it is:
14
+ //
15
+ // OptionalDeviceGuard guard(device_of(tensor));
16
+
17
+ /// Return the Device of a Tensor, if the Tensor is defined.
18
+ inline std::optional<Device> device_of(const Tensor& t) {
19
+ if (t.defined()) {
20
+ return std::make_optional(t.device());
21
+ } else {
22
+ return std::nullopt;
23
+ }
24
+ }
25
+
26
+ inline std::optional<Device> device_of(const std::optional<Tensor>& t) {
27
+ return t.has_value() ? device_of(t.value()) : std::nullopt;
28
+ }
29
+
30
+ /// Return the Device of a TensorList, if the list is non-empty and
31
+ /// the first Tensor is defined. (This function implicitly assumes
32
+ /// that all tensors in the list have the same device.)
33
+ inline std::optional<Device> device_of(ITensorListRef t) {
34
+ if (!t.empty()) {
35
+ return device_of(t.front());
36
+ } else {
37
+ return std::nullopt;
38
+ }
39
+ }
40
+
41
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/Dispatch_v2.h ADDED
@@ -0,0 +1,186 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #include <ATen/Dispatch.h>
2
+
3
+ // This is a new implementation of the AT_DISPATCH macro family from
4
+ // ATen/Dispatch.h
5
+ //
6
+ // The intended usage is:
7
+ //
8
+ // ScalarType scalar_type;
9
+ //
10
+ // AT_DISPATCH_V2(
11
+ // scalar_type,
12
+ // "debug string",
13
+ // AT_WRAP([&] {
14
+ // ... code to specialize with scalar_t ...
15
+ // }),
16
+ // kHalf,
17
+ // AT_EXPAND(AT_ALL_TYPES),
18
+ // ... as many types arguments as needed ...
19
+ // )
20
+ //
21
+ // For example, given an old style:
22
+ //
23
+ // AT_DISPATCH_ALL_TYPES_AND_COMPLEX_AND2(
24
+ // kComplexHalf,
25
+ // kHalf,
26
+ // self.scalar_type(),
27
+ // "_local_scalar_dense_cpu",
28
+ // [&] {
29
+ // scalar_t value = *self.data_ptr<scalar_t>();
30
+ // r = Scalar(value);
31
+ // }
32
+ // )
33
+ //
34
+ // You now write:
35
+ //
36
+ // AT_DISPATCH_V2(
37
+ // self.scalar_type(),
38
+ // "_local_scalar_dense_cpu",
39
+ // AT_WRAP([&] {
40
+ // scalar_t value = *self.data_ptr<scalar_t>();
41
+ // r = Scalar(value);
42
+ // }),
43
+ // AT_EXPAND(AT_ALL_TYPES),
44
+ // AT_EXPAND(AT_COMPLEX_TYPES),
45
+ // kComplexHalf,
46
+ // kHalf,
47
+ // )
48
+ //
49
+ // Notably, it sports the following improvements:
50
+ //
51
+ // - It is not necessary to specify the arity (e.g.,
52
+ // AT_DISPATCH_FLOATING_AND_COMPLEX_TYPES_AND{2,3,4,...})
53
+ // when using the macro
54
+ //
55
+ // - It is not necessary to specify each dtype individually; if
56
+ // there is a set of related dtypes and you want to dispatch
57
+ // over all of them, you can simply say, e.g., AT_EXPAND(AT_INTEGRAL_TYPES)
58
+ // in your argument list.
59
+ //
60
+ // However, you must remember to wrap the payload body in AT_WRAP, or commas
61
+ // inside your lambda will be improperly handled. Furthermore, if you more
62
+ // entries to ScalarType than can be supported by this macro, it will fail
63
+ // with an obscure error (due to attempting to concatenate AT_AP with
64
+ // something that is not a number).
65
+ //
66
+ // The implementation strategy is to use the count arguments trick
67
+ // (e.g., as described in https://stackoverflow.com/a/2124385/23845)
68
+ // to discover how many dtypes have been passed, and then dispatch to a
69
+ // hand-written macro for each arity that applies as many DISPATCH_CASE as
70
+ // necessary. The hand-written macros can be regenerated for other arities
71
+ // with the script below.
72
+ //
73
+ // There is some delicacy in the implementation in controlling when
74
+ // macro expansion occurs, mediated with AT_EXPAND and AT_GUARD. I mostly
75
+ // relied on GPT4 to help me get it right.
76
+
77
+ // Public API macros
78
+
79
+ // See documentation above
80
+ #define AT_DISPATCH_V2(TYPE, NAME, BODY, ...) \
81
+ AT_DISPATCH_SWITCH(TYPE, NAME, AT_AP_VAR(AT_WRAP(BODY), TYPE, __VA_ARGS__))
82
+
83
+ // This macro lets you pass an arbitrary expression that may contain internal
84
+ // commas to another macro without having the commas causing the expression
85
+ // to be interpreted as being multiple arguments
86
+ #define AT_WRAP(...) __VA_ARGS__
87
+
88
+ #define AT_FLOAT8_TYPES \
89
+ c10::kFloat8_e5m2, c10::kFloat8_e5m2fnuz, c10::kFloat8_e4m3fn, \
90
+ c10::kFloat8_e4m3fnuz
91
+
92
+ #define AT_INTEGRAL_TYPES \
93
+ c10::kByte, c10::kChar, c10::kInt, c10::kLong, c10::kShort
94
+ #define AT_FLOATING_TYPES c10::kDouble, c10::kFloat
95
+ #define AT_BAREBONES_UNSIGNED_TYPES c10::kUInt16, c10::kUInt32, c10::kUInt64
96
+ #define AT_INTEGRAL_TYPES_V2 \
97
+ AT_EXPAND(AT_INTEGRAL_TYPES), AT_EXPAND(AT_BAREBONES_UNSIGNED_TYPES)
98
+ #define AT_COMPLEX_TYPES c10::kComplexDouble, c10::kComplexFloat
99
+ #define AT_QINT_TYPES c10::kQInt8, c10::kQUInt8, c10::kQInt32
100
+ // NB: not *actually* all types
101
+ #define AT_ALL_TYPES AT_EXPAND(AT_INTEGRAL_TYPES), AT_EXPAND(AT_FLOATING_TYPES)
102
+ #define AT_ALL_TYPES_AND_COMPLEX \
103
+ AT_EXPAND(AT_ALL_TYPES), AT_EXPAND(AT_COMPLEX_TYPES)
104
+
105
+ // Helper macros
106
+
107
+ #define AT_AP_VAR(N, T, ...) \
108
+ AT_EXPAND(AT_CONCAT(AT_AP, AT_NUM_ARGS(__VA_ARGS__))(AT_WRAP(N), __VA_ARGS__))
109
+ #define AT_CONCAT(a, b) AT_CONCAT_AUX(a, b)
110
+ #define AT_CONCAT_AUX(a, b) a##b
111
+ #define AT_EXPAND(X) X
112
+
113
+ // Ensure we never have too many scalar types for the expansion here to
114
+ // support. To bump this, you must regenerate the macros below.
115
+ static_assert(static_cast<int>(c10::ScalarType::NumOptions) < 45);
116
+
117
+ // Python code to regenerate generate code below:
118
+ #if 0
119
+
120
+ num_args = 45
121
+
122
+ nums = ', '.join(str(i) for i in reversed(range(num_args+1)))
123
+ args = ', '.join(f'_{i}' for i in range(1, num_args+1))
124
+
125
+ print(f'#define AT_NUM_ARGS(...) AT_EXPAND(AT_NUM_ARGS_AUX(__VA_ARGS__, {nums}))')
126
+ print(f'#define AT_NUM_ARGS_AUX({args}, N, ...) N')
127
+
128
+ for i in range(1, num_args+1):
129
+ args = ', '.join(f'_{i}' for i in range(1, i+1))
130
+ cases = ' '.join([f'AT_DISPATCH_CASE(_{j}, N)' for j in range(1, i+1)])
131
+ print(f'#define AT_AP{i}(N, {args}) {cases}')
132
+
133
+ #endif
134
+
135
+ // Begin generated code
136
+ // clang-format off
137
+
138
+ #define AT_NUM_ARGS(...) AT_EXPAND(AT_NUM_ARGS_AUX(__VA_ARGS__, 45, 44, 43, 42, 41, 40, 39, 38, 37, 36, 35, 34, 33, 32, 31, 30, 29, 28, 27, 26, 25, 24, 23, 22, 21, 20, 19, 18, 17, 16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0))
139
+ #define AT_NUM_ARGS_AUX(_1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35, _36, _37, _38, _39, _40, _41, _42, _43, _44, _45, N, ...) N
140
+ #define AT_AP1(N, _1) AT_DISPATCH_CASE(_1, N)
141
+ #define AT_AP2(N, _1, _2) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N)
142
+ #define AT_AP3(N, _1, _2, _3) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N)
143
+ #define AT_AP4(N, _1, _2, _3, _4) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N)
144
+ #define AT_AP5(N, _1, _2, _3, _4, _5) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N)
145
+ #define AT_AP6(N, _1, _2, _3, _4, _5, _6) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N)
146
+ #define AT_AP7(N, _1, _2, _3, _4, _5, _6, _7) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N)
147
+ #define AT_AP8(N, _1, _2, _3, _4, _5, _6, _7, _8) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N)
148
+ #define AT_AP9(N, _1, _2, _3, _4, _5, _6, _7, _8, _9) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N)
149
+ #define AT_AP10(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N)
150
+ #define AT_AP11(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N)
151
+ #define AT_AP12(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N)
152
+ #define AT_AP13(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N)
153
+ #define AT_AP14(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N)
154
+ #define AT_AP15(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N)
155
+ #define AT_AP16(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N)
156
+ #define AT_AP17(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N)
157
+ #define AT_AP18(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N)
158
+ #define AT_AP19(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N)
159
+ #define AT_AP20(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N)
160
+ #define AT_AP21(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N)
161
+ #define AT_AP22(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N)
162
+ #define AT_AP23(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N)
163
+ #define AT_AP24(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N)
164
+ #define AT_AP25(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N)
165
+ #define AT_AP26(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N)
166
+ #define AT_AP27(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N)
167
+ #define AT_AP28(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N)
168
+ #define AT_AP29(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N)
169
+ #define AT_AP30(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N)
170
+ #define AT_AP31(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N)
171
+ #define AT_AP32(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N)
172
+ #define AT_AP33(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N)
173
+ #define AT_AP34(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N)
174
+ #define AT_AP35(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N) AT_DISPATCH_CASE(_35, N)
175
+ #define AT_AP36(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35, _36) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N) AT_DISPATCH_CASE(_35, N) AT_DISPATCH_CASE(_36, N)
176
+ #define AT_AP37(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35, _36, _37) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N) AT_DISPATCH_CASE(_35, N) AT_DISPATCH_CASE(_36, N) AT_DISPATCH_CASE(_37, N)
177
+ #define AT_AP38(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35, _36, _37, _38) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N) AT_DISPATCH_CASE(_35, N) AT_DISPATCH_CASE(_36, N) AT_DISPATCH_CASE(_37, N) AT_DISPATCH_CASE(_38, N)
178
+ #define AT_AP39(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35, _36, _37, _38, _39) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N) AT_DISPATCH_CASE(_35, N) AT_DISPATCH_CASE(_36, N) AT_DISPATCH_CASE(_37, N) AT_DISPATCH_CASE(_38, N) AT_DISPATCH_CASE(_39, N)
179
+ #define AT_AP40(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35, _36, _37, _38, _39, _40) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N) AT_DISPATCH_CASE(_35, N) AT_DISPATCH_CASE(_36, N) AT_DISPATCH_CASE(_37, N) AT_DISPATCH_CASE(_38, N) AT_DISPATCH_CASE(_39, N) AT_DISPATCH_CASE(_40, N)
180
+ #define AT_AP41(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35, _36, _37, _38, _39, _40, _41) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N) AT_DISPATCH_CASE(_35, N) AT_DISPATCH_CASE(_36, N) AT_DISPATCH_CASE(_37, N) AT_DISPATCH_CASE(_38, N) AT_DISPATCH_CASE(_39, N) AT_DISPATCH_CASE(_40, N) AT_DISPATCH_CASE(_41, N)
181
+ #define AT_AP42(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35, _36, _37, _38, _39, _40, _41, _42) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N) AT_DISPATCH_CASE(_35, N) AT_DISPATCH_CASE(_36, N) AT_DISPATCH_CASE(_37, N) AT_DISPATCH_CASE(_38, N) AT_DISPATCH_CASE(_39, N) AT_DISPATCH_CASE(_40, N) AT_DISPATCH_CASE(_41, N) AT_DISPATCH_CASE(_42, N)
182
+ #define AT_AP43(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35, _36, _37, _38, _39, _40, _41, _42, _43) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N) AT_DISPATCH_CASE(_35, N) AT_DISPATCH_CASE(_36, N) AT_DISPATCH_CASE(_37, N) AT_DISPATCH_CASE(_38, N) AT_DISPATCH_CASE(_39, N) AT_DISPATCH_CASE(_40, N) AT_DISPATCH_CASE(_41, N) AT_DISPATCH_CASE(_42, N) AT_DISPATCH_CASE(_43, N)
183
+ #define AT_AP44(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35, _36, _37, _38, _39, _40, _41, _42, _43, _44) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N) AT_DISPATCH_CASE(_35, N) AT_DISPATCH_CASE(_36, N) AT_DISPATCH_CASE(_37, N) AT_DISPATCH_CASE(_38, N) AT_DISPATCH_CASE(_39, N) AT_DISPATCH_CASE(_40, N) AT_DISPATCH_CASE(_41, N) AT_DISPATCH_CASE(_42, N) AT_DISPATCH_CASE(_43, N) AT_DISPATCH_CASE(_44, N)
184
+ #define AT_AP45(N, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _13, _14, _15, _16, _17, _18, _19, _20, _21, _22, _23, _24, _25, _26, _27, _28, _29, _30, _31, _32, _33, _34, _35, _36, _37, _38, _39, _40, _41, _42, _43, _44, _45) AT_DISPATCH_CASE(_1, N) AT_DISPATCH_CASE(_2, N) AT_DISPATCH_CASE(_3, N) AT_DISPATCH_CASE(_4, N) AT_DISPATCH_CASE(_5, N) AT_DISPATCH_CASE(_6, N) AT_DISPATCH_CASE(_7, N) AT_DISPATCH_CASE(_8, N) AT_DISPATCH_CASE(_9, N) AT_DISPATCH_CASE(_10, N) AT_DISPATCH_CASE(_11, N) AT_DISPATCH_CASE(_12, N) AT_DISPATCH_CASE(_13, N) AT_DISPATCH_CASE(_14, N) AT_DISPATCH_CASE(_15, N) AT_DISPATCH_CASE(_16, N) AT_DISPATCH_CASE(_17, N) AT_DISPATCH_CASE(_18, N) AT_DISPATCH_CASE(_19, N) AT_DISPATCH_CASE(_20, N) AT_DISPATCH_CASE(_21, N) AT_DISPATCH_CASE(_22, N) AT_DISPATCH_CASE(_23, N) AT_DISPATCH_CASE(_24, N) AT_DISPATCH_CASE(_25, N) AT_DISPATCH_CASE(_26, N) AT_DISPATCH_CASE(_27, N) AT_DISPATCH_CASE(_28, N) AT_DISPATCH_CASE(_29, N) AT_DISPATCH_CASE(_30, N) AT_DISPATCH_CASE(_31, N) AT_DISPATCH_CASE(_32, N) AT_DISPATCH_CASE(_33, N) AT_DISPATCH_CASE(_34, N) AT_DISPATCH_CASE(_35, N) AT_DISPATCH_CASE(_36, N) AT_DISPATCH_CASE(_37, N) AT_DISPATCH_CASE(_38, N) AT_DISPATCH_CASE(_39, N) AT_DISPATCH_CASE(_40, N) AT_DISPATCH_CASE(_41, N) AT_DISPATCH_CASE(_42, N) AT_DISPATCH_CASE(_43, N) AT_DISPATCH_CASE(_44, N) AT_DISPATCH_CASE(_45, N)
185
+ // End generated code
186
+ // clang-format on
.venv/lib/python3.11/site-packages/torch/include/ATen/DynamicLibrary.h ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <ATen/Utils.h>
4
+ #include <c10/macros/Export.h>
5
+ #include <c10/util/Exception.h>
6
+
7
+ namespace c10 {
8
+
9
+ class DynamicLibraryError : public Error {
10
+ using Error::Error;
11
+ };
12
+
13
+ } // namespace c10
14
+
15
+ namespace at {
16
+
17
+ struct DynamicLibrary {
18
+ AT_DISALLOW_COPY_AND_ASSIGN(DynamicLibrary);
19
+
20
+ TORCH_API DynamicLibrary(
21
+ const char* name,
22
+ const char* alt_name = nullptr,
23
+ bool leak_handle = false);
24
+
25
+ TORCH_API void* sym(const char* name);
26
+
27
+ TORCH_API ~DynamicLibrary();
28
+
29
+ private:
30
+ bool leak_handle;
31
+ void* handle = nullptr;
32
+ };
33
+
34
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/ExpandUtils.h ADDED
@@ -0,0 +1,527 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #ifndef AT_PER_OPERATOR_HEADERS
4
+ #include <ATen/Functions.h>
5
+ #else
6
+ #include <ATen/ops/view.h>
7
+ #include <ATen/ops/view_copy.h>
8
+ #endif
9
+
10
+ #include <ATen/Tensor.h>
11
+ #include <ATen/core/DimVector.h>
12
+ #include <c10/util/Exception.h>
13
+ #include <c10/util/MaybeOwned.h>
14
+ #include <c10/util/irange.h>
15
+
16
+ #include <functional>
17
+ #include <tuple>
18
+ #include <utility>
19
+
20
+ namespace at {
21
+
22
+ TORCH_API std::vector<int64_t> infer_size(IntArrayRef a, IntArrayRef b);
23
+ TORCH_API std::vector<SymInt> infer_size_symint(
24
+ SymIntArrayRef a,
25
+ SymIntArrayRef b);
26
+ TORCH_API DimVector infer_size_dimvector(IntArrayRef a, IntArrayRef b);
27
+ TORCH_API SymDimVector
28
+ infer_size_symdimvector(SymIntArrayRef a, SymIntArrayRef b);
29
+
30
+ // Named type instead of a pair/tuple so that we can be sure to
31
+ // construct the vectors in place and get NRVO.
32
+ template <typename Container>
33
+ struct InferExpandGeometryResult {
34
+ Container sizes;
35
+ Container strides;
36
+ explicit InferExpandGeometryResult(size_t ndim)
37
+ : sizes(ndim), strides(ndim) {}
38
+ explicit InferExpandGeometryResult(IntArrayRef sizes_, size_t ndim)
39
+ : sizes(sizes_.begin(), sizes_.end()), strides(ndim) {}
40
+ };
41
+
42
+ TORCH_API std::tuple<std::vector<int64_t>, std::vector<int64_t>>
43
+ inferExpandGeometry(
44
+ IntArrayRef tensor_sizes,
45
+ IntArrayRef tensor_strides,
46
+ IntArrayRef sizes);
47
+
48
+ TORCH_API InferExpandGeometryResult<DimVector> inferExpandGeometry_dimvector(
49
+ IntArrayRef tensor_sizes,
50
+ IntArrayRef tensor_strides,
51
+ IntArrayRef sizes);
52
+
53
+ TORCH_API std::vector<int64_t> infer_dense_strides(
54
+ IntArrayRef tensor_sizes,
55
+ IntArrayRef tensor_strides);
56
+
57
+ // True if input shapes are expandable
58
+ // NOTE: infer_size did a similar check, please keep them sync if change is
59
+ // needed
60
+ inline bool are_expandable(IntArrayRef shape1, IntArrayRef shape2) {
61
+ size_t ndim1 = shape1.size();
62
+ size_t ndim2 = shape2.size();
63
+ size_t ndim = ndim1 < ndim2 ? ndim1 : ndim2;
64
+
65
+ for (int64_t i = static_cast<int64_t>(ndim) - 1; i >= 0; --i) {
66
+ if (shape1[--ndim1] == shape2[--ndim2] || shape1[ndim1] == 1 ||
67
+ shape2[ndim2] == 1) {
68
+ continue;
69
+ }
70
+ return false;
71
+ }
72
+ return true;
73
+ }
74
+
75
+ // avoid copy-construction of Tensor by using a reference_wrapper.
76
+ inline void check_defined(
77
+ std::initializer_list<std::reference_wrapper<const Tensor>> tensors,
78
+ const char* api_name) {
79
+ for (auto& t : tensors) {
80
+ if (!t.get().defined()) {
81
+ AT_ERROR(api_name, "(...) called with an undefined Tensor");
82
+ }
83
+ }
84
+ }
85
+
86
+ // NOTE [ ExpandUtils Borrowing ]
87
+ //
88
+ // Functions in ExpandUtils return `c10::MaybeOwned<Tensor>` because
89
+ // expansion may not actually be needed, in which case we can improve
90
+ // efficiency by returning
91
+ // `c10::MaybeOwned<Tensor>::borrowed(to_expand)`. However, this means
92
+ // that you need to be careful: the returned `c10::MaybeOwned<Tensor>`
93
+ // must not outlive the original `Tensor` object that `to_expand`
94
+ // referred to! The deleted rvalue reference overloads of these
95
+ // functions help with this by preventing trivial use of a temporary
96
+ // resulting from a function call, but it is still possible to make a
97
+ // mistake.
98
+
99
+ inline c10::MaybeOwned<Tensor> expand_inplace(
100
+ const Tensor& tensor,
101
+ const Tensor& to_expand) {
102
+ if (tensor.sym_sizes().equals(to_expand.sym_sizes())) {
103
+ return c10::MaybeOwned<Tensor>::borrowed(to_expand);
104
+ }
105
+ return c10::MaybeOwned<Tensor>::owned(
106
+ to_expand.expand_symint(tensor.sym_sizes()));
107
+ }
108
+
109
+ inline c10::MaybeOwned<Tensor> expand_inplace(
110
+ const Tensor& tensor,
111
+ Tensor&& to_expand) = delete;
112
+
113
+ inline c10::MaybeOwned<Tensor> expand_inplace(
114
+ const Tensor& tensor,
115
+ const Tensor& to_expand,
116
+ const char* api_name) {
117
+ check_defined({tensor, to_expand}, api_name);
118
+ return expand_inplace(tensor, to_expand);
119
+ }
120
+
121
+ inline c10::MaybeOwned<Tensor> expand_inplace(
122
+ const Tensor& tensor,
123
+ Tensor&& to_expand,
124
+ const char* api_name) = delete;
125
+
126
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
127
+ expand_inplace(
128
+ const Tensor& tensor,
129
+ const Tensor& to_expand1,
130
+ const Tensor& to_expand2) {
131
+ if (tensor.sizes().equals(to_expand1.sizes()) &&
132
+ tensor.sizes().equals((to_expand2.sizes()))) {
133
+ return std::make_tuple(
134
+ c10::MaybeOwned<Tensor>::borrowed(to_expand1),
135
+ c10::MaybeOwned<Tensor>::borrowed(to_expand2));
136
+ }
137
+
138
+ return std::make_tuple(
139
+ c10::MaybeOwned<Tensor>::owned(to_expand1.expand(tensor.sizes())),
140
+ c10::MaybeOwned<Tensor>::owned(to_expand2.expand(tensor.sizes())));
141
+ }
142
+
143
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
144
+ expand_inplace(
145
+ const Tensor& tensor,
146
+ Tensor&& to_expand1,
147
+ const Tensor& to_expand2) = delete;
148
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
149
+ expand_inplace(
150
+ const Tensor& tensor,
151
+ const Tensor& to_expand1,
152
+ Tensor&& to_expand2) = delete;
153
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
154
+ expand_inplace(const Tensor& tensor, Tensor&& to_expand1, Tensor&& to_expand2) =
155
+ delete;
156
+
157
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
158
+ expand_inplace(
159
+ const Tensor& tensor,
160
+ const Tensor& to_expand1,
161
+ const Tensor& to_expand2,
162
+ const char* api_name) {
163
+ check_defined({tensor, to_expand1, to_expand2}, api_name);
164
+ return expand_inplace(tensor, to_expand1, to_expand2);
165
+ }
166
+
167
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
168
+ expand_inplace(
169
+ const Tensor& tensor,
170
+ Tensor&& to_expand1,
171
+ const Tensor& to_expand2,
172
+ const char* api_name) = delete;
173
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
174
+ expand_inplace(
175
+ const Tensor& tensor,
176
+ const Tensor& to_expand1,
177
+ Tensor&& to_expand2,
178
+ const char* api_name) = delete;
179
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
180
+ expand_inplace(
181
+ const Tensor& tensor,
182
+ Tensor&& to_expand1,
183
+ Tensor&& to_expand2,
184
+ const char* api_name) = delete;
185
+
186
+ // See NOTE [ ExpandUtils Borrowing ] above for `MaybeOwned` explanation.
187
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
188
+ expand_outplace(const Tensor& to_expand1, const Tensor& to_expand2) {
189
+ auto s1 = to_expand1.sym_sizes();
190
+ auto s2 = to_expand2.sym_sizes();
191
+ if (s1.equals(s2)) {
192
+ return std::make_tuple(
193
+ c10::MaybeOwned<Tensor>::borrowed(to_expand1),
194
+ c10::MaybeOwned<Tensor>::borrowed(to_expand2));
195
+ }
196
+
197
+ auto expanded_size = infer_size_symdimvector(s1, s2);
198
+ return std::make_tuple(
199
+ c10::MaybeOwned<Tensor>::owned(to_expand1.expand_symint(expanded_size)),
200
+ c10::MaybeOwned<Tensor>::owned(to_expand2.expand_symint(expanded_size)));
201
+ }
202
+
203
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
204
+ expand_outplace(Tensor&& to_expand1, const Tensor& to_expand2) = delete;
205
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
206
+ expand_outplace(const Tensor& to_expand1, Tensor&& to_expand2) = delete;
207
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
208
+ expand_outplace(Tensor&& to_expand1, Tensor&& to_expand2) = delete;
209
+
210
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
211
+ expand_outplace(
212
+ const Tensor& to_expand1,
213
+ const Tensor& to_expand2,
214
+ const char* api_name) {
215
+ check_defined({to_expand1, to_expand2}, api_name);
216
+ return expand_outplace(to_expand1, to_expand2);
217
+ }
218
+
219
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
220
+ expand_outplace(
221
+ Tensor&& to_expand1,
222
+ const Tensor& to_expand2,
223
+ const char* api_name) = delete;
224
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
225
+ expand_outplace(
226
+ const Tensor& to_expand1,
227
+ Tensor&& to_expand2,
228
+ const char* api_name) = delete;
229
+ inline std::tuple<c10::MaybeOwned<Tensor>, c10::MaybeOwned<Tensor>>
230
+ expand_outplace(
231
+ Tensor&& to_expand1,
232
+ Tensor&& to_expand2,
233
+ const char* api_name) = delete;
234
+
235
+ inline std::tuple<
236
+ c10::MaybeOwned<Tensor>,
237
+ c10::MaybeOwned<Tensor>,
238
+ c10::MaybeOwned<Tensor>>
239
+ expand_outplace(
240
+ const Tensor& to_expand1,
241
+ const Tensor& to_expand2,
242
+ const Tensor& to_expand3) {
243
+ if (to_expand1.sizes().equals(to_expand2.sizes()) &&
244
+ to_expand1.sizes().equals(to_expand3.sizes())) {
245
+ return std::make_tuple(
246
+ c10::MaybeOwned<Tensor>::borrowed(to_expand1),
247
+ c10::MaybeOwned<Tensor>::borrowed(to_expand2),
248
+ c10::MaybeOwned<Tensor>::borrowed(to_expand3));
249
+ }
250
+
251
+ auto expanded_size12 =
252
+ infer_size_dimvector(to_expand1.sizes(), to_expand2.sizes());
253
+ auto expanded_size =
254
+ infer_size_dimvector(expanded_size12, to_expand3.sizes());
255
+ return std::make_tuple(
256
+ c10::MaybeOwned<Tensor>::owned(to_expand1.expand(expanded_size)),
257
+ c10::MaybeOwned<Tensor>::owned(to_expand2.expand(expanded_size)),
258
+ c10::MaybeOwned<Tensor>::owned(to_expand3.expand(expanded_size)));
259
+ }
260
+
261
+ inline std::tuple<
262
+ c10::MaybeOwned<Tensor>,
263
+ c10::MaybeOwned<Tensor>,
264
+ c10::MaybeOwned<Tensor>>
265
+ expand_outplace(
266
+ Tensor&& to_expand1,
267
+ const Tensor& to_expand2,
268
+ const Tensor& to_expand3) = delete;
269
+ inline std::tuple<
270
+ c10::MaybeOwned<Tensor>,
271
+ c10::MaybeOwned<Tensor>,
272
+ c10::MaybeOwned<Tensor>>
273
+ expand_outplace(
274
+ const Tensor& to_expand1,
275
+ Tensor&& to_expand2,
276
+ const Tensor& to_expand3) = delete;
277
+ inline std::tuple<
278
+ c10::MaybeOwned<Tensor>,
279
+ c10::MaybeOwned<Tensor>,
280
+ c10::MaybeOwned<Tensor>>
281
+ expand_outplace(
282
+ Tensor&& to_expand1,
283
+ Tensor&& to_expand2,
284
+ const Tensor& to_expand3) = delete;
285
+ inline std::tuple<
286
+ c10::MaybeOwned<Tensor>,
287
+ c10::MaybeOwned<Tensor>,
288
+ c10::MaybeOwned<Tensor>>
289
+ expand_outplace(
290
+ const Tensor& to_expand1,
291
+ const Tensor& to_expand2,
292
+ Tensor&& to_expand3) = delete;
293
+ inline std::tuple<
294
+ c10::MaybeOwned<Tensor>,
295
+ c10::MaybeOwned<Tensor>,
296
+ c10::MaybeOwned<Tensor>>
297
+ expand_outplace(
298
+ Tensor&& to_expand1,
299
+ const Tensor& to_expand2,
300
+ Tensor&& to_expand3) = delete;
301
+ inline std::tuple<
302
+ c10::MaybeOwned<Tensor>,
303
+ c10::MaybeOwned<Tensor>,
304
+ c10::MaybeOwned<Tensor>>
305
+ expand_outplace(
306
+ const Tensor& to_expand1,
307
+ Tensor&& to_expand2,
308
+ Tensor&& to_expand3) = delete;
309
+ inline std::tuple<
310
+ c10::MaybeOwned<Tensor>,
311
+ c10::MaybeOwned<Tensor>,
312
+ c10::MaybeOwned<Tensor>>
313
+ expand_outplace(Tensor&& to_expand1, Tensor&& to_expand2, Tensor&& to_expand3) =
314
+ delete;
315
+
316
+ inline std::tuple<
317
+ c10::MaybeOwned<Tensor>,
318
+ c10::MaybeOwned<Tensor>,
319
+ c10::MaybeOwned<Tensor>>
320
+ expand_outplace(
321
+ const Tensor& to_expand1,
322
+ const Tensor& to_expand2,
323
+ const Tensor& to_expand3,
324
+ const char* api_name) {
325
+ check_defined({to_expand1, to_expand2, to_expand3}, api_name);
326
+ return expand_outplace(to_expand1, to_expand2, to_expand3);
327
+ }
328
+
329
+ inline std::tuple<
330
+ c10::MaybeOwned<Tensor>,
331
+ c10::MaybeOwned<Tensor>,
332
+ c10::MaybeOwned<Tensor>>
333
+ expand_outplace(
334
+ Tensor&& to_expand1,
335
+ const Tensor& to_expand2,
336
+ const Tensor& to_expand3,
337
+ const char* api_name) = delete;
338
+ inline std::tuple<
339
+ c10::MaybeOwned<Tensor>,
340
+ c10::MaybeOwned<Tensor>,
341
+ c10::MaybeOwned<Tensor>>
342
+ expand_outplace(
343
+ const Tensor& to_expand1,
344
+ Tensor&& to_expand2,
345
+ const Tensor& to_expand3,
346
+ const char* api_name) = delete;
347
+ inline std::tuple<
348
+ c10::MaybeOwned<Tensor>,
349
+ c10::MaybeOwned<Tensor>,
350
+ c10::MaybeOwned<Tensor>>
351
+ expand_outplace(
352
+ Tensor&& to_expand1,
353
+ Tensor&& to_expand2,
354
+ const Tensor& to_expand3,
355
+ const char* api_name) = delete;
356
+ inline std::tuple<
357
+ c10::MaybeOwned<Tensor>,
358
+ c10::MaybeOwned<Tensor>,
359
+ c10::MaybeOwned<Tensor>>
360
+ expand_outplace(
361
+ const Tensor& to_expand1,
362
+ const Tensor& to_expand2,
363
+ Tensor&& to_expand3,
364
+ const char* api_name) = delete;
365
+ inline std::tuple<
366
+ c10::MaybeOwned<Tensor>,
367
+ c10::MaybeOwned<Tensor>,
368
+ c10::MaybeOwned<Tensor>>
369
+ expand_outplace(
370
+ Tensor&& to_expand1,
371
+ const Tensor& to_expand2,
372
+ Tensor&& to_expand3,
373
+ const char* api_name) = delete;
374
+ inline std::tuple<
375
+ c10::MaybeOwned<Tensor>,
376
+ c10::MaybeOwned<Tensor>,
377
+ c10::MaybeOwned<Tensor>>
378
+ expand_outplace(
379
+ const Tensor& to_expand1,
380
+ Tensor&& to_expand2,
381
+ Tensor&& to_expand3,
382
+ const char* api_name) = delete;
383
+ inline std::tuple<
384
+ c10::MaybeOwned<Tensor>,
385
+ c10::MaybeOwned<Tensor>,
386
+ c10::MaybeOwned<Tensor>>
387
+ expand_outplace(
388
+ Tensor&& to_expand1,
389
+ Tensor&& to_expand2,
390
+ Tensor&& to_expand3,
391
+ const char* api_name) = delete;
392
+
393
+ inline c10::MaybeOwned<Tensor> expand_size(
394
+ const Tensor& to_expand,
395
+ IntArrayRef sizes) {
396
+ if (to_expand.sizes().equals(sizes)) {
397
+ return c10::MaybeOwned<Tensor>::borrowed(to_expand);
398
+ }
399
+
400
+ return c10::MaybeOwned<Tensor>::owned(to_expand.expand(sizes));
401
+ }
402
+
403
+ inline c10::MaybeOwned<Tensor> expand_size(
404
+ Tensor&& to_expand,
405
+ IntArrayRef sizes) = delete;
406
+
407
+ inline c10::MaybeOwned<Tensor> expand_size(
408
+ const Tensor& to_expand,
409
+ IntArrayRef sizes,
410
+ const char* api_name) {
411
+ check_defined({to_expand}, api_name);
412
+ return expand_size(to_expand, sizes);
413
+ }
414
+
415
+ inline c10::MaybeOwned<Tensor> expand_size(
416
+ Tensor&& to_expand,
417
+ IntArrayRef sizes,
418
+ const char* api_name) = delete;
419
+
420
+ inline std::vector<Tensor> expand_outplace(TensorList to_expand) {
421
+ // expands a list of Tensors; ignores undefined (null) tensors
422
+ bool first = true;
423
+ DimVector sizes;
424
+ for (const auto i : c10::irange(to_expand.size())) {
425
+ if (!to_expand[i].defined()) {
426
+ continue;
427
+ } else if (first) {
428
+ sizes = to_expand[i].sizes();
429
+ first = false;
430
+ } else {
431
+ sizes = infer_size_dimvector(sizes, to_expand[i].sizes());
432
+ }
433
+ }
434
+
435
+ std::vector<Tensor> result(to_expand.size());
436
+ for (const auto i : c10::irange(to_expand.size())) {
437
+ if (!to_expand[i].defined()) {
438
+ continue;
439
+ } else if (to_expand[i].sizes().equals(sizes)) {
440
+ result[i] = to_expand[i];
441
+ } else {
442
+ result[i] = to_expand[i].expand(sizes);
443
+ }
444
+ }
445
+ return result;
446
+ }
447
+
448
+ template <typename T>
449
+ inline Tensor _sum_to(
450
+ Tensor tensor,
451
+ const c10::ArrayRef<T> shape,
452
+ bool always_return_non_view = false) {
453
+ if (shape.size() == 0) {
454
+ return tensor.sum();
455
+ }
456
+
457
+ auto sizes = at::symint::sizes<T>(tensor);
458
+ c10::SmallVector<int64_t, 8> reduce_dims;
459
+ const int64_t leading_dims = sizes.size() - shape.size();
460
+ for (const auto i : c10::irange(leading_dims)) {
461
+ reduce_dims.push_back(i);
462
+ }
463
+ for (int64_t i = leading_dims; i < static_cast<int64_t>(sizes.size()); ++i) {
464
+ if (TORCH_GUARD_SIZE_OBLIVIOUS(sym_eq(shape[i - leading_dims], 1)) &&
465
+ TORCH_GUARD_SIZE_OBLIVIOUS(sym_ne(sizes[i], 1))) {
466
+ reduce_dims.push_back(i);
467
+ }
468
+ }
469
+
470
+ if (!reduce_dims.empty()) {
471
+ tensor = tensor.sum(reduce_dims, /*keepdim=*/true);
472
+ }
473
+
474
+ if (always_return_non_view) {
475
+ // This is only actually used by the functionalization pass.
476
+ // We want to be able to guarantee that this function doesn't return a view
477
+ // of the input.
478
+ return leading_dims > 0 ? at::symint::view_copy<T>(tensor, shape)
479
+ : tensor.clone();
480
+ } else {
481
+ return leading_dims > 0 ? at::symint::view<T>(tensor, shape) : tensor;
482
+ }
483
+ }
484
+
485
+ inline Tensor sum_to(
486
+ Tensor tensor,
487
+ const c10::SymIntArrayRef shape,
488
+ bool always_return_non_view = false) {
489
+ return _sum_to(std::move(tensor), shape, always_return_non_view);
490
+ }
491
+
492
+ // Sums `tensor` repeatedly to produce a tensor of shape `shape`.
493
+ // Precondition: is_expandable_to(shape, tensor.sizes()) must be true
494
+ inline Tensor sum_to(
495
+ Tensor tensor,
496
+ const IntArrayRef shape,
497
+ bool always_return_non_view = false) {
498
+ return _sum_to(std::move(tensor), shape, always_return_non_view);
499
+ }
500
+
501
+ inline bool is_expandable_to(
502
+ SymIntArrayRef shape,
503
+ c10::SymIntArrayRef desired) {
504
+ size_t ndim = shape.size();
505
+ size_t target_dim = desired.size();
506
+ if (ndim > target_dim) {
507
+ return false;
508
+ }
509
+ for (const auto i : c10::irange(ndim)) {
510
+ const auto& size = shape[ndim - i - 1];
511
+ const auto& target = desired[target_dim - i - 1];
512
+ if (size != target && size != 1) {
513
+ return false;
514
+ }
515
+ }
516
+ return true;
517
+ }
518
+
519
+ inline bool is_expandable_to(IntArrayRef shape, IntArrayRef desired) {
520
+ auto sym_shape = c10::SymIntArrayRef(
521
+ reinterpret_cast<const c10::SymInt*>(shape.data()), shape.size());
522
+ auto sym_desired = c10::SymIntArrayRef(
523
+ reinterpret_cast<const c10::SymInt*>(desired.data()), desired.size());
524
+ return is_expandable_to(sym_shape, sym_desired);
525
+ }
526
+
527
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/Functions.h ADDED
@@ -0,0 +1,1454 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ // @generated by torchgen/gen.py from Functions.h
4
+
5
+ #ifdef TORCH_ASSERT_NO_OPERATORS
6
+ #error This change adds a dependency on native_functions.yaml, \
7
+ meaning the file will need to be re-compiled every time an operator \
8
+ is changed or added. Consider if your change would be better placed in \
9
+ another file, or if a more specific header might achieve the same goal. \
10
+ See NOTE: [Tensor vs. TensorBase]
11
+ #endif
12
+
13
+ #if defined(AT_PER_OPERATOR_HEADERS) && defined(TORCH_ASSERT_ONLY_METHOD_OPERATORS)
14
+ #error This change adds a dependency on all pytorch operators, meaning the \
15
+ file will need to be re-compiled every time an operator is changed or added. \
16
+ Consider including a specific operator from <ATen/ops/{my_operator}.h> and \
17
+ see NOTE [TORCH_ASSERT_ONLY_METHOD_OPERATORS].
18
+ #endif
19
+
20
+ // NOTE: [TORCH_ASSERT_ONLY_METHOD_OPERATORS]
21
+ //
22
+ // In ATen, certain generated headers files include the definitions of
23
+ // every single operator in PyTorch. Unfortunately this means every
24
+ // time an operator signature is updated or changed in
25
+ // native_functions.yaml, you (and every other PyTorch developer) need
26
+ // to recompile every source file that includes any of these headers.
27
+ //
28
+ // To break up these header dependencies, and improve incremental
29
+ // build times for all PyTorch developers. These headers are split
30
+ // into per-operator headers in the `ATen/ops` folder. This limits
31
+ // incremental builds to only changes to methods of `Tensor`, or files
32
+ // that use the specific operator being changed. With `at::sum` as an
33
+ // example, you should include
34
+ //
35
+ // <ATen/ops/sum.h> // instead of ATen/Functions.h
36
+ // <ATen/ops/sum_native.h> // instead of ATen/NativeFunctions.h
37
+ // <ATen/ops/sum_ops.h> // instead of ATen/Operators.h
38
+ // <ATen/ops/sum_cpu_dispatch.h> // instead of ATen/CPUFunctions.h
39
+ //
40
+ // However, even if you're careful to use this in your own code.
41
+ // `Functions.h` might be included indirectly through another header
42
+ // without you realising. To avoid this, you can add
43
+ //
44
+ // #define TORCH_ASSERT_ONLY_METHOD_OPERATORS
45
+ //
46
+ // to the top of your source file. This way any time the non-specific
47
+ // headers are included, the compiler will error out.
48
+ //
49
+ // Also, be aware that `ops` are not available in all build
50
+ // configurations (namely fb-internal) so you must guard these
51
+ // includes with `#ifdef AT_PER_OPERATOR_HEADERS`. e.g.
52
+ //
53
+ // #ifndef AT_PER_OPERATOR_HEADERS
54
+ // #include <ATen/Functions.h>
55
+ // #else
56
+ // #include <ATen/ops/sum.h>
57
+ // #endif
58
+
59
+ #include <ATen/Context.h>
60
+ #include <ATen/DeviceGuard.h>
61
+ #include <ATen/TensorUtils.h>
62
+ #include <ATen/TracerMode.h>
63
+ #include <ATen/core/Generator.h>
64
+ #include <ATen/core/Reduction.h>
65
+ #include <c10/core/SymInt.h>
66
+ #include <ATen/core/Tensor.h>
67
+ #include <c10/core/Scalar.h>
68
+ #include <c10/core/Storage.h>
69
+ #include <c10/core/TensorOptions.h>
70
+ #include <c10/util/Deprecated.h>
71
+ #include <optional>
72
+ #include <c10/util/OptionalArrayRef.h>
73
+
74
+ #include <ATen/ops/from_blob.h>
75
+ #include <ATen/ops/tensor.h>
76
+
77
+ #include <ATen/ops/_adaptive_avg_pool2d.h>
78
+ #include <ATen/ops/_adaptive_avg_pool2d_backward.h>
79
+ #include <ATen/ops/_adaptive_avg_pool3d.h>
80
+ #include <ATen/ops/_adaptive_avg_pool3d_backward.h>
81
+ #include <ATen/ops/_add_batch_dim.h>
82
+ #include <ATen/ops/_add_relu.h>
83
+ #include <ATen/ops/_addmm_activation.h>
84
+ #include <ATen/ops/_aminmax.h>
85
+ #include <ATen/ops/_amp_foreach_non_finite_check_and_unscale.h>
86
+ #include <ATen/ops/_amp_update_scale.h>
87
+ #include <ATen/ops/_assert_async.h>
88
+ #include <ATen/ops/_assert_scalar.h>
89
+ #include <ATen/ops/_assert_tensor_metadata.h>
90
+ #include <ATen/ops/_autocast_to_full_precision.h>
91
+ #include <ATen/ops/_autocast_to_reduced_precision.h>
92
+ #include <ATen/ops/_backward.h>
93
+ #include <ATen/ops/_batch_norm_impl_index.h>
94
+ #include <ATen/ops/_batch_norm_impl_index_backward.h>
95
+ #include <ATen/ops/_batch_norm_no_update.h>
96
+ #include <ATen/ops/_batch_norm_with_update.h>
97
+ #include <ATen/ops/_cast_Byte.h>
98
+ #include <ATen/ops/_cast_Char.h>
99
+ #include <ATen/ops/_cast_Double.h>
100
+ #include <ATen/ops/_cast_Float.h>
101
+ #include <ATen/ops/_cast_Half.h>
102
+ #include <ATen/ops/_cast_Int.h>
103
+ #include <ATen/ops/_cast_Long.h>
104
+ #include <ATen/ops/_cast_Short.h>
105
+ #include <ATen/ops/_cdist_backward.h>
106
+ #include <ATen/ops/_cdist_forward.h>
107
+ #include <ATen/ops/_cholesky_solve_helper.h>
108
+ #include <ATen/ops/_choose_qparams_per_tensor.h>
109
+ #include <ATen/ops/_chunk_cat.h>
110
+ #include <ATen/ops/_coalesce.h>
111
+ #include <ATen/ops/_coalesced.h>
112
+ #include <ATen/ops/_compute_linear_combination.h>
113
+ #include <ATen/ops/_conj.h>
114
+ #include <ATen/ops/_conj_copy.h>
115
+ #include <ATen/ops/_conj_physical.h>
116
+ #include <ATen/ops/_conv_depthwise2d.h>
117
+ #include <ATen/ops/_convert_indices_from_coo_to_csr.h>
118
+ #include <ATen/ops/_convert_indices_from_csr_to_coo.h>
119
+ #include <ATen/ops/_convert_weight_to_int4pack.h>
120
+ #include <ATen/ops/_convolution.h>
121
+ #include <ATen/ops/_convolution_double_backward.h>
122
+ #include <ATen/ops/_convolution_mode.h>
123
+ #include <ATen/ops/_copy_from.h>
124
+ #include <ATen/ops/_copy_from_and_resize.h>
125
+ #include <ATen/ops/_cslt_compress.h>
126
+ #include <ATen/ops/_cslt_sparse_mm.h>
127
+ #include <ATen/ops/_cslt_sparse_mm_search.h>
128
+ #include <ATen/ops/_ctc_loss.h>
129
+ #include <ATen/ops/_ctc_loss_backward.h>
130
+ #include <ATen/ops/_cudnn_ctc_loss.h>
131
+ #include <ATen/ops/_cudnn_init_dropout_state.h>
132
+ #include <ATen/ops/_cudnn_rnn.h>
133
+ #include <ATen/ops/_cudnn_rnn_backward.h>
134
+ #include <ATen/ops/_cudnn_rnn_flatten_weight.h>
135
+ #include <ATen/ops/_cufft_clear_plan_cache.h>
136
+ #include <ATen/ops/_cufft_get_plan_cache_max_size.h>
137
+ #include <ATen/ops/_cufft_get_plan_cache_size.h>
138
+ #include <ATen/ops/_cufft_set_plan_cache_max_size.h>
139
+ #include <ATen/ops/_cummax_helper.h>
140
+ #include <ATen/ops/_cummin_helper.h>
141
+ #include <ATen/ops/_debug_has_internal_overlap.h>
142
+ #include <ATen/ops/_dimI.h>
143
+ #include <ATen/ops/_dimV.h>
144
+ #include <ATen/ops/_dim_arange.h>
145
+ #include <ATen/ops/_dirichlet_grad.h>
146
+ #include <ATen/ops/_efficient_attention_backward.h>
147
+ #include <ATen/ops/_efficient_attention_forward.h>
148
+ #include <ATen/ops/_efficientzerotensor.h>
149
+ #include <ATen/ops/_embedding_bag.h>
150
+ #include <ATen/ops/_embedding_bag_backward.h>
151
+ #include <ATen/ops/_embedding_bag_dense_backward.h>
152
+ #include <ATen/ops/_embedding_bag_forward_only.h>
153
+ #include <ATen/ops/_embedding_bag_per_sample_weights_backward.h>
154
+ #include <ATen/ops/_embedding_bag_sparse_backward.h>
155
+ #include <ATen/ops/_empty_affine_quantized.h>
156
+ #include <ATen/ops/_empty_per_channel_affine_quantized.h>
157
+ #include <ATen/ops/_euclidean_dist.h>
158
+ #include <ATen/ops/_fake_quantize_learnable_per_channel_affine.h>
159
+ #include <ATen/ops/_fake_quantize_learnable_per_channel_affine_backward.h>
160
+ #include <ATen/ops/_fake_quantize_learnable_per_tensor_affine.h>
161
+ #include <ATen/ops/_fake_quantize_learnable_per_tensor_affine_backward.h>
162
+ #include <ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams.h>
163
+ #include <ATen/ops/_fft_c2c.h>
164
+ #include <ATen/ops/_fft_c2r.h>
165
+ #include <ATen/ops/_fft_r2c.h>
166
+ #include <ATen/ops/_fill_mem_eff_dropout_mask.h>
167
+ #include <ATen/ops/_flash_attention_backward.h>
168
+ #include <ATen/ops/_flash_attention_forward.h>
169
+ #include <ATen/ops/_foobar.h>
170
+ #include <ATen/ops/_foreach_abs.h>
171
+ #include <ATen/ops/_foreach_acos.h>
172
+ #include <ATen/ops/_foreach_add.h>
173
+ #include <ATen/ops/_foreach_addcdiv.h>
174
+ #include <ATen/ops/_foreach_addcmul.h>
175
+ #include <ATen/ops/_foreach_asin.h>
176
+ #include <ATen/ops/_foreach_atan.h>
177
+ #include <ATen/ops/_foreach_ceil.h>
178
+ #include <ATen/ops/_foreach_clamp_max.h>
179
+ #include <ATen/ops/_foreach_clamp_min.h>
180
+ #include <ATen/ops/_foreach_copy.h>
181
+ #include <ATen/ops/_foreach_cos.h>
182
+ #include <ATen/ops/_foreach_cosh.h>
183
+ #include <ATen/ops/_foreach_div.h>
184
+ #include <ATen/ops/_foreach_erf.h>
185
+ #include <ATen/ops/_foreach_erfc.h>
186
+ #include <ATen/ops/_foreach_exp.h>
187
+ #include <ATen/ops/_foreach_expm1.h>
188
+ #include <ATen/ops/_foreach_floor.h>
189
+ #include <ATen/ops/_foreach_frac.h>
190
+ #include <ATen/ops/_foreach_lerp.h>
191
+ #include <ATen/ops/_foreach_lgamma.h>
192
+ #include <ATen/ops/_foreach_log.h>
193
+ #include <ATen/ops/_foreach_log10.h>
194
+ #include <ATen/ops/_foreach_log1p.h>
195
+ #include <ATen/ops/_foreach_log2.h>
196
+ #include <ATen/ops/_foreach_max.h>
197
+ #include <ATen/ops/_foreach_maximum.h>
198
+ #include <ATen/ops/_foreach_minimum.h>
199
+ #include <ATen/ops/_foreach_mul.h>
200
+ #include <ATen/ops/_foreach_neg.h>
201
+ #include <ATen/ops/_foreach_norm.h>
202
+ #include <ATen/ops/_foreach_pow.h>
203
+ #include <ATen/ops/_foreach_reciprocal.h>
204
+ #include <ATen/ops/_foreach_round.h>
205
+ #include <ATen/ops/_foreach_sigmoid.h>
206
+ #include <ATen/ops/_foreach_sign.h>
207
+ #include <ATen/ops/_foreach_sin.h>
208
+ #include <ATen/ops/_foreach_sinh.h>
209
+ #include <ATen/ops/_foreach_sqrt.h>
210
+ #include <ATen/ops/_foreach_sub.h>
211
+ #include <ATen/ops/_foreach_tan.h>
212
+ #include <ATen/ops/_foreach_tanh.h>
213
+ #include <ATen/ops/_foreach_trunc.h>
214
+ #include <ATen/ops/_foreach_zero.h>
215
+ #include <ATen/ops/_functional_assert_async.h>
216
+ #include <ATen/ops/_functional_assert_scalar.h>
217
+ #include <ATen/ops/_functional_sym_constrain_range.h>
218
+ #include <ATen/ops/_functional_sym_constrain_range_for_size.h>
219
+ #include <ATen/ops/_fused_adagrad.h>
220
+ #include <ATen/ops/_fused_adam.h>
221
+ #include <ATen/ops/_fused_adamw.h>
222
+ #include <ATen/ops/_fused_dropout.h>
223
+ #include <ATen/ops/_fused_moving_avg_obs_fq_helper.h>
224
+ #include <ATen/ops/_fused_sdp_choice.h>
225
+ #include <ATen/ops/_fused_sgd.h>
226
+ #include <ATen/ops/_fw_primal.h>
227
+ #include <ATen/ops/_fw_primal_copy.h>
228
+ #include <ATen/ops/_gather_sparse_backward.h>
229
+ #include <ATen/ops/_grid_sampler_2d_cpu_fallback.h>
230
+ #include <ATen/ops/_grid_sampler_2d_cpu_fallback_backward.h>
231
+ #include <ATen/ops/_has_compatible_shallow_copy_type.h>
232
+ #include <ATen/ops/_has_same_storage_numel.h>
233
+ #include <ATen/ops/_histogramdd_bin_edges.h>
234
+ #include <ATen/ops/_histogramdd_from_bin_cts.h>
235
+ #include <ATen/ops/_histogramdd_from_bin_tensors.h>
236
+ #include <ATen/ops/_index_put_impl.h>
237
+ #include <ATen/ops/_indices.h>
238
+ #include <ATen/ops/_indices_copy.h>
239
+ #include <ATen/ops/_int_mm.h>
240
+ #include <ATen/ops/_is_all_true.h>
241
+ #include <ATen/ops/_is_any_true.h>
242
+ #include <ATen/ops/_is_zerotensor.h>
243
+ #include <ATen/ops/_jagged_to_padded_dense_forward.h>
244
+ #include <ATen/ops/_lazy_clone.h>
245
+ #include <ATen/ops/_linalg_check_errors.h>
246
+ #include <ATen/ops/_linalg_det.h>
247
+ #include <ATen/ops/_linalg_eigh.h>
248
+ #include <ATen/ops/_linalg_eigvals.h>
249
+ #include <ATen/ops/_linalg_slogdet.h>
250
+ #include <ATen/ops/_linalg_solve_ex.h>
251
+ #include <ATen/ops/_linalg_svd.h>
252
+ #include <ATen/ops/_local_scalar_dense.h>
253
+ #include <ATen/ops/_log_softmax.h>
254
+ #include <ATen/ops/_log_softmax_backward_data.h>
255
+ #include <ATen/ops/_logcumsumexp.h>
256
+ #include <ATen/ops/_lstm_mps.h>
257
+ #include <ATen/ops/_lu_with_info.h>
258
+ #include <ATen/ops/_make_dep_token.h>
259
+ #include <ATen/ops/_make_dual.h>
260
+ #include <ATen/ops/_make_dual_copy.h>
261
+ #include <ATen/ops/_make_per_channel_quantized_tensor.h>
262
+ #include <ATen/ops/_make_per_tensor_quantized_tensor.h>
263
+ #include <ATen/ops/_masked_scale.h>
264
+ #include <ATen/ops/_masked_softmax.h>
265
+ #include <ATen/ops/_masked_softmax_backward.h>
266
+ #include <ATen/ops/_mixed_dtypes_linear.h>
267
+ #include <ATen/ops/_mkldnn_reshape.h>
268
+ #include <ATen/ops/_mkldnn_transpose.h>
269
+ #include <ATen/ops/_mps_convolution.h>
270
+ #include <ATen/ops/_mps_convolution_transpose.h>
271
+ #include <ATen/ops/_native_batch_norm_legit.h>
272
+ #include <ATen/ops/_native_batch_norm_legit_no_training.h>
273
+ #include <ATen/ops/_native_multi_head_attention.h>
274
+ #include <ATen/ops/_neg_view.h>
275
+ #include <ATen/ops/_neg_view_copy.h>
276
+ #include <ATen/ops/_nested_compute_contiguous_strides_offsets.h>
277
+ #include <ATen/ops/_nested_from_padded.h>
278
+ #include <ATen/ops/_nested_from_padded_and_nested_example.h>
279
+ #include <ATen/ops/_nested_get_jagged_dummy.h>
280
+ #include <ATen/ops/_nested_get_lengths.h>
281
+ #include <ATen/ops/_nested_get_max_seqlen.h>
282
+ #include <ATen/ops/_nested_get_min_seqlen.h>
283
+ #include <ATen/ops/_nested_get_offsets.h>
284
+ #include <ATen/ops/_nested_get_ragged_idx.h>
285
+ #include <ATen/ops/_nested_get_values.h>
286
+ #include <ATen/ops/_nested_get_values_copy.h>
287
+ #include <ATen/ops/_nested_select_backward.h>
288
+ #include <ATen/ops/_nested_sum_backward.h>
289
+ #include <ATen/ops/_nested_tensor_from_mask.h>
290
+ #include <ATen/ops/_nested_tensor_from_mask_left_aligned.h>
291
+ #include <ATen/ops/_nested_tensor_from_tensor_list.h>
292
+ #include <ATen/ops/_nested_tensor_size.h>
293
+ #include <ATen/ops/_nested_tensor_softmax_with_shape.h>
294
+ #include <ATen/ops/_nested_tensor_storage_offsets.h>
295
+ #include <ATen/ops/_nested_tensor_strides.h>
296
+ #include <ATen/ops/_nested_view_from_buffer.h>
297
+ #include <ATen/ops/_nested_view_from_buffer_copy.h>
298
+ #include <ATen/ops/_nested_view_from_jagged.h>
299
+ #include <ATen/ops/_nested_view_from_jagged_copy.h>
300
+ #include <ATen/ops/_new_zeros_with_same_feature_meta.h>
301
+ #include <ATen/ops/_nnpack_available.h>
302
+ #include <ATen/ops/_nnpack_spatial_convolution.h>
303
+ #include <ATen/ops/_nnz.h>
304
+ #include <ATen/ops/_pack_padded_sequence.h>
305
+ #include <ATen/ops/_pack_padded_sequence_backward.h>
306
+ #include <ATen/ops/_pad_circular.h>
307
+ #include <ATen/ops/_pad_enum.h>
308
+ #include <ATen/ops/_pad_packed_sequence.h>
309
+ #include <ATen/ops/_padded_dense_to_jagged_forward.h>
310
+ #include <ATen/ops/_pdist_backward.h>
311
+ #include <ATen/ops/_pdist_forward.h>
312
+ #include <ATen/ops/_pin_memory.h>
313
+ #include <ATen/ops/_prelu_kernel.h>
314
+ #include <ATen/ops/_prelu_kernel_backward.h>
315
+ #include <ATen/ops/_print.h>
316
+ #include <ATen/ops/_propagate_xla_data.h>
317
+ #include <ATen/ops/_remove_batch_dim.h>
318
+ #include <ATen/ops/_reshape_alias.h>
319
+ #include <ATen/ops/_reshape_alias_copy.h>
320
+ #include <ATen/ops/_reshape_copy.h>
321
+ #include <ATen/ops/_reshape_from_tensor.h>
322
+ #include <ATen/ops/_resize_output.h>
323
+ #include <ATen/ops/_rowwise_prune.h>
324
+ #include <ATen/ops/_safe_softmax.h>
325
+ #include <ATen/ops/_sample_dirichlet.h>
326
+ #include <ATen/ops/_saturate_weight_to_fp16.h>
327
+ #include <ATen/ops/_scaled_dot_product_attention_math.h>
328
+ #include <ATen/ops/_scaled_dot_product_attention_math_for_mps.h>
329
+ #include <ATen/ops/_scaled_dot_product_cudnn_attention.h>
330
+ #include <ATen/ops/_scaled_dot_product_cudnn_attention_backward.h>
331
+ #include <ATen/ops/_scaled_dot_product_efficient_attention.h>
332
+ #include <ATen/ops/_scaled_dot_product_efficient_attention_backward.h>
333
+ #include <ATen/ops/_scaled_dot_product_flash_attention.h>
334
+ #include <ATen/ops/_scaled_dot_product_flash_attention_backward.h>
335
+ #include <ATen/ops/_scaled_dot_product_flash_attention_for_cpu.h>
336
+ #include <ATen/ops/_scaled_dot_product_flash_attention_for_cpu_backward.h>
337
+ #include <ATen/ops/_scaled_dot_product_fused_attention_overrideable.h>
338
+ #include <ATen/ops/_scaled_dot_product_fused_attention_overrideable_backward.h>
339
+ #include <ATen/ops/_scaled_mm.h>
340
+ #include <ATen/ops/_segment_reduce_backward.h>
341
+ #include <ATen/ops/_shape_as_tensor.h>
342
+ #include <ATen/ops/_slow_conv2d_backward.h>
343
+ #include <ATen/ops/_slow_conv2d_forward.h>
344
+ #include <ATen/ops/_sobol_engine_draw.h>
345
+ #include <ATen/ops/_sobol_engine_ff.h>
346
+ #include <ATen/ops/_sobol_engine_initialize_state.h>
347
+ #include <ATen/ops/_sobol_engine_scramble.h>
348
+ #include <ATen/ops/_softmax.h>
349
+ #include <ATen/ops/_softmax_backward_data.h>
350
+ #include <ATen/ops/_sparse_addmm.h>
351
+ #include <ATen/ops/_sparse_broadcast_to.h>
352
+ #include <ATen/ops/_sparse_broadcast_to_copy.h>
353
+ #include <ATen/ops/_sparse_bsc_tensor_unsafe.h>
354
+ #include <ATen/ops/_sparse_bsr_tensor_unsafe.h>
355
+ #include <ATen/ops/_sparse_compressed_tensor_unsafe.h>
356
+ #include <ATen/ops/_sparse_compressed_tensor_with_dims.h>
357
+ #include <ATen/ops/_sparse_coo_tensor_unsafe.h>
358
+ #include <ATen/ops/_sparse_coo_tensor_with_dims.h>
359
+ #include <ATen/ops/_sparse_coo_tensor_with_dims_and_tensors.h>
360
+ #include <ATen/ops/_sparse_csc_tensor_unsafe.h>
361
+ #include <ATen/ops/_sparse_csr_prod.h>
362
+ #include <ATen/ops/_sparse_csr_sum.h>
363
+ #include <ATen/ops/_sparse_csr_tensor_unsafe.h>
364
+ #include <ATen/ops/_sparse_log_softmax.h>
365
+ #include <ATen/ops/_sparse_log_softmax_backward_data.h>
366
+ #include <ATen/ops/_sparse_mask_projection.h>
367
+ #include <ATen/ops/_sparse_mm.h>
368
+ #include <ATen/ops/_sparse_mm_reduce_impl.h>
369
+ #include <ATen/ops/_sparse_mm_reduce_impl_backward.h>
370
+ #include <ATen/ops/_sparse_semi_structured_addmm.h>
371
+ #include <ATen/ops/_sparse_semi_structured_apply.h>
372
+ #include <ATen/ops/_sparse_semi_structured_apply_dense.h>
373
+ #include <ATen/ops/_sparse_semi_structured_linear.h>
374
+ #include <ATen/ops/_sparse_semi_structured_mm.h>
375
+ #include <ATen/ops/_sparse_semi_structured_tile.h>
376
+ #include <ATen/ops/_sparse_softmax.h>
377
+ #include <ATen/ops/_sparse_softmax_backward_data.h>
378
+ #include <ATen/ops/_sparse_sparse_matmul.h>
379
+ #include <ATen/ops/_sparse_sum.h>
380
+ #include <ATen/ops/_sparse_sum_backward.h>
381
+ #include <ATen/ops/_spdiags.h>
382
+ #include <ATen/ops/_spsolve.h>
383
+ #include <ATen/ops/_stack.h>
384
+ #include <ATen/ops/_standard_gamma.h>
385
+ #include <ATen/ops/_standard_gamma_grad.h>
386
+ #include <ATen/ops/_test_ambiguous_defaults.h>
387
+ #include <ATen/ops/_test_autograd_multiple_dispatch.h>
388
+ #include <ATen/ops/_test_autograd_multiple_dispatch_view.h>
389
+ #include <ATen/ops/_test_autograd_multiple_dispatch_view_copy.h>
390
+ #include <ATen/ops/_test_check_tensor.h>
391
+ #include <ATen/ops/_test_functorch_fallback.h>
392
+ #include <ATen/ops/_test_optional_filled_intlist.h>
393
+ #include <ATen/ops/_test_optional_floatlist.h>
394
+ #include <ATen/ops/_test_optional_intlist.h>
395
+ #include <ATen/ops/_test_parallel_materialize.h>
396
+ #include <ATen/ops/_test_serialization_subcmul.h>
397
+ #include <ATen/ops/_test_string_default.h>
398
+ #include <ATen/ops/_test_warn_in_autograd.h>
399
+ #include <ATen/ops/_thnn_differentiable_gru_cell_backward.h>
400
+ #include <ATen/ops/_thnn_differentiable_lstm_cell_backward.h>
401
+ #include <ATen/ops/_thnn_fused_gru_cell.h>
402
+ #include <ATen/ops/_thnn_fused_gru_cell_backward.h>
403
+ #include <ATen/ops/_thnn_fused_lstm_cell.h>
404
+ #include <ATen/ops/_thnn_fused_lstm_cell_backward.h>
405
+ #include <ATen/ops/_thnn_fused_lstm_cell_backward_impl.h>
406
+ #include <ATen/ops/_to_copy.h>
407
+ #include <ATen/ops/_to_cpu.h>
408
+ #include <ATen/ops/_to_dense.h>
409
+ #include <ATen/ops/_to_sparse.h>
410
+ #include <ATen/ops/_to_sparse_bsc.h>
411
+ #include <ATen/ops/_to_sparse_bsr.h>
412
+ #include <ATen/ops/_to_sparse_csc.h>
413
+ #include <ATen/ops/_to_sparse_csr.h>
414
+ #include <ATen/ops/_to_sparse_semi_structured.h>
415
+ #include <ATen/ops/_transform_bias_rescale_qkv.h>
416
+ #include <ATen/ops/_transformer_encoder_layer_fwd.h>
417
+ #include <ATen/ops/_trilinear.h>
418
+ #include <ATen/ops/_triton_multi_head_attention.h>
419
+ #include <ATen/ops/_triton_scaled_dot_attention.h>
420
+ #include <ATen/ops/_unique.h>
421
+ #include <ATen/ops/_unique2.h>
422
+ #include <ATen/ops/_unpack_dual.h>
423
+ #include <ATen/ops/_unsafe_index.h>
424
+ #include <ATen/ops/_unsafe_index_put.h>
425
+ #include <ATen/ops/_unsafe_masked_index.h>
426
+ #include <ATen/ops/_unsafe_masked_index_put_accumulate.h>
427
+ #include <ATen/ops/_unsafe_view.h>
428
+ #include <ATen/ops/_upsample_bicubic2d_aa.h>
429
+ #include <ATen/ops/_upsample_bicubic2d_aa_backward.h>
430
+ #include <ATen/ops/_upsample_bilinear2d_aa.h>
431
+ #include <ATen/ops/_upsample_bilinear2d_aa_backward.h>
432
+ #include <ATen/ops/_upsample_nearest_exact1d.h>
433
+ #include <ATen/ops/_upsample_nearest_exact1d_backward.h>
434
+ #include <ATen/ops/_upsample_nearest_exact2d.h>
435
+ #include <ATen/ops/_upsample_nearest_exact2d_backward.h>
436
+ #include <ATen/ops/_upsample_nearest_exact3d.h>
437
+ #include <ATen/ops/_upsample_nearest_exact3d_backward.h>
438
+ #include <ATen/ops/_use_cudnn_ctc_loss.h>
439
+ #include <ATen/ops/_use_cudnn_rnn_flatten_weight.h>
440
+ #include <ATen/ops/_validate_compressed_sparse_indices.h>
441
+ #include <ATen/ops/_validate_sparse_bsc_tensor_args.h>
442
+ #include <ATen/ops/_validate_sparse_bsr_tensor_args.h>
443
+ #include <ATen/ops/_validate_sparse_compressed_tensor_args.h>
444
+ #include <ATen/ops/_validate_sparse_coo_tensor_args.h>
445
+ #include <ATen/ops/_validate_sparse_csc_tensor_args.h>
446
+ #include <ATen/ops/_validate_sparse_csr_tensor_args.h>
447
+ #include <ATen/ops/_values.h>
448
+ #include <ATen/ops/_values_copy.h>
449
+ #include <ATen/ops/_version.h>
450
+ #include <ATen/ops/_weight_int4pack_mm.h>
451
+ #include <ATen/ops/_weight_int8pack_mm.h>
452
+ #include <ATen/ops/_weight_norm.h>
453
+ #include <ATen/ops/_weight_norm_differentiable_backward.h>
454
+ #include <ATen/ops/_weight_norm_interface.h>
455
+ #include <ATen/ops/_weight_norm_interface_backward.h>
456
+ #include <ATen/ops/_wrapped_linear_prepack.h>
457
+ #include <ATen/ops/_wrapped_quantized_linear_prepacked.h>
458
+ #include <ATen/ops/abs.h>
459
+ #include <ATen/ops/absolute.h>
460
+ #include <ATen/ops/acos.h>
461
+ #include <ATen/ops/acosh.h>
462
+ #include <ATen/ops/adaptive_avg_pool1d.h>
463
+ #include <ATen/ops/adaptive_avg_pool2d.h>
464
+ #include <ATen/ops/adaptive_avg_pool3d.h>
465
+ #include <ATen/ops/adaptive_avg_pool3d_backward.h>
466
+ #include <ATen/ops/adaptive_max_pool1d.h>
467
+ #include <ATen/ops/adaptive_max_pool2d.h>
468
+ #include <ATen/ops/adaptive_max_pool2d_backward.h>
469
+ #include <ATen/ops/adaptive_max_pool3d.h>
470
+ #include <ATen/ops/adaptive_max_pool3d_backward.h>
471
+ #include <ATen/ops/add.h>
472
+ #include <ATen/ops/addbmm.h>
473
+ #include <ATen/ops/addcdiv.h>
474
+ #include <ATen/ops/addcmul.h>
475
+ #include <ATen/ops/addmm.h>
476
+ #include <ATen/ops/addmv.h>
477
+ #include <ATen/ops/addr.h>
478
+ #include <ATen/ops/adjoint.h>
479
+ #include <ATen/ops/affine_grid_generator.h>
480
+ #include <ATen/ops/affine_grid_generator_backward.h>
481
+ #include <ATen/ops/alias.h>
482
+ #include <ATen/ops/alias_copy.h>
483
+ #include <ATen/ops/align_as.h>
484
+ #include <ATen/ops/align_tensors.h>
485
+ #include <ATen/ops/align_to.h>
486
+ #include <ATen/ops/all.h>
487
+ #include <ATen/ops/allclose.h>
488
+ #include <ATen/ops/alpha_dropout.h>
489
+ #include <ATen/ops/amax.h>
490
+ #include <ATen/ops/amin.h>
491
+ #include <ATen/ops/aminmax.h>
492
+ #include <ATen/ops/and.h>
493
+ #include <ATen/ops/angle.h>
494
+ #include <ATen/ops/any.h>
495
+ #include <ATen/ops/arange.h>
496
+ #include <ATen/ops/arccos.h>
497
+ #include <ATen/ops/arccosh.h>
498
+ #include <ATen/ops/arcsin.h>
499
+ #include <ATen/ops/arcsinh.h>
500
+ #include <ATen/ops/arctan.h>
501
+ #include <ATen/ops/arctan2.h>
502
+ #include <ATen/ops/arctanh.h>
503
+ #include <ATen/ops/argmax.h>
504
+ #include <ATen/ops/argmin.h>
505
+ #include <ATen/ops/argsort.h>
506
+ #include <ATen/ops/argwhere.h>
507
+ #include <ATen/ops/as_strided.h>
508
+ #include <ATen/ops/as_strided_copy.h>
509
+ #include <ATen/ops/as_strided_scatter.h>
510
+ #include <ATen/ops/asin.h>
511
+ #include <ATen/ops/asinh.h>
512
+ #include <ATen/ops/atan.h>
513
+ #include <ATen/ops/atan2.h>
514
+ #include <ATen/ops/atanh.h>
515
+ #include <ATen/ops/atleast_1d.h>
516
+ #include <ATen/ops/atleast_2d.h>
517
+ #include <ATen/ops/atleast_3d.h>
518
+ #include <ATen/ops/avg_pool1d.h>
519
+ #include <ATen/ops/avg_pool2d.h>
520
+ #include <ATen/ops/avg_pool2d_backward.h>
521
+ #include <ATen/ops/avg_pool3d.h>
522
+ #include <ATen/ops/avg_pool3d_backward.h>
523
+ #include <ATen/ops/baddbmm.h>
524
+ #include <ATen/ops/bartlett_window.h>
525
+ #include <ATen/ops/batch_norm.h>
526
+ #include <ATen/ops/batch_norm_backward.h>
527
+ #include <ATen/ops/batch_norm_backward_elemt.h>
528
+ #include <ATen/ops/batch_norm_backward_reduce.h>
529
+ #include <ATen/ops/batch_norm_elemt.h>
530
+ #include <ATen/ops/batch_norm_gather_stats.h>
531
+ #include <ATen/ops/batch_norm_gather_stats_with_counts.h>
532
+ #include <ATen/ops/batch_norm_stats.h>
533
+ #include <ATen/ops/batch_norm_update_stats.h>
534
+ #include <ATen/ops/bernoulli.h>
535
+ #include <ATen/ops/bilinear.h>
536
+ #include <ATen/ops/binary_cross_entropy.h>
537
+ #include <ATen/ops/binary_cross_entropy_backward.h>
538
+ #include <ATen/ops/binary_cross_entropy_with_logits.h>
539
+ #include <ATen/ops/bincount.h>
540
+ #include <ATen/ops/binomial.h>
541
+ #include <ATen/ops/bitwise_and.h>
542
+ #include <ATen/ops/bitwise_left_shift.h>
543
+ #include <ATen/ops/bitwise_not.h>
544
+ #include <ATen/ops/bitwise_or.h>
545
+ #include <ATen/ops/bitwise_right_shift.h>
546
+ #include <ATen/ops/bitwise_xor.h>
547
+ #include <ATen/ops/blackman_window.h>
548
+ #include <ATen/ops/block_diag.h>
549
+ #include <ATen/ops/bmm.h>
550
+ #include <ATen/ops/broadcast_tensors.h>
551
+ #include <ATen/ops/broadcast_to.h>
552
+ #include <ATen/ops/bucketize.h>
553
+ #include <ATen/ops/can_cast.h>
554
+ #include <ATen/ops/cartesian_prod.h>
555
+ #include <ATen/ops/cat.h>
556
+ #include <ATen/ops/cauchy.h>
557
+ #include <ATen/ops/ccol_indices.h>
558
+ #include <ATen/ops/ccol_indices_copy.h>
559
+ #include <ATen/ops/cdist.h>
560
+ #include <ATen/ops/ceil.h>
561
+ #include <ATen/ops/celu.h>
562
+ #include <ATen/ops/chain_matmul.h>
563
+ #include <ATen/ops/chalf.h>
564
+ #include <ATen/ops/channel_shuffle.h>
565
+ #include <ATen/ops/cholesky.h>
566
+ #include <ATen/ops/cholesky_inverse.h>
567
+ #include <ATen/ops/cholesky_solve.h>
568
+ #include <ATen/ops/choose_qparams_optimized.h>
569
+ #include <ATen/ops/chunk.h>
570
+ #include <ATen/ops/clamp.h>
571
+ #include <ATen/ops/clamp_max.h>
572
+ #include <ATen/ops/clamp_min.h>
573
+ #include <ATen/ops/clip.h>
574
+ #include <ATen/ops/clone.h>
575
+ #include <ATen/ops/coalesce.h>
576
+ #include <ATen/ops/col2im.h>
577
+ #include <ATen/ops/col_indices.h>
578
+ #include <ATen/ops/col_indices_copy.h>
579
+ #include <ATen/ops/column_stack.h>
580
+ #include <ATen/ops/combinations.h>
581
+ #include <ATen/ops/complex.h>
582
+ #include <ATen/ops/concat.h>
583
+ #include <ATen/ops/concatenate.h>
584
+ #include <ATen/ops/conj.h>
585
+ #include <ATen/ops/conj_physical.h>
586
+ #include <ATen/ops/constant_pad_nd.h>
587
+ #include <ATen/ops/contiguous.h>
588
+ #include <ATen/ops/conv1d.h>
589
+ #include <ATen/ops/conv2d.h>
590
+ #include <ATen/ops/conv3d.h>
591
+ #include <ATen/ops/conv_depthwise3d.h>
592
+ #include <ATen/ops/conv_tbc.h>
593
+ #include <ATen/ops/conv_tbc_backward.h>
594
+ #include <ATen/ops/conv_transpose1d.h>
595
+ #include <ATen/ops/conv_transpose2d.h>
596
+ #include <ATen/ops/conv_transpose3d.h>
597
+ #include <ATen/ops/convolution.h>
598
+ #include <ATen/ops/convolution_backward.h>
599
+ #include <ATen/ops/convolution_backward_overrideable.h>
600
+ #include <ATen/ops/convolution_overrideable.h>
601
+ #include <ATen/ops/copy.h>
602
+ #include <ATen/ops/copy_sparse_to_sparse.h>
603
+ #include <ATen/ops/copysign.h>
604
+ #include <ATen/ops/corrcoef.h>
605
+ #include <ATen/ops/cos.h>
606
+ #include <ATen/ops/cosh.h>
607
+ #include <ATen/ops/cosine_embedding_loss.h>
608
+ #include <ATen/ops/cosine_similarity.h>
609
+ #include <ATen/ops/count_nonzero.h>
610
+ #include <ATen/ops/cov.h>
611
+ #include <ATen/ops/cross.h>
612
+ #include <ATen/ops/cross_entropy_loss.h>
613
+ #include <ATen/ops/crow_indices.h>
614
+ #include <ATen/ops/crow_indices_copy.h>
615
+ #include <ATen/ops/ctc_loss.h>
616
+ #include <ATen/ops/cudnn_affine_grid_generator.h>
617
+ #include <ATen/ops/cudnn_affine_grid_generator_backward.h>
618
+ #include <ATen/ops/cudnn_batch_norm.h>
619
+ #include <ATen/ops/cudnn_batch_norm_backward.h>
620
+ #include <ATen/ops/cudnn_convolution.h>
621
+ #include <ATen/ops/cudnn_convolution_add_relu.h>
622
+ #include <ATen/ops/cudnn_convolution_relu.h>
623
+ #include <ATen/ops/cudnn_convolution_transpose.h>
624
+ #include <ATen/ops/cudnn_grid_sampler.h>
625
+ #include <ATen/ops/cudnn_grid_sampler_backward.h>
626
+ #include <ATen/ops/cudnn_is_acceptable.h>
627
+ #include <ATen/ops/cummax.h>
628
+ #include <ATen/ops/cummaxmin_backward.h>
629
+ #include <ATen/ops/cummin.h>
630
+ #include <ATen/ops/cumprod.h>
631
+ #include <ATen/ops/cumprod_backward.h>
632
+ #include <ATen/ops/cumsum.h>
633
+ #include <ATen/ops/cumulative_trapezoid.h>
634
+ #include <ATen/ops/data.h>
635
+ #include <ATen/ops/deg2rad.h>
636
+ #include <ATen/ops/dense_dim.h>
637
+ #include <ATen/ops/dequantize.h>
638
+ #include <ATen/ops/det.h>
639
+ #include <ATen/ops/detach.h>
640
+ #include <ATen/ops/detach_copy.h>
641
+ #include <ATen/ops/diag.h>
642
+ #include <ATen/ops/diag_embed.h>
643
+ #include <ATen/ops/diagflat.h>
644
+ #include <ATen/ops/diagonal.h>
645
+ #include <ATen/ops/diagonal_backward.h>
646
+ #include <ATen/ops/diagonal_copy.h>
647
+ #include <ATen/ops/diagonal_scatter.h>
648
+ #include <ATen/ops/diff.h>
649
+ #include <ATen/ops/digamma.h>
650
+ #include <ATen/ops/dist.h>
651
+ #include <ATen/ops/div.h>
652
+ #include <ATen/ops/divide.h>
653
+ #include <ATen/ops/dot.h>
654
+ #include <ATen/ops/dropout.h>
655
+ #include <ATen/ops/dsplit.h>
656
+ #include <ATen/ops/dstack.h>
657
+ #include <ATen/ops/einsum.h>
658
+ #include <ATen/ops/elu.h>
659
+ #include <ATen/ops/elu_backward.h>
660
+ #include <ATen/ops/embedding.h>
661
+ #include <ATen/ops/embedding_backward.h>
662
+ #include <ATen/ops/embedding_bag.h>
663
+ #include <ATen/ops/embedding_dense_backward.h>
664
+ #include <ATen/ops/embedding_renorm.h>
665
+ #include <ATen/ops/embedding_sparse_backward.h>
666
+ #include <ATen/ops/empty.h>
667
+ #include <ATen/ops/empty_like.h>
668
+ #include <ATen/ops/empty_permuted.h>
669
+ #include <ATen/ops/empty_quantized.h>
670
+ #include <ATen/ops/empty_strided.h>
671
+ #include <ATen/ops/eq.h>
672
+ #include <ATen/ops/equal.h>
673
+ #include <ATen/ops/erf.h>
674
+ #include <ATen/ops/erfc.h>
675
+ #include <ATen/ops/erfinv.h>
676
+ #include <ATen/ops/exp.h>
677
+ #include <ATen/ops/exp2.h>
678
+ #include <ATen/ops/expand.h>
679
+ #include <ATen/ops/expand_as.h>
680
+ #include <ATen/ops/expand_copy.h>
681
+ #include <ATen/ops/expm1.h>
682
+ #include <ATen/ops/exponential.h>
683
+ #include <ATen/ops/eye.h>
684
+ #include <ATen/ops/fake_quantize_per_channel_affine.h>
685
+ #include <ATen/ops/fake_quantize_per_channel_affine_cachemask.h>
686
+ #include <ATen/ops/fake_quantize_per_channel_affine_cachemask_backward.h>
687
+ #include <ATen/ops/fake_quantize_per_tensor_affine.h>
688
+ #include <ATen/ops/fake_quantize_per_tensor_affine_cachemask.h>
689
+ #include <ATen/ops/fake_quantize_per_tensor_affine_cachemask_backward.h>
690
+ #include <ATen/ops/fbgemm_linear_fp16_weight.h>
691
+ #include <ATen/ops/fbgemm_linear_fp16_weight_fp32_activation.h>
692
+ #include <ATen/ops/fbgemm_linear_int8_weight.h>
693
+ #include <ATen/ops/fbgemm_linear_int8_weight_fp32_activation.h>
694
+ #include <ATen/ops/fbgemm_linear_quantize_weight.h>
695
+ #include <ATen/ops/fbgemm_pack_gemm_matrix_fp16.h>
696
+ #include <ATen/ops/fbgemm_pack_quantized_matrix.h>
697
+ #include <ATen/ops/feature_alpha_dropout.h>
698
+ #include <ATen/ops/feature_dropout.h>
699
+ #include <ATen/ops/fft_fft.h>
700
+ #include <ATen/ops/fft_fft2.h>
701
+ #include <ATen/ops/fft_fftfreq.h>
702
+ #include <ATen/ops/fft_fftn.h>
703
+ #include <ATen/ops/fft_fftshift.h>
704
+ #include <ATen/ops/fft_hfft.h>
705
+ #include <ATen/ops/fft_hfft2.h>
706
+ #include <ATen/ops/fft_hfftn.h>
707
+ #include <ATen/ops/fft_ifft.h>
708
+ #include <ATen/ops/fft_ifft2.h>
709
+ #include <ATen/ops/fft_ifftn.h>
710
+ #include <ATen/ops/fft_ifftshift.h>
711
+ #include <ATen/ops/fft_ihfft.h>
712
+ #include <ATen/ops/fft_ihfft2.h>
713
+ #include <ATen/ops/fft_ihfftn.h>
714
+ #include <ATen/ops/fft_irfft.h>
715
+ #include <ATen/ops/fft_irfft2.h>
716
+ #include <ATen/ops/fft_irfftn.h>
717
+ #include <ATen/ops/fft_rfft.h>
718
+ #include <ATen/ops/fft_rfft2.h>
719
+ #include <ATen/ops/fft_rfftfreq.h>
720
+ #include <ATen/ops/fft_rfftn.h>
721
+ #include <ATen/ops/fill.h>
722
+ #include <ATen/ops/fill_diagonal.h>
723
+ #include <ATen/ops/fix.h>
724
+ #include <ATen/ops/flatten.h>
725
+ #include <ATen/ops/flatten_dense_tensors.h>
726
+ #include <ATen/ops/flip.h>
727
+ #include <ATen/ops/fliplr.h>
728
+ #include <ATen/ops/flipud.h>
729
+ #include <ATen/ops/float_power.h>
730
+ #include <ATen/ops/floor.h>
731
+ #include <ATen/ops/floor_divide.h>
732
+ #include <ATen/ops/fmax.h>
733
+ #include <ATen/ops/fmin.h>
734
+ #include <ATen/ops/fmod.h>
735
+ #include <ATen/ops/frac.h>
736
+ #include <ATen/ops/fractional_max_pool2d.h>
737
+ #include <ATen/ops/fractional_max_pool2d_backward.h>
738
+ #include <ATen/ops/fractional_max_pool3d.h>
739
+ #include <ATen/ops/fractional_max_pool3d_backward.h>
740
+ #include <ATen/ops/frexp.h>
741
+ #include <ATen/ops/frobenius_norm.h>
742
+ #include <ATen/ops/from_file.h>
743
+ #include <ATen/ops/full.h>
744
+ #include <ATen/ops/full_like.h>
745
+ #include <ATen/ops/fused_moving_avg_obs_fake_quant.h>
746
+ #include <ATen/ops/gather.h>
747
+ #include <ATen/ops/gather_backward.h>
748
+ #include <ATen/ops/gcd.h>
749
+ #include <ATen/ops/ge.h>
750
+ #include <ATen/ops/gelu.h>
751
+ #include <ATen/ops/gelu_backward.h>
752
+ #include <ATen/ops/geometric.h>
753
+ #include <ATen/ops/geqrf.h>
754
+ #include <ATen/ops/ger.h>
755
+ #include <ATen/ops/glu.h>
756
+ #include <ATen/ops/glu_backward.h>
757
+ #include <ATen/ops/glu_backward_jvp.h>
758
+ #include <ATen/ops/glu_jvp.h>
759
+ #include <ATen/ops/gradient.h>
760
+ #include <ATen/ops/greater.h>
761
+ #include <ATen/ops/greater_equal.h>
762
+ #include <ATen/ops/grid_sampler.h>
763
+ #include <ATen/ops/grid_sampler_2d.h>
764
+ #include <ATen/ops/grid_sampler_2d_backward.h>
765
+ #include <ATen/ops/grid_sampler_3d.h>
766
+ #include <ATen/ops/grid_sampler_3d_backward.h>
767
+ #include <ATen/ops/group_norm.h>
768
+ #include <ATen/ops/gru.h>
769
+ #include <ATen/ops/gru_cell.h>
770
+ #include <ATen/ops/gt.h>
771
+ #include <ATen/ops/hamming_window.h>
772
+ #include <ATen/ops/hann_window.h>
773
+ #include <ATen/ops/hardshrink.h>
774
+ #include <ATen/ops/hardshrink_backward.h>
775
+ #include <ATen/ops/hardsigmoid.h>
776
+ #include <ATen/ops/hardsigmoid_backward.h>
777
+ #include <ATen/ops/hardswish.h>
778
+ #include <ATen/ops/hardswish_backward.h>
779
+ #include <ATen/ops/hardtanh.h>
780
+ #include <ATen/ops/hardtanh_backward.h>
781
+ #include <ATen/ops/heaviside.h>
782
+ #include <ATen/ops/hinge_embedding_loss.h>
783
+ #include <ATen/ops/histc.h>
784
+ #include <ATen/ops/histogram.h>
785
+ #include <ATen/ops/histogramdd.h>
786
+ #include <ATen/ops/hsplit.h>
787
+ #include <ATen/ops/hspmm.h>
788
+ #include <ATen/ops/hstack.h>
789
+ #include <ATen/ops/huber_loss.h>
790
+ #include <ATen/ops/huber_loss_backward.h>
791
+ #include <ATen/ops/hypot.h>
792
+ #include <ATen/ops/i0.h>
793
+ #include <ATen/ops/igamma.h>
794
+ #include <ATen/ops/igammac.h>
795
+ #include <ATen/ops/im2col.h>
796
+ #include <ATen/ops/imag.h>
797
+ #include <ATen/ops/index.h>
798
+ #include <ATen/ops/index_add.h>
799
+ #include <ATen/ops/index_copy.h>
800
+ #include <ATen/ops/index_fill.h>
801
+ #include <ATen/ops/index_put.h>
802
+ #include <ATen/ops/index_reduce.h>
803
+ #include <ATen/ops/index_select.h>
804
+ #include <ATen/ops/index_select_backward.h>
805
+ #include <ATen/ops/indices.h>
806
+ #include <ATen/ops/indices_copy.h>
807
+ #include <ATen/ops/infinitely_differentiable_gelu_backward.h>
808
+ #include <ATen/ops/inner.h>
809
+ #include <ATen/ops/instance_norm.h>
810
+ #include <ATen/ops/int_repr.h>
811
+ #include <ATen/ops/inverse.h>
812
+ #include <ATen/ops/is_coalesced.h>
813
+ #include <ATen/ops/is_complex.h>
814
+ #include <ATen/ops/is_conj.h>
815
+ #include <ATen/ops/is_distributed.h>
816
+ #include <ATen/ops/is_floating_point.h>
817
+ #include <ATen/ops/is_inference.h>
818
+ #include <ATen/ops/is_leaf.h>
819
+ #include <ATen/ops/is_neg.h>
820
+ #include <ATen/ops/is_nonzero.h>
821
+ #include <ATen/ops/is_pinned.h>
822
+ #include <ATen/ops/is_same_size.h>
823
+ #include <ATen/ops/is_set_to.h>
824
+ #include <ATen/ops/is_signed.h>
825
+ #include <ATen/ops/is_vulkan_available.h>
826
+ #include <ATen/ops/isclose.h>
827
+ #include <ATen/ops/isfinite.h>
828
+ #include <ATen/ops/isin.h>
829
+ #include <ATen/ops/isinf.h>
830
+ #include <ATen/ops/isnan.h>
831
+ #include <ATen/ops/isneginf.h>
832
+ #include <ATen/ops/isposinf.h>
833
+ #include <ATen/ops/isreal.h>
834
+ #include <ATen/ops/istft.h>
835
+ #include <ATen/ops/item.h>
836
+ #include <ATen/ops/kaiser_window.h>
837
+ #include <ATen/ops/kl_div.h>
838
+ #include <ATen/ops/kron.h>
839
+ #include <ATen/ops/kthvalue.h>
840
+ #include <ATen/ops/l1_loss.h>
841
+ #include <ATen/ops/layer_norm.h>
842
+ #include <ATen/ops/lcm.h>
843
+ #include <ATen/ops/ldexp.h>
844
+ #include <ATen/ops/le.h>
845
+ #include <ATen/ops/leaky_relu.h>
846
+ #include <ATen/ops/leaky_relu_backward.h>
847
+ #include <ATen/ops/lerp.h>
848
+ #include <ATen/ops/less.h>
849
+ #include <ATen/ops/less_equal.h>
850
+ #include <ATen/ops/lgamma.h>
851
+ #include <ATen/ops/lift.h>
852
+ #include <ATen/ops/lift_fresh.h>
853
+ #include <ATen/ops/lift_fresh_copy.h>
854
+ #include <ATen/ops/linalg_cholesky.h>
855
+ #include <ATen/ops/linalg_cholesky_ex.h>
856
+ #include <ATen/ops/linalg_cond.h>
857
+ #include <ATen/ops/linalg_cross.h>
858
+ #include <ATen/ops/linalg_det.h>
859
+ #include <ATen/ops/linalg_diagonal.h>
860
+ #include <ATen/ops/linalg_eig.h>
861
+ #include <ATen/ops/linalg_eigh.h>
862
+ #include <ATen/ops/linalg_eigvals.h>
863
+ #include <ATen/ops/linalg_eigvalsh.h>
864
+ #include <ATen/ops/linalg_householder_product.h>
865
+ #include <ATen/ops/linalg_inv.h>
866
+ #include <ATen/ops/linalg_inv_ex.h>
867
+ #include <ATen/ops/linalg_ldl_factor.h>
868
+ #include <ATen/ops/linalg_ldl_factor_ex.h>
869
+ #include <ATen/ops/linalg_ldl_solve.h>
870
+ #include <ATen/ops/linalg_lstsq.h>
871
+ #include <ATen/ops/linalg_lu.h>
872
+ #include <ATen/ops/linalg_lu_factor.h>
873
+ #include <ATen/ops/linalg_lu_factor_ex.h>
874
+ #include <ATen/ops/linalg_lu_solve.h>
875
+ #include <ATen/ops/linalg_matmul.h>
876
+ #include <ATen/ops/linalg_matrix_exp.h>
877
+ #include <ATen/ops/linalg_matrix_norm.h>
878
+ #include <ATen/ops/linalg_matrix_power.h>
879
+ #include <ATen/ops/linalg_matrix_rank.h>
880
+ #include <ATen/ops/linalg_multi_dot.h>
881
+ #include <ATen/ops/linalg_norm.h>
882
+ #include <ATen/ops/linalg_pinv.h>
883
+ #include <ATen/ops/linalg_qr.h>
884
+ #include <ATen/ops/linalg_slogdet.h>
885
+ #include <ATen/ops/linalg_solve.h>
886
+ #include <ATen/ops/linalg_solve_ex.h>
887
+ #include <ATen/ops/linalg_solve_triangular.h>
888
+ #include <ATen/ops/linalg_svd.h>
889
+ #include <ATen/ops/linalg_svdvals.h>
890
+ #include <ATen/ops/linalg_tensorinv.h>
891
+ #include <ATen/ops/linalg_tensorsolve.h>
892
+ #include <ATen/ops/linalg_vander.h>
893
+ #include <ATen/ops/linalg_vecdot.h>
894
+ #include <ATen/ops/linalg_vector_norm.h>
895
+ #include <ATen/ops/linear.h>
896
+ #include <ATen/ops/linear_backward.h>
897
+ #include <ATen/ops/linspace.h>
898
+ #include <ATen/ops/log.h>
899
+ #include <ATen/ops/log10.h>
900
+ #include <ATen/ops/log1p.h>
901
+ #include <ATen/ops/log2.h>
902
+ #include <ATen/ops/log_normal.h>
903
+ #include <ATen/ops/log_sigmoid.h>
904
+ #include <ATen/ops/log_sigmoid_backward.h>
905
+ #include <ATen/ops/log_sigmoid_forward.h>
906
+ #include <ATen/ops/log_softmax.h>
907
+ #include <ATen/ops/logaddexp.h>
908
+ #include <ATen/ops/logaddexp2.h>
909
+ #include <ATen/ops/logcumsumexp.h>
910
+ #include <ATen/ops/logdet.h>
911
+ #include <ATen/ops/logical_and.h>
912
+ #include <ATen/ops/logical_not.h>
913
+ #include <ATen/ops/logical_or.h>
914
+ #include <ATen/ops/logical_xor.h>
915
+ #include <ATen/ops/logit.h>
916
+ #include <ATen/ops/logit_backward.h>
917
+ #include <ATen/ops/logspace.h>
918
+ #include <ATen/ops/logsumexp.h>
919
+ #include <ATen/ops/lshift.h>
920
+ #include <ATen/ops/lstm.h>
921
+ #include <ATen/ops/lstm_cell.h>
922
+ #include <ATen/ops/lstm_mps_backward.h>
923
+ #include <ATen/ops/lt.h>
924
+ #include <ATen/ops/lu_solve.h>
925
+ #include <ATen/ops/lu_unpack.h>
926
+ #include <ATen/ops/mH.h>
927
+ #include <ATen/ops/mT.h>
928
+ #include <ATen/ops/margin_ranking_loss.h>
929
+ #include <ATen/ops/masked_fill.h>
930
+ #include <ATen/ops/masked_scatter.h>
931
+ #include <ATen/ops/masked_scatter_backward.h>
932
+ #include <ATen/ops/masked_select.h>
933
+ #include <ATen/ops/masked_select_backward.h>
934
+ #include <ATen/ops/matmul.h>
935
+ #include <ATen/ops/matmul_backward.h>
936
+ #include <ATen/ops/matrix_H.h>
937
+ #include <ATen/ops/matrix_exp.h>
938
+ #include <ATen/ops/matrix_exp_backward.h>
939
+ #include <ATen/ops/matrix_power.h>
940
+ #include <ATen/ops/max.h>
941
+ #include <ATen/ops/max_pool1d.h>
942
+ #include <ATen/ops/max_pool1d_with_indices.h>
943
+ #include <ATen/ops/max_pool2d.h>
944
+ #include <ATen/ops/max_pool2d_backward.h>
945
+ #include <ATen/ops/max_pool2d_with_indices.h>
946
+ #include <ATen/ops/max_pool2d_with_indices_backward.h>
947
+ #include <ATen/ops/max_pool3d.h>
948
+ #include <ATen/ops/max_pool3d_with_indices.h>
949
+ #include <ATen/ops/max_pool3d_with_indices_backward.h>
950
+ #include <ATen/ops/max_unpool2d.h>
951
+ #include <ATen/ops/max_unpool3d.h>
952
+ #include <ATen/ops/maximum.h>
953
+ #include <ATen/ops/mean.h>
954
+ #include <ATen/ops/median.h>
955
+ #include <ATen/ops/meshgrid.h>
956
+ #include <ATen/ops/min.h>
957
+ #include <ATen/ops/minimum.h>
958
+ #include <ATen/ops/miopen_batch_norm.h>
959
+ #include <ATen/ops/miopen_batch_norm_backward.h>
960
+ #include <ATen/ops/miopen_convolution.h>
961
+ #include <ATen/ops/miopen_convolution_add_relu.h>
962
+ #include <ATen/ops/miopen_convolution_relu.h>
963
+ #include <ATen/ops/miopen_convolution_transpose.h>
964
+ #include <ATen/ops/miopen_depthwise_convolution.h>
965
+ #include <ATen/ops/miopen_rnn.h>
966
+ #include <ATen/ops/miopen_rnn_backward.h>
967
+ #include <ATen/ops/mish.h>
968
+ #include <ATen/ops/mish_backward.h>
969
+ #include <ATen/ops/mkldnn_adaptive_avg_pool2d.h>
970
+ #include <ATen/ops/mkldnn_adaptive_avg_pool2d_backward.h>
971
+ #include <ATen/ops/mkldnn_convolution.h>
972
+ #include <ATen/ops/mkldnn_linear.h>
973
+ #include <ATen/ops/mkldnn_linear_backward.h>
974
+ #include <ATen/ops/mkldnn_linear_backward_input.h>
975
+ #include <ATen/ops/mkldnn_linear_backward_weights.h>
976
+ #include <ATen/ops/mkldnn_max_pool2d.h>
977
+ #include <ATen/ops/mkldnn_max_pool2d_backward.h>
978
+ #include <ATen/ops/mkldnn_max_pool3d.h>
979
+ #include <ATen/ops/mkldnn_max_pool3d_backward.h>
980
+ #include <ATen/ops/mkldnn_reorder_conv2d_weight.h>
981
+ #include <ATen/ops/mkldnn_reorder_conv3d_weight.h>
982
+ #include <ATen/ops/mkldnn_rnn_layer.h>
983
+ #include <ATen/ops/mkldnn_rnn_layer_backward.h>
984
+ #include <ATen/ops/mm.h>
985
+ #include <ATen/ops/mode.h>
986
+ #include <ATen/ops/moveaxis.h>
987
+ #include <ATen/ops/movedim.h>
988
+ #include <ATen/ops/mps_convolution_backward.h>
989
+ #include <ATen/ops/mps_convolution_transpose_backward.h>
990
+ #include <ATen/ops/mse_loss.h>
991
+ #include <ATen/ops/mse_loss_backward.h>
992
+ #include <ATen/ops/msort.h>
993
+ #include <ATen/ops/mul.h>
994
+ #include <ATen/ops/multi_margin_loss.h>
995
+ #include <ATen/ops/multi_margin_loss_backward.h>
996
+ #include <ATen/ops/multilabel_margin_loss.h>
997
+ #include <ATen/ops/multilabel_margin_loss_backward.h>
998
+ #include <ATen/ops/multilabel_margin_loss_forward.h>
999
+ #include <ATen/ops/multinomial.h>
1000
+ #include <ATen/ops/multiply.h>
1001
+ #include <ATen/ops/mv.h>
1002
+ #include <ATen/ops/mvlgamma.h>
1003
+ #include <ATen/ops/nan_to_num.h>
1004
+ #include <ATen/ops/nanmean.h>
1005
+ #include <ATen/ops/nanmedian.h>
1006
+ #include <ATen/ops/nanquantile.h>
1007
+ #include <ATen/ops/nansum.h>
1008
+ #include <ATen/ops/narrow.h>
1009
+ #include <ATen/ops/narrow_copy.h>
1010
+ #include <ATen/ops/native_batch_norm.h>
1011
+ #include <ATen/ops/native_batch_norm_backward.h>
1012
+ #include <ATen/ops/native_channel_shuffle.h>
1013
+ #include <ATen/ops/native_dropout.h>
1014
+ #include <ATen/ops/native_dropout_backward.h>
1015
+ #include <ATen/ops/native_group_norm.h>
1016
+ #include <ATen/ops/native_group_norm_backward.h>
1017
+ #include <ATen/ops/native_layer_norm.h>
1018
+ #include <ATen/ops/native_layer_norm_backward.h>
1019
+ #include <ATen/ops/native_norm.h>
1020
+ #include <ATen/ops/ne.h>
1021
+ #include <ATen/ops/neg.h>
1022
+ #include <ATen/ops/negative.h>
1023
+ #include <ATen/ops/nested_to_padded_tensor.h>
1024
+ #include <ATen/ops/new_empty.h>
1025
+ #include <ATen/ops/new_empty_strided.h>
1026
+ #include <ATen/ops/new_full.h>
1027
+ #include <ATen/ops/new_ones.h>
1028
+ #include <ATen/ops/new_zeros.h>
1029
+ #include <ATen/ops/nextafter.h>
1030
+ #include <ATen/ops/nll_loss.h>
1031
+ #include <ATen/ops/nll_loss2d.h>
1032
+ #include <ATen/ops/nll_loss2d_backward.h>
1033
+ #include <ATen/ops/nll_loss2d_forward.h>
1034
+ #include <ATen/ops/nll_loss_backward.h>
1035
+ #include <ATen/ops/nll_loss_forward.h>
1036
+ #include <ATen/ops/nll_loss_nd.h>
1037
+ #include <ATen/ops/nonzero.h>
1038
+ #include <ATen/ops/nonzero_numpy.h>
1039
+ #include <ATen/ops/nonzero_static.h>
1040
+ #include <ATen/ops/norm.h>
1041
+ #include <ATen/ops/norm_except_dim.h>
1042
+ #include <ATen/ops/normal.h>
1043
+ #include <ATen/ops/not_equal.h>
1044
+ #include <ATen/ops/nuclear_norm.h>
1045
+ #include <ATen/ops/numpy_T.h>
1046
+ #include <ATen/ops/one_hot.h>
1047
+ #include <ATen/ops/ones.h>
1048
+ #include <ATen/ops/ones_like.h>
1049
+ #include <ATen/ops/or.h>
1050
+ #include <ATen/ops/orgqr.h>
1051
+ #include <ATen/ops/ormqr.h>
1052
+ #include <ATen/ops/outer.h>
1053
+ #include <ATen/ops/output_nr.h>
1054
+ #include <ATen/ops/pad.h>
1055
+ #include <ATen/ops/pad_sequence.h>
1056
+ #include <ATen/ops/pairwise_distance.h>
1057
+ #include <ATen/ops/pdist.h>
1058
+ #include <ATen/ops/permute.h>
1059
+ #include <ATen/ops/permute_copy.h>
1060
+ #include <ATen/ops/pin_memory.h>
1061
+ #include <ATen/ops/pinverse.h>
1062
+ #include <ATen/ops/pixel_shuffle.h>
1063
+ #include <ATen/ops/pixel_unshuffle.h>
1064
+ #include <ATen/ops/poisson.h>
1065
+ #include <ATen/ops/poisson_nll_loss.h>
1066
+ #include <ATen/ops/polar.h>
1067
+ #include <ATen/ops/polygamma.h>
1068
+ #include <ATen/ops/positive.h>
1069
+ #include <ATen/ops/pow.h>
1070
+ #include <ATen/ops/prelu.h>
1071
+ #include <ATen/ops/prod.h>
1072
+ #include <ATen/ops/promote_types.h>
1073
+ #include <ATen/ops/put.h>
1074
+ #include <ATen/ops/q_per_channel_axis.h>
1075
+ #include <ATen/ops/q_per_channel_scales.h>
1076
+ #include <ATen/ops/q_per_channel_zero_points.h>
1077
+ #include <ATen/ops/q_scale.h>
1078
+ #include <ATen/ops/q_zero_point.h>
1079
+ #include <ATen/ops/qr.h>
1080
+ #include <ATen/ops/qscheme.h>
1081
+ #include <ATen/ops/quantile.h>
1082
+ #include <ATen/ops/quantize_per_channel.h>
1083
+ #include <ATen/ops/quantize_per_tensor.h>
1084
+ #include <ATen/ops/quantize_per_tensor_dynamic.h>
1085
+ #include <ATen/ops/quantized_batch_norm.h>
1086
+ #include <ATen/ops/quantized_gru_cell.h>
1087
+ #include <ATen/ops/quantized_lstm_cell.h>
1088
+ #include <ATen/ops/quantized_max_pool1d.h>
1089
+ #include <ATen/ops/quantized_max_pool2d.h>
1090
+ #include <ATen/ops/quantized_max_pool3d.h>
1091
+ #include <ATen/ops/quantized_rnn_relu_cell.h>
1092
+ #include <ATen/ops/quantized_rnn_tanh_cell.h>
1093
+ #include <ATen/ops/rad2deg.h>
1094
+ #include <ATen/ops/rand.h>
1095
+ #include <ATen/ops/rand_like.h>
1096
+ #include <ATen/ops/randint.h>
1097
+ #include <ATen/ops/randint_like.h>
1098
+ #include <ATen/ops/randn.h>
1099
+ #include <ATen/ops/randn_like.h>
1100
+ #include <ATen/ops/random.h>
1101
+ #include <ATen/ops/randperm.h>
1102
+ #include <ATen/ops/range.h>
1103
+ #include <ATen/ops/ravel.h>
1104
+ #include <ATen/ops/real.h>
1105
+ #include <ATen/ops/reciprocal.h>
1106
+ #include <ATen/ops/record_stream.h>
1107
+ #include <ATen/ops/refine_names.h>
1108
+ #include <ATen/ops/reflection_pad1d.h>
1109
+ #include <ATen/ops/reflection_pad1d_backward.h>
1110
+ #include <ATen/ops/reflection_pad2d.h>
1111
+ #include <ATen/ops/reflection_pad2d_backward.h>
1112
+ #include <ATen/ops/reflection_pad3d.h>
1113
+ #include <ATen/ops/reflection_pad3d_backward.h>
1114
+ #include <ATen/ops/relu.h>
1115
+ #include <ATen/ops/relu6.h>
1116
+ #include <ATen/ops/remainder.h>
1117
+ #include <ATen/ops/rename.h>
1118
+ #include <ATen/ops/renorm.h>
1119
+ #include <ATen/ops/repeat.h>
1120
+ #include <ATen/ops/repeat_interleave.h>
1121
+ #include <ATen/ops/replication_pad1d.h>
1122
+ #include <ATen/ops/replication_pad1d_backward.h>
1123
+ #include <ATen/ops/replication_pad2d.h>
1124
+ #include <ATen/ops/replication_pad2d_backward.h>
1125
+ #include <ATen/ops/replication_pad3d.h>
1126
+ #include <ATen/ops/replication_pad3d_backward.h>
1127
+ #include <ATen/ops/requires_grad.h>
1128
+ #include <ATen/ops/reshape.h>
1129
+ #include <ATen/ops/reshape_as.h>
1130
+ #include <ATen/ops/resize.h>
1131
+ #include <ATen/ops/resize_as.h>
1132
+ #include <ATen/ops/resize_as_sparse.h>
1133
+ #include <ATen/ops/resolve_conj.h>
1134
+ #include <ATen/ops/resolve_neg.h>
1135
+ #include <ATen/ops/result_type.h>
1136
+ #include <ATen/ops/retain_grad.h>
1137
+ #include <ATen/ops/retains_grad.h>
1138
+ #include <ATen/ops/rms_norm.h>
1139
+ #include <ATen/ops/rnn_relu.h>
1140
+ #include <ATen/ops/rnn_relu_cell.h>
1141
+ #include <ATen/ops/rnn_tanh.h>
1142
+ #include <ATen/ops/rnn_tanh_cell.h>
1143
+ #include <ATen/ops/roll.h>
1144
+ #include <ATen/ops/rot90.h>
1145
+ #include <ATen/ops/round.h>
1146
+ #include <ATen/ops/row_indices.h>
1147
+ #include <ATen/ops/row_indices_copy.h>
1148
+ #include <ATen/ops/row_stack.h>
1149
+ #include <ATen/ops/rrelu.h>
1150
+ #include <ATen/ops/rrelu_with_noise.h>
1151
+ #include <ATen/ops/rrelu_with_noise_backward.h>
1152
+ #include <ATen/ops/rshift.h>
1153
+ #include <ATen/ops/rsqrt.h>
1154
+ #include <ATen/ops/rsub.h>
1155
+ #include <ATen/ops/scalar_tensor.h>
1156
+ #include <ATen/ops/scaled_dot_product_attention.h>
1157
+ #include <ATen/ops/scatter.h>
1158
+ #include <ATen/ops/scatter_add.h>
1159
+ #include <ATen/ops/scatter_reduce.h>
1160
+ #include <ATen/ops/searchsorted.h>
1161
+ #include <ATen/ops/segment_reduce.h>
1162
+ #include <ATen/ops/select.h>
1163
+ #include <ATen/ops/select_backward.h>
1164
+ #include <ATen/ops/select_copy.h>
1165
+ #include <ATen/ops/select_scatter.h>
1166
+ #include <ATen/ops/selu.h>
1167
+ #include <ATen/ops/set.h>
1168
+ #include <ATen/ops/set_data.h>
1169
+ #include <ATen/ops/sgn.h>
1170
+ #include <ATen/ops/sigmoid.h>
1171
+ #include <ATen/ops/sigmoid_backward.h>
1172
+ #include <ATen/ops/sign.h>
1173
+ #include <ATen/ops/signbit.h>
1174
+ #include <ATen/ops/silu.h>
1175
+ #include <ATen/ops/silu_backward.h>
1176
+ #include <ATen/ops/sin.h>
1177
+ #include <ATen/ops/sinc.h>
1178
+ #include <ATen/ops/sinh.h>
1179
+ #include <ATen/ops/size.h>
1180
+ #include <ATen/ops/slice.h>
1181
+ #include <ATen/ops/slice_backward.h>
1182
+ #include <ATen/ops/slice_copy.h>
1183
+ #include <ATen/ops/slice_inverse.h>
1184
+ #include <ATen/ops/slice_scatter.h>
1185
+ #include <ATen/ops/slogdet.h>
1186
+ #include <ATen/ops/slow_conv3d.h>
1187
+ #include <ATen/ops/slow_conv3d_forward.h>
1188
+ #include <ATen/ops/slow_conv_dilated2d.h>
1189
+ #include <ATen/ops/slow_conv_dilated3d.h>
1190
+ #include <ATen/ops/slow_conv_transpose2d.h>
1191
+ #include <ATen/ops/slow_conv_transpose3d.h>
1192
+ #include <ATen/ops/smm.h>
1193
+ #include <ATen/ops/smooth_l1_loss.h>
1194
+ #include <ATen/ops/smooth_l1_loss_backward.h>
1195
+ #include <ATen/ops/soft_margin_loss.h>
1196
+ #include <ATen/ops/soft_margin_loss_backward.h>
1197
+ #include <ATen/ops/softmax.h>
1198
+ #include <ATen/ops/softplus.h>
1199
+ #include <ATen/ops/softplus_backward.h>
1200
+ #include <ATen/ops/softshrink.h>
1201
+ #include <ATen/ops/softshrink_backward.h>
1202
+ #include <ATen/ops/sort.h>
1203
+ #include <ATen/ops/sparse_bsc_tensor.h>
1204
+ #include <ATen/ops/sparse_bsr_tensor.h>
1205
+ #include <ATen/ops/sparse_compressed_tensor.h>
1206
+ #include <ATen/ops/sparse_coo_tensor.h>
1207
+ #include <ATen/ops/sparse_csc_tensor.h>
1208
+ #include <ATen/ops/sparse_csr_tensor.h>
1209
+ #include <ATen/ops/sparse_dim.h>
1210
+ #include <ATen/ops/sparse_mask.h>
1211
+ #include <ATen/ops/sparse_resize.h>
1212
+ #include <ATen/ops/sparse_resize_and_clear.h>
1213
+ #include <ATen/ops/sparse_sampled_addmm.h>
1214
+ #include <ATen/ops/special_airy_ai.h>
1215
+ #include <ATen/ops/special_bessel_j0.h>
1216
+ #include <ATen/ops/special_bessel_j1.h>
1217
+ #include <ATen/ops/special_bessel_y0.h>
1218
+ #include <ATen/ops/special_bessel_y1.h>
1219
+ #include <ATen/ops/special_chebyshev_polynomial_t.h>
1220
+ #include <ATen/ops/special_chebyshev_polynomial_u.h>
1221
+ #include <ATen/ops/special_chebyshev_polynomial_v.h>
1222
+ #include <ATen/ops/special_chebyshev_polynomial_w.h>
1223
+ #include <ATen/ops/special_digamma.h>
1224
+ #include <ATen/ops/special_entr.h>
1225
+ #include <ATen/ops/special_erf.h>
1226
+ #include <ATen/ops/special_erfc.h>
1227
+ #include <ATen/ops/special_erfcx.h>
1228
+ #include <ATen/ops/special_erfinv.h>
1229
+ #include <ATen/ops/special_exp2.h>
1230
+ #include <ATen/ops/special_expit.h>
1231
+ #include <ATen/ops/special_expm1.h>
1232
+ #include <ATen/ops/special_gammainc.h>
1233
+ #include <ATen/ops/special_gammaincc.h>
1234
+ #include <ATen/ops/special_gammaln.h>
1235
+ #include <ATen/ops/special_hermite_polynomial_h.h>
1236
+ #include <ATen/ops/special_hermite_polynomial_he.h>
1237
+ #include <ATen/ops/special_i0.h>
1238
+ #include <ATen/ops/special_i0e.h>
1239
+ #include <ATen/ops/special_i1.h>
1240
+ #include <ATen/ops/special_i1e.h>
1241
+ #include <ATen/ops/special_laguerre_polynomial_l.h>
1242
+ #include <ATen/ops/special_legendre_polynomial_p.h>
1243
+ #include <ATen/ops/special_log1p.h>
1244
+ #include <ATen/ops/special_log_ndtr.h>
1245
+ #include <ATen/ops/special_log_softmax.h>
1246
+ #include <ATen/ops/special_logit.h>
1247
+ #include <ATen/ops/special_logsumexp.h>
1248
+ #include <ATen/ops/special_modified_bessel_i0.h>
1249
+ #include <ATen/ops/special_modified_bessel_i1.h>
1250
+ #include <ATen/ops/special_modified_bessel_k0.h>
1251
+ #include <ATen/ops/special_modified_bessel_k1.h>
1252
+ #include <ATen/ops/special_multigammaln.h>
1253
+ #include <ATen/ops/special_ndtr.h>
1254
+ #include <ATen/ops/special_ndtri.h>
1255
+ #include <ATen/ops/special_polygamma.h>
1256
+ #include <ATen/ops/special_psi.h>
1257
+ #include <ATen/ops/special_round.h>
1258
+ #include <ATen/ops/special_scaled_modified_bessel_k0.h>
1259
+ #include <ATen/ops/special_scaled_modified_bessel_k1.h>
1260
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_t.h>
1261
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_u.h>
1262
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_v.h>
1263
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_w.h>
1264
+ #include <ATen/ops/special_sinc.h>
1265
+ #include <ATen/ops/special_softmax.h>
1266
+ #include <ATen/ops/special_spherical_bessel_j0.h>
1267
+ #include <ATen/ops/special_xlog1py.h>
1268
+ #include <ATen/ops/special_xlogy.h>
1269
+ #include <ATen/ops/special_zeta.h>
1270
+ #include <ATen/ops/split.h>
1271
+ #include <ATen/ops/split_copy.h>
1272
+ #include <ATen/ops/split_with_sizes.h>
1273
+ #include <ATen/ops/split_with_sizes_copy.h>
1274
+ #include <ATen/ops/sqrt.h>
1275
+ #include <ATen/ops/square.h>
1276
+ #include <ATen/ops/squeeze.h>
1277
+ #include <ATen/ops/squeeze_copy.h>
1278
+ #include <ATen/ops/sspaddmm.h>
1279
+ #include <ATen/ops/stack.h>
1280
+ #include <ATen/ops/std.h>
1281
+ #include <ATen/ops/std_mean.h>
1282
+ #include <ATen/ops/stft.h>
1283
+ #include <ATen/ops/stride.h>
1284
+ #include <ATen/ops/sub.h>
1285
+ #include <ATen/ops/subtract.h>
1286
+ #include <ATen/ops/sum.h>
1287
+ #include <ATen/ops/sum_to_size.h>
1288
+ #include <ATen/ops/svd.h>
1289
+ #include <ATen/ops/swapaxes.h>
1290
+ #include <ATen/ops/swapdims.h>
1291
+ #include <ATen/ops/sym_constrain_range.h>
1292
+ #include <ATen/ops/sym_constrain_range_for_size.h>
1293
+ #include <ATen/ops/sym_numel.h>
1294
+ #include <ATen/ops/sym_size.h>
1295
+ #include <ATen/ops/sym_storage_offset.h>
1296
+ #include <ATen/ops/sym_stride.h>
1297
+ #include <ATen/ops/t.h>
1298
+ #include <ATen/ops/t_copy.h>
1299
+ #include <ATen/ops/take.h>
1300
+ #include <ATen/ops/take_along_dim.h>
1301
+ #include <ATen/ops/tan.h>
1302
+ #include <ATen/ops/tanh.h>
1303
+ #include <ATen/ops/tanh_backward.h>
1304
+ #include <ATen/ops/tensor_split.h>
1305
+ #include <ATen/ops/tensordot.h>
1306
+ #include <ATen/ops/thnn_conv2d.h>
1307
+ #include <ATen/ops/threshold.h>
1308
+ #include <ATen/ops/threshold_backward.h>
1309
+ #include <ATen/ops/tile.h>
1310
+ #include <ATen/ops/to.h>
1311
+ #include <ATen/ops/to_dense.h>
1312
+ #include <ATen/ops/to_dense_backward.h>
1313
+ #include <ATen/ops/to_mkldnn.h>
1314
+ #include <ATen/ops/to_mkldnn_backward.h>
1315
+ #include <ATen/ops/to_padded_tensor.h>
1316
+ #include <ATen/ops/to_sparse.h>
1317
+ #include <ATen/ops/to_sparse_bsc.h>
1318
+ #include <ATen/ops/to_sparse_bsr.h>
1319
+ #include <ATen/ops/to_sparse_csc.h>
1320
+ #include <ATen/ops/to_sparse_csr.h>
1321
+ #include <ATen/ops/topk.h>
1322
+ #include <ATen/ops/trace.h>
1323
+ #include <ATen/ops/trace_backward.h>
1324
+ #include <ATen/ops/transpose.h>
1325
+ #include <ATen/ops/transpose_copy.h>
1326
+ #include <ATen/ops/trapezoid.h>
1327
+ #include <ATen/ops/trapz.h>
1328
+ #include <ATen/ops/triangular_solve.h>
1329
+ #include <ATen/ops/tril.h>
1330
+ #include <ATen/ops/tril_indices.h>
1331
+ #include <ATen/ops/triplet_margin_loss.h>
1332
+ #include <ATen/ops/triu.h>
1333
+ #include <ATen/ops/triu_indices.h>
1334
+ #include <ATen/ops/true_divide.h>
1335
+ #include <ATen/ops/trunc.h>
1336
+ #include <ATen/ops/type_as.h>
1337
+ #include <ATen/ops/unbind.h>
1338
+ #include <ATen/ops/unbind_copy.h>
1339
+ #include <ATen/ops/unflatten.h>
1340
+ #include <ATen/ops/unflatten_dense_tensors.h>
1341
+ #include <ATen/ops/unfold.h>
1342
+ #include <ATen/ops/unfold_backward.h>
1343
+ #include <ATen/ops/unfold_copy.h>
1344
+ #include <ATen/ops/uniform.h>
1345
+ #include <ATen/ops/unique_consecutive.h>
1346
+ #include <ATen/ops/unique_dim.h>
1347
+ #include <ATen/ops/unique_dim_consecutive.h>
1348
+ #include <ATen/ops/unsafe_chunk.h>
1349
+ #include <ATen/ops/unsafe_split.h>
1350
+ #include <ATen/ops/unsafe_split_with_sizes.h>
1351
+ #include <ATen/ops/unsqueeze.h>
1352
+ #include <ATen/ops/unsqueeze_copy.h>
1353
+ #include <ATen/ops/upsample_bicubic2d.h>
1354
+ #include <ATen/ops/upsample_bicubic2d_backward.h>
1355
+ #include <ATen/ops/upsample_bilinear2d.h>
1356
+ #include <ATen/ops/upsample_bilinear2d_backward.h>
1357
+ #include <ATen/ops/upsample_linear1d.h>
1358
+ #include <ATen/ops/upsample_linear1d_backward.h>
1359
+ #include <ATen/ops/upsample_nearest1d.h>
1360
+ #include <ATen/ops/upsample_nearest1d_backward.h>
1361
+ #include <ATen/ops/upsample_nearest2d.h>
1362
+ #include <ATen/ops/upsample_nearest2d_backward.h>
1363
+ #include <ATen/ops/upsample_nearest3d.h>
1364
+ #include <ATen/ops/upsample_nearest3d_backward.h>
1365
+ #include <ATen/ops/upsample_trilinear3d.h>
1366
+ #include <ATen/ops/upsample_trilinear3d_backward.h>
1367
+ #include <ATen/ops/value_selecting_reduction_backward.h>
1368
+ #include <ATen/ops/values.h>
1369
+ #include <ATen/ops/values_copy.h>
1370
+ #include <ATen/ops/vander.h>
1371
+ #include <ATen/ops/var.h>
1372
+ #include <ATen/ops/var_mean.h>
1373
+ #include <ATen/ops/vdot.h>
1374
+ #include <ATen/ops/view.h>
1375
+ #include <ATen/ops/view_as.h>
1376
+ #include <ATen/ops/view_as_complex.h>
1377
+ #include <ATen/ops/view_as_complex_copy.h>
1378
+ #include <ATen/ops/view_as_real.h>
1379
+ #include <ATen/ops/view_as_real_copy.h>
1380
+ #include <ATen/ops/view_copy.h>
1381
+ #include <ATen/ops/vsplit.h>
1382
+ #include <ATen/ops/vstack.h>
1383
+ #include <ATen/ops/where.h>
1384
+ #include <ATen/ops/xlogy.h>
1385
+ #include <ATen/ops/xor.h>
1386
+ #include <ATen/ops/zero.h>
1387
+ #include <ATen/ops/zeros.h>
1388
+ #include <ATen/ops/zeros_like.h>
1389
+
1390
+ namespace at {
1391
+
1392
+
1393
+
1394
+ // Special C++ only overloads for std()-like functions (See gh-40287)
1395
+ // These are needed because int -> bool conversion takes precedence over int -> IntArrayRef
1396
+ // So, for example std(0) would select the std(unbiased=False) overload
1397
+ TORCH_API inline Tensor var(const Tensor& self, int dim) {
1398
+ return at::var(self, IntArrayRef{dim});
1399
+ }
1400
+ TORCH_API inline std::tuple<Tensor, Tensor> var_mean(const Tensor& self, int dim) {
1401
+ return at::var_mean(self, IntArrayRef{dim});
1402
+ }
1403
+ TORCH_API inline Tensor std(const Tensor& self, int dim) {
1404
+ return at::std(self, IntArrayRef{dim});
1405
+ }
1406
+ TORCH_API inline std::tuple<Tensor, Tensor> std_mean(const Tensor& self, int dim) {
1407
+ return at::std_mean(self, IntArrayRef{dim});
1408
+ }
1409
+
1410
+ inline int64_t numel(const Tensor& tensor) {
1411
+ return tensor.numel();
1412
+ }
1413
+
1414
+ inline int64_t size(const Tensor& tensor, int64_t dim) {
1415
+ return tensor.size(dim);
1416
+ }
1417
+
1418
+ inline int64_t stride(const Tensor& tensor, int64_t dim) {
1419
+ return tensor.stride(dim);
1420
+ }
1421
+
1422
+ inline bool is_complex(const Tensor& tensor) {
1423
+ return tensor.is_complex();
1424
+ }
1425
+
1426
+ inline bool is_floating_point(const Tensor& tensor) {
1427
+ return tensor.is_floating_point();
1428
+ }
1429
+
1430
+ inline bool is_signed(const Tensor& tensor) {
1431
+ return tensor.is_signed();
1432
+ }
1433
+
1434
+ inline bool is_inference(const Tensor& tensor) {
1435
+ return tensor.is_inference();
1436
+ }
1437
+
1438
+ inline bool _is_zerotensor(const Tensor& tensor) {
1439
+ return tensor._is_zerotensor();
1440
+ }
1441
+
1442
+ inline bool is_conj(const Tensor& tensor) {
1443
+ return tensor.is_conj();
1444
+ }
1445
+
1446
+ inline Tensor conj(const Tensor& tensor) {
1447
+ return tensor.conj();
1448
+ }
1449
+
1450
+ inline bool is_neg(const Tensor& tensor) {
1451
+ return tensor.is_neg();
1452
+ }
1453
+
1454
+ }
.venv/lib/python3.11/site-packages/torch/include/ATen/Generator.h ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ #pragma once
2
+ #include <ATen/core/Generator.h>
.venv/lib/python3.11/site-packages/torch/include/ATen/LinalgBackend.h ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <c10/util/Exception.h>
4
+
5
+ #include <ostream>
6
+ #include <string>
7
+
8
+ namespace at {
9
+
10
+ enum class LinalgBackend : int8_t { Default, Cusolver, Magma };
11
+
12
+ inline std::string LinalgBackendToString(at::LinalgBackend backend) {
13
+ switch (backend) {
14
+ case LinalgBackend::Default:
15
+ return "at::LinalgBackend::Default";
16
+ case LinalgBackend::Cusolver:
17
+ return "at::LinalgBackend::Cusolver";
18
+ case LinalgBackend::Magma:
19
+ return "at::LinalgBackend::Magma";
20
+ default:
21
+ TORCH_CHECK(false, "Unknown linalg backend");
22
+ }
23
+ }
24
+
25
+ inline std::ostream& operator<<(
26
+ std::ostream& stream,
27
+ at::LinalgBackend backend) {
28
+ return stream << LinalgBackendToString(backend);
29
+ }
30
+
31
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/MemoryOverlap.h ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <c10/macros/Export.h>
4
+
5
+ namespace c10 {
6
+ struct TensorImpl;
7
+ }
8
+
9
+ namespace at {
10
+ class TensorBase;
11
+
12
+ // MemOverlap: Whether or not there is memory overlap
13
+ //
14
+ // No: Absolutely no memory overlap
15
+ // Yes: Absolutely yes memory overlap
16
+ // TooHard: There might be memory overlap, but it was too expensive to compute.
17
+ //
18
+ // NB: Please update the python test for these if you renumber them.
19
+ enum class MemOverlap { No, Yes, TooHard };
20
+
21
+ enum class MemOverlapStatus { Full, Partial, No, TooHard };
22
+
23
+ TORCH_API MemOverlap has_internal_overlap(const TensorBase& t);
24
+ TORCH_API MemOverlap has_internal_overlap(c10::TensorImpl* t);
25
+
26
+ TORCH_API void assert_no_internal_overlap(const TensorBase& t);
27
+ TORCH_API void assert_no_internal_overlap(c10::TensorImpl* t);
28
+
29
+ TORCH_API MemOverlapStatus
30
+ get_overlap_status(const TensorBase& a, const TensorBase& b);
31
+ TORCH_API MemOverlapStatus
32
+ get_overlap_status(const c10::TensorImpl* a, const c10::TensorImpl* b);
33
+
34
+ TORCH_API void assert_no_partial_overlap(
35
+ const TensorBase& a,
36
+ const TensorBase& b);
37
+ void assert_no_partial_overlap(c10::TensorImpl* a, c10::TensorImpl* b);
38
+
39
+ TORCH_API void assert_no_overlap(const TensorBase& a, const TensorBase& b);
40
+ TORCH_API void assert_no_overlap(c10::TensorImpl* a, c10::TensorImpl* b);
41
+
42
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/NativeMetaFunctions.h ADDED
@@ -0,0 +1,1330 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ // @generated by torchgen/gen.py from NativeMetaFunctions.h
4
+
5
+ #include <ATen/core/Tensor.h>
6
+ #include <ATen/core/IListRef.h>
7
+ #include <ATen/TensorMeta.h>
8
+ #include <ATen/TensorIterator.h>
9
+
10
+ #include <ATen/ops/_adaptive_avg_pool2d_meta.h>
11
+ #include <ATen/ops/_adaptive_avg_pool2d_backward_meta.h>
12
+ #include <ATen/ops/_adaptive_avg_pool3d_meta.h>
13
+ #include <ATen/ops/_adaptive_avg_pool3d_backward_meta.h>
14
+ #include <ATen/ops/_add_batch_dim_meta.h>
15
+ #include <ATen/ops/_add_relu_meta.h>
16
+ #include <ATen/ops/_addmm_activation_meta.h>
17
+ #include <ATen/ops/_aminmax_meta.h>
18
+ #include <ATen/ops/_amp_foreach_non_finite_check_and_unscale_meta.h>
19
+ #include <ATen/ops/_amp_update_scale_meta.h>
20
+ #include <ATen/ops/_assert_async_meta.h>
21
+ #include <ATen/ops/_assert_scalar_meta.h>
22
+ #include <ATen/ops/_assert_tensor_metadata_meta.h>
23
+ #include <ATen/ops/_autocast_to_full_precision_meta.h>
24
+ #include <ATen/ops/_autocast_to_reduced_precision_meta.h>
25
+ #include <ATen/ops/_backward_meta.h>
26
+ #include <ATen/ops/_batch_norm_impl_index_meta.h>
27
+ #include <ATen/ops/_batch_norm_impl_index_backward_meta.h>
28
+ #include <ATen/ops/_batch_norm_no_update_meta.h>
29
+ #include <ATen/ops/_batch_norm_with_update_meta.h>
30
+ #include <ATen/ops/_cast_Byte_meta.h>
31
+ #include <ATen/ops/_cast_Char_meta.h>
32
+ #include <ATen/ops/_cast_Double_meta.h>
33
+ #include <ATen/ops/_cast_Float_meta.h>
34
+ #include <ATen/ops/_cast_Half_meta.h>
35
+ #include <ATen/ops/_cast_Int_meta.h>
36
+ #include <ATen/ops/_cast_Long_meta.h>
37
+ #include <ATen/ops/_cast_Short_meta.h>
38
+ #include <ATen/ops/_cdist_backward_meta.h>
39
+ #include <ATen/ops/_cdist_forward_meta.h>
40
+ #include <ATen/ops/_cholesky_solve_helper_meta.h>
41
+ #include <ATen/ops/_choose_qparams_per_tensor_meta.h>
42
+ #include <ATen/ops/_chunk_cat_meta.h>
43
+ #include <ATen/ops/_coalesce_meta.h>
44
+ #include <ATen/ops/_coalesced_meta.h>
45
+ #include <ATen/ops/_compute_linear_combination_meta.h>
46
+ #include <ATen/ops/_conj_meta.h>
47
+ #include <ATen/ops/_conj_copy_meta.h>
48
+ #include <ATen/ops/_conj_physical_meta.h>
49
+ #include <ATen/ops/_conv_depthwise2d_meta.h>
50
+ #include <ATen/ops/_convert_indices_from_coo_to_csr_meta.h>
51
+ #include <ATen/ops/_convert_indices_from_csr_to_coo_meta.h>
52
+ #include <ATen/ops/_convert_weight_to_int4pack_meta.h>
53
+ #include <ATen/ops/_convolution_meta.h>
54
+ #include <ATen/ops/_convolution_double_backward_meta.h>
55
+ #include <ATen/ops/_convolution_mode_meta.h>
56
+ #include <ATen/ops/_copy_from_meta.h>
57
+ #include <ATen/ops/_copy_from_and_resize_meta.h>
58
+ #include <ATen/ops/_cslt_compress_meta.h>
59
+ #include <ATen/ops/_cslt_sparse_mm_meta.h>
60
+ #include <ATen/ops/_cslt_sparse_mm_search_meta.h>
61
+ #include <ATen/ops/_ctc_loss_meta.h>
62
+ #include <ATen/ops/_ctc_loss_backward_meta.h>
63
+ #include <ATen/ops/_cudnn_ctc_loss_meta.h>
64
+ #include <ATen/ops/_cudnn_init_dropout_state_meta.h>
65
+ #include <ATen/ops/_cudnn_rnn_meta.h>
66
+ #include <ATen/ops/_cudnn_rnn_backward_meta.h>
67
+ #include <ATen/ops/_cudnn_rnn_flatten_weight_meta.h>
68
+ #include <ATen/ops/_cufft_clear_plan_cache_meta.h>
69
+ #include <ATen/ops/_cufft_get_plan_cache_max_size_meta.h>
70
+ #include <ATen/ops/_cufft_get_plan_cache_size_meta.h>
71
+ #include <ATen/ops/_cufft_set_plan_cache_max_size_meta.h>
72
+ #include <ATen/ops/_cummax_helper_meta.h>
73
+ #include <ATen/ops/_cummin_helper_meta.h>
74
+ #include <ATen/ops/_debug_has_internal_overlap_meta.h>
75
+ #include <ATen/ops/_dimI_meta.h>
76
+ #include <ATen/ops/_dimV_meta.h>
77
+ #include <ATen/ops/_dim_arange_meta.h>
78
+ #include <ATen/ops/_dirichlet_grad_meta.h>
79
+ #include <ATen/ops/_efficient_attention_backward_meta.h>
80
+ #include <ATen/ops/_efficient_attention_forward_meta.h>
81
+ #include <ATen/ops/_efficientzerotensor_meta.h>
82
+ #include <ATen/ops/_embedding_bag_meta.h>
83
+ #include <ATen/ops/_embedding_bag_backward_meta.h>
84
+ #include <ATen/ops/_embedding_bag_dense_backward_meta.h>
85
+ #include <ATen/ops/_embedding_bag_forward_only_meta.h>
86
+ #include <ATen/ops/_embedding_bag_per_sample_weights_backward_meta.h>
87
+ #include <ATen/ops/_embedding_bag_sparse_backward_meta.h>
88
+ #include <ATen/ops/_empty_affine_quantized_meta.h>
89
+ #include <ATen/ops/_empty_per_channel_affine_quantized_meta.h>
90
+ #include <ATen/ops/_euclidean_dist_meta.h>
91
+ #include <ATen/ops/_fake_quantize_learnable_per_channel_affine_meta.h>
92
+ #include <ATen/ops/_fake_quantize_learnable_per_channel_affine_backward_meta.h>
93
+ #include <ATen/ops/_fake_quantize_learnable_per_tensor_affine_meta.h>
94
+ #include <ATen/ops/_fake_quantize_learnable_per_tensor_affine_backward_meta.h>
95
+ #include <ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams_meta.h>
96
+ #include <ATen/ops/_fft_c2c_meta.h>
97
+ #include <ATen/ops/_fft_c2r_meta.h>
98
+ #include <ATen/ops/_fft_r2c_meta.h>
99
+ #include <ATen/ops/_fill_mem_eff_dropout_mask_meta.h>
100
+ #include <ATen/ops/_flash_attention_backward_meta.h>
101
+ #include <ATen/ops/_flash_attention_forward_meta.h>
102
+ #include <ATen/ops/_foobar_meta.h>
103
+ #include <ATen/ops/_foreach_abs_meta.h>
104
+ #include <ATen/ops/_foreach_acos_meta.h>
105
+ #include <ATen/ops/_foreach_add_meta.h>
106
+ #include <ATen/ops/_foreach_addcdiv_meta.h>
107
+ #include <ATen/ops/_foreach_addcmul_meta.h>
108
+ #include <ATen/ops/_foreach_asin_meta.h>
109
+ #include <ATen/ops/_foreach_atan_meta.h>
110
+ #include <ATen/ops/_foreach_ceil_meta.h>
111
+ #include <ATen/ops/_foreach_clamp_max_meta.h>
112
+ #include <ATen/ops/_foreach_clamp_min_meta.h>
113
+ #include <ATen/ops/_foreach_copy_meta.h>
114
+ #include <ATen/ops/_foreach_cos_meta.h>
115
+ #include <ATen/ops/_foreach_cosh_meta.h>
116
+ #include <ATen/ops/_foreach_div_meta.h>
117
+ #include <ATen/ops/_foreach_erf_meta.h>
118
+ #include <ATen/ops/_foreach_erfc_meta.h>
119
+ #include <ATen/ops/_foreach_exp_meta.h>
120
+ #include <ATen/ops/_foreach_expm1_meta.h>
121
+ #include <ATen/ops/_foreach_floor_meta.h>
122
+ #include <ATen/ops/_foreach_frac_meta.h>
123
+ #include <ATen/ops/_foreach_lerp_meta.h>
124
+ #include <ATen/ops/_foreach_lgamma_meta.h>
125
+ #include <ATen/ops/_foreach_log_meta.h>
126
+ #include <ATen/ops/_foreach_log10_meta.h>
127
+ #include <ATen/ops/_foreach_log1p_meta.h>
128
+ #include <ATen/ops/_foreach_log2_meta.h>
129
+ #include <ATen/ops/_foreach_max_meta.h>
130
+ #include <ATen/ops/_foreach_maximum_meta.h>
131
+ #include <ATen/ops/_foreach_minimum_meta.h>
132
+ #include <ATen/ops/_foreach_mul_meta.h>
133
+ #include <ATen/ops/_foreach_neg_meta.h>
134
+ #include <ATen/ops/_foreach_norm_meta.h>
135
+ #include <ATen/ops/_foreach_pow_meta.h>
136
+ #include <ATen/ops/_foreach_reciprocal_meta.h>
137
+ #include <ATen/ops/_foreach_round_meta.h>
138
+ #include <ATen/ops/_foreach_sigmoid_meta.h>
139
+ #include <ATen/ops/_foreach_sign_meta.h>
140
+ #include <ATen/ops/_foreach_sin_meta.h>
141
+ #include <ATen/ops/_foreach_sinh_meta.h>
142
+ #include <ATen/ops/_foreach_sqrt_meta.h>
143
+ #include <ATen/ops/_foreach_sub_meta.h>
144
+ #include <ATen/ops/_foreach_tan_meta.h>
145
+ #include <ATen/ops/_foreach_tanh_meta.h>
146
+ #include <ATen/ops/_foreach_trunc_meta.h>
147
+ #include <ATen/ops/_foreach_zero_meta.h>
148
+ #include <ATen/ops/_functional_assert_async_meta.h>
149
+ #include <ATen/ops/_functional_assert_scalar_meta.h>
150
+ #include <ATen/ops/_functional_sym_constrain_range_meta.h>
151
+ #include <ATen/ops/_functional_sym_constrain_range_for_size_meta.h>
152
+ #include <ATen/ops/_fused_adagrad_meta.h>
153
+ #include <ATen/ops/_fused_adam_meta.h>
154
+ #include <ATen/ops/_fused_adamw_meta.h>
155
+ #include <ATen/ops/_fused_dropout_meta.h>
156
+ #include <ATen/ops/_fused_moving_avg_obs_fq_helper_meta.h>
157
+ #include <ATen/ops/_fused_sdp_choice_meta.h>
158
+ #include <ATen/ops/_fused_sgd_meta.h>
159
+ #include <ATen/ops/_fw_primal_meta.h>
160
+ #include <ATen/ops/_fw_primal_copy_meta.h>
161
+ #include <ATen/ops/_gather_sparse_backward_meta.h>
162
+ #include <ATen/ops/_grid_sampler_2d_cpu_fallback_meta.h>
163
+ #include <ATen/ops/_grid_sampler_2d_cpu_fallback_backward_meta.h>
164
+ #include <ATen/ops/_has_compatible_shallow_copy_type_meta.h>
165
+ #include <ATen/ops/_has_same_storage_numel_meta.h>
166
+ #include <ATen/ops/_histogramdd_bin_edges_meta.h>
167
+ #include <ATen/ops/_histogramdd_from_bin_cts_meta.h>
168
+ #include <ATen/ops/_histogramdd_from_bin_tensors_meta.h>
169
+ #include <ATen/ops/_index_put_impl_meta.h>
170
+ #include <ATen/ops/_indices_meta.h>
171
+ #include <ATen/ops/_indices_copy_meta.h>
172
+ #include <ATen/ops/_int_mm_meta.h>
173
+ #include <ATen/ops/_is_all_true_meta.h>
174
+ #include <ATen/ops/_is_any_true_meta.h>
175
+ #include <ATen/ops/_is_zerotensor_meta.h>
176
+ #include <ATen/ops/_jagged_to_padded_dense_forward_meta.h>
177
+ #include <ATen/ops/_lazy_clone_meta.h>
178
+ #include <ATen/ops/_linalg_check_errors_meta.h>
179
+ #include <ATen/ops/_linalg_det_meta.h>
180
+ #include <ATen/ops/_linalg_eigh_meta.h>
181
+ #include <ATen/ops/_linalg_eigvals_meta.h>
182
+ #include <ATen/ops/_linalg_slogdet_meta.h>
183
+ #include <ATen/ops/_linalg_solve_ex_meta.h>
184
+ #include <ATen/ops/_linalg_svd_meta.h>
185
+ #include <ATen/ops/_local_scalar_dense_meta.h>
186
+ #include <ATen/ops/_log_softmax_meta.h>
187
+ #include <ATen/ops/_log_softmax_backward_data_meta.h>
188
+ #include <ATen/ops/_logcumsumexp_meta.h>
189
+ #include <ATen/ops/_lstm_mps_meta.h>
190
+ #include <ATen/ops/_lu_with_info_meta.h>
191
+ #include <ATen/ops/_make_dep_token_meta.h>
192
+ #include <ATen/ops/_make_dual_meta.h>
193
+ #include <ATen/ops/_make_dual_copy_meta.h>
194
+ #include <ATen/ops/_make_per_channel_quantized_tensor_meta.h>
195
+ #include <ATen/ops/_make_per_tensor_quantized_tensor_meta.h>
196
+ #include <ATen/ops/_masked_scale_meta.h>
197
+ #include <ATen/ops/_masked_softmax_meta.h>
198
+ #include <ATen/ops/_masked_softmax_backward_meta.h>
199
+ #include <ATen/ops/_mixed_dtypes_linear_meta.h>
200
+ #include <ATen/ops/_mkldnn_reshape_meta.h>
201
+ #include <ATen/ops/_mkldnn_transpose_meta.h>
202
+ #include <ATen/ops/_mps_convolution_meta.h>
203
+ #include <ATen/ops/_mps_convolution_transpose_meta.h>
204
+ #include <ATen/ops/_native_batch_norm_legit_meta.h>
205
+ #include <ATen/ops/_native_batch_norm_legit_no_training_meta.h>
206
+ #include <ATen/ops/_native_multi_head_attention_meta.h>
207
+ #include <ATen/ops/_neg_view_meta.h>
208
+ #include <ATen/ops/_neg_view_copy_meta.h>
209
+ #include <ATen/ops/_nested_compute_contiguous_strides_offsets_meta.h>
210
+ #include <ATen/ops/_nested_from_padded_meta.h>
211
+ #include <ATen/ops/_nested_from_padded_and_nested_example_meta.h>
212
+ #include <ATen/ops/_nested_get_jagged_dummy_meta.h>
213
+ #include <ATen/ops/_nested_get_lengths_meta.h>
214
+ #include <ATen/ops/_nested_get_max_seqlen_meta.h>
215
+ #include <ATen/ops/_nested_get_min_seqlen_meta.h>
216
+ #include <ATen/ops/_nested_get_offsets_meta.h>
217
+ #include <ATen/ops/_nested_get_ragged_idx_meta.h>
218
+ #include <ATen/ops/_nested_get_values_meta.h>
219
+ #include <ATen/ops/_nested_get_values_copy_meta.h>
220
+ #include <ATen/ops/_nested_select_backward_meta.h>
221
+ #include <ATen/ops/_nested_sum_backward_meta.h>
222
+ #include <ATen/ops/_nested_tensor_from_mask_meta.h>
223
+ #include <ATen/ops/_nested_tensor_from_mask_left_aligned_meta.h>
224
+ #include <ATen/ops/_nested_tensor_from_tensor_list_meta.h>
225
+ #include <ATen/ops/_nested_tensor_size_meta.h>
226
+ #include <ATen/ops/_nested_tensor_softmax_with_shape_meta.h>
227
+ #include <ATen/ops/_nested_tensor_storage_offsets_meta.h>
228
+ #include <ATen/ops/_nested_tensor_strides_meta.h>
229
+ #include <ATen/ops/_nested_view_from_buffer_meta.h>
230
+ #include <ATen/ops/_nested_view_from_buffer_copy_meta.h>
231
+ #include <ATen/ops/_nested_view_from_jagged_meta.h>
232
+ #include <ATen/ops/_nested_view_from_jagged_copy_meta.h>
233
+ #include <ATen/ops/_new_zeros_with_same_feature_meta_meta.h>
234
+ #include <ATen/ops/_nnpack_available_meta.h>
235
+ #include <ATen/ops/_nnpack_spatial_convolution_meta.h>
236
+ #include <ATen/ops/_nnz_meta.h>
237
+ #include <ATen/ops/_pack_padded_sequence_meta.h>
238
+ #include <ATen/ops/_pack_padded_sequence_backward_meta.h>
239
+ #include <ATen/ops/_pad_circular_meta.h>
240
+ #include <ATen/ops/_pad_enum_meta.h>
241
+ #include <ATen/ops/_pad_packed_sequence_meta.h>
242
+ #include <ATen/ops/_padded_dense_to_jagged_forward_meta.h>
243
+ #include <ATen/ops/_pdist_backward_meta.h>
244
+ #include <ATen/ops/_pdist_forward_meta.h>
245
+ #include <ATen/ops/_pin_memory_meta.h>
246
+ #include <ATen/ops/_prelu_kernel_meta.h>
247
+ #include <ATen/ops/_prelu_kernel_backward_meta.h>
248
+ #include <ATen/ops/_print_meta.h>
249
+ #include <ATen/ops/_propagate_xla_data_meta.h>
250
+ #include <ATen/ops/_remove_batch_dim_meta.h>
251
+ #include <ATen/ops/_reshape_alias_meta.h>
252
+ #include <ATen/ops/_reshape_alias_copy_meta.h>
253
+ #include <ATen/ops/_reshape_copy_meta.h>
254
+ #include <ATen/ops/_reshape_from_tensor_meta.h>
255
+ #include <ATen/ops/_resize_output_meta.h>
256
+ #include <ATen/ops/_rowwise_prune_meta.h>
257
+ #include <ATen/ops/_safe_softmax_meta.h>
258
+ #include <ATen/ops/_sample_dirichlet_meta.h>
259
+ #include <ATen/ops/_saturate_weight_to_fp16_meta.h>
260
+ #include <ATen/ops/_scaled_dot_product_attention_math_meta.h>
261
+ #include <ATen/ops/_scaled_dot_product_attention_math_for_mps_meta.h>
262
+ #include <ATen/ops/_scaled_dot_product_cudnn_attention_meta.h>
263
+ #include <ATen/ops/_scaled_dot_product_cudnn_attention_backward_meta.h>
264
+ #include <ATen/ops/_scaled_dot_product_efficient_attention_meta.h>
265
+ #include <ATen/ops/_scaled_dot_product_efficient_attention_backward_meta.h>
266
+ #include <ATen/ops/_scaled_dot_product_flash_attention_meta.h>
267
+ #include <ATen/ops/_scaled_dot_product_flash_attention_backward_meta.h>
268
+ #include <ATen/ops/_scaled_dot_product_flash_attention_for_cpu_meta.h>
269
+ #include <ATen/ops/_scaled_dot_product_flash_attention_for_cpu_backward_meta.h>
270
+ #include <ATen/ops/_scaled_dot_product_fused_attention_overrideable_meta.h>
271
+ #include <ATen/ops/_scaled_dot_product_fused_attention_overrideable_backward_meta.h>
272
+ #include <ATen/ops/_scaled_mm_meta.h>
273
+ #include <ATen/ops/_segment_reduce_backward_meta.h>
274
+ #include <ATen/ops/_shape_as_tensor_meta.h>
275
+ #include <ATen/ops/_slow_conv2d_backward_meta.h>
276
+ #include <ATen/ops/_slow_conv2d_forward_meta.h>
277
+ #include <ATen/ops/_sobol_engine_draw_meta.h>
278
+ #include <ATen/ops/_sobol_engine_ff_meta.h>
279
+ #include <ATen/ops/_sobol_engine_initialize_state_meta.h>
280
+ #include <ATen/ops/_sobol_engine_scramble_meta.h>
281
+ #include <ATen/ops/_softmax_meta.h>
282
+ #include <ATen/ops/_softmax_backward_data_meta.h>
283
+ #include <ATen/ops/_sparse_addmm_meta.h>
284
+ #include <ATen/ops/_sparse_broadcast_to_meta.h>
285
+ #include <ATen/ops/_sparse_broadcast_to_copy_meta.h>
286
+ #include <ATen/ops/_sparse_bsc_tensor_unsafe_meta.h>
287
+ #include <ATen/ops/_sparse_bsr_tensor_unsafe_meta.h>
288
+ #include <ATen/ops/_sparse_compressed_tensor_unsafe_meta.h>
289
+ #include <ATen/ops/_sparse_compressed_tensor_with_dims_meta.h>
290
+ #include <ATen/ops/_sparse_coo_tensor_unsafe_meta.h>
291
+ #include <ATen/ops/_sparse_coo_tensor_with_dims_meta.h>
292
+ #include <ATen/ops/_sparse_coo_tensor_with_dims_and_tensors_meta.h>
293
+ #include <ATen/ops/_sparse_csc_tensor_unsafe_meta.h>
294
+ #include <ATen/ops/_sparse_csr_prod_meta.h>
295
+ #include <ATen/ops/_sparse_csr_sum_meta.h>
296
+ #include <ATen/ops/_sparse_csr_tensor_unsafe_meta.h>
297
+ #include <ATen/ops/_sparse_log_softmax_meta.h>
298
+ #include <ATen/ops/_sparse_log_softmax_backward_data_meta.h>
299
+ #include <ATen/ops/_sparse_mask_projection_meta.h>
300
+ #include <ATen/ops/_sparse_mm_meta.h>
301
+ #include <ATen/ops/_sparse_mm_reduce_impl_meta.h>
302
+ #include <ATen/ops/_sparse_mm_reduce_impl_backward_meta.h>
303
+ #include <ATen/ops/_sparse_semi_structured_addmm_meta.h>
304
+ #include <ATen/ops/_sparse_semi_structured_apply_meta.h>
305
+ #include <ATen/ops/_sparse_semi_structured_apply_dense_meta.h>
306
+ #include <ATen/ops/_sparse_semi_structured_linear_meta.h>
307
+ #include <ATen/ops/_sparse_semi_structured_mm_meta.h>
308
+ #include <ATen/ops/_sparse_semi_structured_tile_meta.h>
309
+ #include <ATen/ops/_sparse_softmax_meta.h>
310
+ #include <ATen/ops/_sparse_softmax_backward_data_meta.h>
311
+ #include <ATen/ops/_sparse_sparse_matmul_meta.h>
312
+ #include <ATen/ops/_sparse_sum_meta.h>
313
+ #include <ATen/ops/_sparse_sum_backward_meta.h>
314
+ #include <ATen/ops/_spdiags_meta.h>
315
+ #include <ATen/ops/_spsolve_meta.h>
316
+ #include <ATen/ops/_stack_meta.h>
317
+ #include <ATen/ops/_standard_gamma_meta.h>
318
+ #include <ATen/ops/_standard_gamma_grad_meta.h>
319
+ #include <ATen/ops/_test_ambiguous_defaults_meta.h>
320
+ #include <ATen/ops/_test_autograd_multiple_dispatch_meta.h>
321
+ #include <ATen/ops/_test_autograd_multiple_dispatch_view_meta.h>
322
+ #include <ATen/ops/_test_autograd_multiple_dispatch_view_copy_meta.h>
323
+ #include <ATen/ops/_test_check_tensor_meta.h>
324
+ #include <ATen/ops/_test_functorch_fallback_meta.h>
325
+ #include <ATen/ops/_test_optional_filled_intlist_meta.h>
326
+ #include <ATen/ops/_test_optional_floatlist_meta.h>
327
+ #include <ATen/ops/_test_optional_intlist_meta.h>
328
+ #include <ATen/ops/_test_parallel_materialize_meta.h>
329
+ #include <ATen/ops/_test_serialization_subcmul_meta.h>
330
+ #include <ATen/ops/_test_string_default_meta.h>
331
+ #include <ATen/ops/_test_warn_in_autograd_meta.h>
332
+ #include <ATen/ops/_thnn_differentiable_gru_cell_backward_meta.h>
333
+ #include <ATen/ops/_thnn_differentiable_lstm_cell_backward_meta.h>
334
+ #include <ATen/ops/_thnn_fused_gru_cell_meta.h>
335
+ #include <ATen/ops/_thnn_fused_gru_cell_backward_meta.h>
336
+ #include <ATen/ops/_thnn_fused_lstm_cell_meta.h>
337
+ #include <ATen/ops/_thnn_fused_lstm_cell_backward_meta.h>
338
+ #include <ATen/ops/_thnn_fused_lstm_cell_backward_impl_meta.h>
339
+ #include <ATen/ops/_to_copy_meta.h>
340
+ #include <ATen/ops/_to_cpu_meta.h>
341
+ #include <ATen/ops/_to_dense_meta.h>
342
+ #include <ATen/ops/_to_sparse_meta.h>
343
+ #include <ATen/ops/_to_sparse_bsc_meta.h>
344
+ #include <ATen/ops/_to_sparse_bsr_meta.h>
345
+ #include <ATen/ops/_to_sparse_csc_meta.h>
346
+ #include <ATen/ops/_to_sparse_csr_meta.h>
347
+ #include <ATen/ops/_to_sparse_semi_structured_meta.h>
348
+ #include <ATen/ops/_transform_bias_rescale_qkv_meta.h>
349
+ #include <ATen/ops/_transformer_encoder_layer_fwd_meta.h>
350
+ #include <ATen/ops/_trilinear_meta.h>
351
+ #include <ATen/ops/_triton_multi_head_attention_meta.h>
352
+ #include <ATen/ops/_triton_scaled_dot_attention_meta.h>
353
+ #include <ATen/ops/_unique_meta.h>
354
+ #include <ATen/ops/_unique2_meta.h>
355
+ #include <ATen/ops/_unpack_dual_meta.h>
356
+ #include <ATen/ops/_unsafe_index_meta.h>
357
+ #include <ATen/ops/_unsafe_index_put_meta.h>
358
+ #include <ATen/ops/_unsafe_masked_index_meta.h>
359
+ #include <ATen/ops/_unsafe_masked_index_put_accumulate_meta.h>
360
+ #include <ATen/ops/_unsafe_view_meta.h>
361
+ #include <ATen/ops/_upsample_bicubic2d_aa_meta.h>
362
+ #include <ATen/ops/_upsample_bicubic2d_aa_backward_meta.h>
363
+ #include <ATen/ops/_upsample_bilinear2d_aa_meta.h>
364
+ #include <ATen/ops/_upsample_bilinear2d_aa_backward_meta.h>
365
+ #include <ATen/ops/_upsample_nearest_exact1d_meta.h>
366
+ #include <ATen/ops/_upsample_nearest_exact1d_backward_meta.h>
367
+ #include <ATen/ops/_upsample_nearest_exact2d_meta.h>
368
+ #include <ATen/ops/_upsample_nearest_exact2d_backward_meta.h>
369
+ #include <ATen/ops/_upsample_nearest_exact3d_meta.h>
370
+ #include <ATen/ops/_upsample_nearest_exact3d_backward_meta.h>
371
+ #include <ATen/ops/_use_cudnn_ctc_loss_meta.h>
372
+ #include <ATen/ops/_use_cudnn_rnn_flatten_weight_meta.h>
373
+ #include <ATen/ops/_validate_compressed_sparse_indices_meta.h>
374
+ #include <ATen/ops/_validate_sparse_bsc_tensor_args_meta.h>
375
+ #include <ATen/ops/_validate_sparse_bsr_tensor_args_meta.h>
376
+ #include <ATen/ops/_validate_sparse_compressed_tensor_args_meta.h>
377
+ #include <ATen/ops/_validate_sparse_coo_tensor_args_meta.h>
378
+ #include <ATen/ops/_validate_sparse_csc_tensor_args_meta.h>
379
+ #include <ATen/ops/_validate_sparse_csr_tensor_args_meta.h>
380
+ #include <ATen/ops/_values_meta.h>
381
+ #include <ATen/ops/_values_copy_meta.h>
382
+ #include <ATen/ops/_version_meta.h>
383
+ #include <ATen/ops/_weight_int4pack_mm_meta.h>
384
+ #include <ATen/ops/_weight_int8pack_mm_meta.h>
385
+ #include <ATen/ops/_weight_norm_meta.h>
386
+ #include <ATen/ops/_weight_norm_differentiable_backward_meta.h>
387
+ #include <ATen/ops/_weight_norm_interface_meta.h>
388
+ #include <ATen/ops/_weight_norm_interface_backward_meta.h>
389
+ #include <ATen/ops/_wrapped_linear_prepack_meta.h>
390
+ #include <ATen/ops/_wrapped_quantized_linear_prepacked_meta.h>
391
+ #include <ATen/ops/abs_meta.h>
392
+ #include <ATen/ops/absolute_meta.h>
393
+ #include <ATen/ops/acos_meta.h>
394
+ #include <ATen/ops/acosh_meta.h>
395
+ #include <ATen/ops/adaptive_avg_pool1d_meta.h>
396
+ #include <ATen/ops/adaptive_avg_pool2d_meta.h>
397
+ #include <ATen/ops/adaptive_avg_pool3d_meta.h>
398
+ #include <ATen/ops/adaptive_avg_pool3d_backward_meta.h>
399
+ #include <ATen/ops/adaptive_max_pool1d_meta.h>
400
+ #include <ATen/ops/adaptive_max_pool2d_meta.h>
401
+ #include <ATen/ops/adaptive_max_pool2d_backward_meta.h>
402
+ #include <ATen/ops/adaptive_max_pool3d_meta.h>
403
+ #include <ATen/ops/adaptive_max_pool3d_backward_meta.h>
404
+ #include <ATen/ops/add_meta.h>
405
+ #include <ATen/ops/addbmm_meta.h>
406
+ #include <ATen/ops/addcdiv_meta.h>
407
+ #include <ATen/ops/addcmul_meta.h>
408
+ #include <ATen/ops/addmm_meta.h>
409
+ #include <ATen/ops/addmv_meta.h>
410
+ #include <ATen/ops/addr_meta.h>
411
+ #include <ATen/ops/adjoint_meta.h>
412
+ #include <ATen/ops/affine_grid_generator_meta.h>
413
+ #include <ATen/ops/affine_grid_generator_backward_meta.h>
414
+ #include <ATen/ops/alias_meta.h>
415
+ #include <ATen/ops/alias_copy_meta.h>
416
+ #include <ATen/ops/align_as_meta.h>
417
+ #include <ATen/ops/align_tensors_meta.h>
418
+ #include <ATen/ops/align_to_meta.h>
419
+ #include <ATen/ops/all_meta.h>
420
+ #include <ATen/ops/allclose_meta.h>
421
+ #include <ATen/ops/alpha_dropout_meta.h>
422
+ #include <ATen/ops/amax_meta.h>
423
+ #include <ATen/ops/amin_meta.h>
424
+ #include <ATen/ops/aminmax_meta.h>
425
+ #include <ATen/ops/and_meta.h>
426
+ #include <ATen/ops/angle_meta.h>
427
+ #include <ATen/ops/any_meta.h>
428
+ #include <ATen/ops/arange_meta.h>
429
+ #include <ATen/ops/arccos_meta.h>
430
+ #include <ATen/ops/arccosh_meta.h>
431
+ #include <ATen/ops/arcsin_meta.h>
432
+ #include <ATen/ops/arcsinh_meta.h>
433
+ #include <ATen/ops/arctan_meta.h>
434
+ #include <ATen/ops/arctan2_meta.h>
435
+ #include <ATen/ops/arctanh_meta.h>
436
+ #include <ATen/ops/argmax_meta.h>
437
+ #include <ATen/ops/argmin_meta.h>
438
+ #include <ATen/ops/argsort_meta.h>
439
+ #include <ATen/ops/argwhere_meta.h>
440
+ #include <ATen/ops/as_strided_meta.h>
441
+ #include <ATen/ops/as_strided_copy_meta.h>
442
+ #include <ATen/ops/as_strided_scatter_meta.h>
443
+ #include <ATen/ops/asin_meta.h>
444
+ #include <ATen/ops/asinh_meta.h>
445
+ #include <ATen/ops/atan_meta.h>
446
+ #include <ATen/ops/atan2_meta.h>
447
+ #include <ATen/ops/atanh_meta.h>
448
+ #include <ATen/ops/atleast_1d_meta.h>
449
+ #include <ATen/ops/atleast_2d_meta.h>
450
+ #include <ATen/ops/atleast_3d_meta.h>
451
+ #include <ATen/ops/avg_pool1d_meta.h>
452
+ #include <ATen/ops/avg_pool2d_meta.h>
453
+ #include <ATen/ops/avg_pool2d_backward_meta.h>
454
+ #include <ATen/ops/avg_pool3d_meta.h>
455
+ #include <ATen/ops/avg_pool3d_backward_meta.h>
456
+ #include <ATen/ops/baddbmm_meta.h>
457
+ #include <ATen/ops/bartlett_window_meta.h>
458
+ #include <ATen/ops/batch_norm_meta.h>
459
+ #include <ATen/ops/batch_norm_backward_meta.h>
460
+ #include <ATen/ops/batch_norm_backward_elemt_meta.h>
461
+ #include <ATen/ops/batch_norm_backward_reduce_meta.h>
462
+ #include <ATen/ops/batch_norm_elemt_meta.h>
463
+ #include <ATen/ops/batch_norm_gather_stats_meta.h>
464
+ #include <ATen/ops/batch_norm_gather_stats_with_counts_meta.h>
465
+ #include <ATen/ops/batch_norm_stats_meta.h>
466
+ #include <ATen/ops/batch_norm_update_stats_meta.h>
467
+ #include <ATen/ops/bernoulli_meta.h>
468
+ #include <ATen/ops/bilinear_meta.h>
469
+ #include <ATen/ops/binary_cross_entropy_meta.h>
470
+ #include <ATen/ops/binary_cross_entropy_backward_meta.h>
471
+ #include <ATen/ops/binary_cross_entropy_with_logits_meta.h>
472
+ #include <ATen/ops/bincount_meta.h>
473
+ #include <ATen/ops/binomial_meta.h>
474
+ #include <ATen/ops/bitwise_and_meta.h>
475
+ #include <ATen/ops/bitwise_left_shift_meta.h>
476
+ #include <ATen/ops/bitwise_not_meta.h>
477
+ #include <ATen/ops/bitwise_or_meta.h>
478
+ #include <ATen/ops/bitwise_right_shift_meta.h>
479
+ #include <ATen/ops/bitwise_xor_meta.h>
480
+ #include <ATen/ops/blackman_window_meta.h>
481
+ #include <ATen/ops/block_diag_meta.h>
482
+ #include <ATen/ops/bmm_meta.h>
483
+ #include <ATen/ops/broadcast_tensors_meta.h>
484
+ #include <ATen/ops/broadcast_to_meta.h>
485
+ #include <ATen/ops/bucketize_meta.h>
486
+ #include <ATen/ops/can_cast_meta.h>
487
+ #include <ATen/ops/cartesian_prod_meta.h>
488
+ #include <ATen/ops/cat_meta.h>
489
+ #include <ATen/ops/cauchy_meta.h>
490
+ #include <ATen/ops/ccol_indices_meta.h>
491
+ #include <ATen/ops/ccol_indices_copy_meta.h>
492
+ #include <ATen/ops/cdist_meta.h>
493
+ #include <ATen/ops/ceil_meta.h>
494
+ #include <ATen/ops/celu_meta.h>
495
+ #include <ATen/ops/chain_matmul_meta.h>
496
+ #include <ATen/ops/chalf_meta.h>
497
+ #include <ATen/ops/channel_shuffle_meta.h>
498
+ #include <ATen/ops/cholesky_meta.h>
499
+ #include <ATen/ops/cholesky_inverse_meta.h>
500
+ #include <ATen/ops/cholesky_solve_meta.h>
501
+ #include <ATen/ops/choose_qparams_optimized_meta.h>
502
+ #include <ATen/ops/chunk_meta.h>
503
+ #include <ATen/ops/clamp_meta.h>
504
+ #include <ATen/ops/clamp_max_meta.h>
505
+ #include <ATen/ops/clamp_min_meta.h>
506
+ #include <ATen/ops/clip_meta.h>
507
+ #include <ATen/ops/clone_meta.h>
508
+ #include <ATen/ops/coalesce_meta.h>
509
+ #include <ATen/ops/col2im_meta.h>
510
+ #include <ATen/ops/col_indices_meta.h>
511
+ #include <ATen/ops/col_indices_copy_meta.h>
512
+ #include <ATen/ops/column_stack_meta.h>
513
+ #include <ATen/ops/combinations_meta.h>
514
+ #include <ATen/ops/complex_meta.h>
515
+ #include <ATen/ops/concat_meta.h>
516
+ #include <ATen/ops/concatenate_meta.h>
517
+ #include <ATen/ops/conj_meta.h>
518
+ #include <ATen/ops/conj_physical_meta.h>
519
+ #include <ATen/ops/constant_pad_nd_meta.h>
520
+ #include <ATen/ops/contiguous_meta.h>
521
+ #include <ATen/ops/conv1d_meta.h>
522
+ #include <ATen/ops/conv2d_meta.h>
523
+ #include <ATen/ops/conv3d_meta.h>
524
+ #include <ATen/ops/conv_depthwise3d_meta.h>
525
+ #include <ATen/ops/conv_tbc_meta.h>
526
+ #include <ATen/ops/conv_tbc_backward_meta.h>
527
+ #include <ATen/ops/conv_transpose1d_meta.h>
528
+ #include <ATen/ops/conv_transpose2d_meta.h>
529
+ #include <ATen/ops/conv_transpose3d_meta.h>
530
+ #include <ATen/ops/convolution_meta.h>
531
+ #include <ATen/ops/convolution_backward_meta.h>
532
+ #include <ATen/ops/convolution_backward_overrideable_meta.h>
533
+ #include <ATen/ops/convolution_overrideable_meta.h>
534
+ #include <ATen/ops/copy_meta.h>
535
+ #include <ATen/ops/copy_sparse_to_sparse_meta.h>
536
+ #include <ATen/ops/copysign_meta.h>
537
+ #include <ATen/ops/corrcoef_meta.h>
538
+ #include <ATen/ops/cos_meta.h>
539
+ #include <ATen/ops/cosh_meta.h>
540
+ #include <ATen/ops/cosine_embedding_loss_meta.h>
541
+ #include <ATen/ops/cosine_similarity_meta.h>
542
+ #include <ATen/ops/count_nonzero_meta.h>
543
+ #include <ATen/ops/cov_meta.h>
544
+ #include <ATen/ops/cross_meta.h>
545
+ #include <ATen/ops/cross_entropy_loss_meta.h>
546
+ #include <ATen/ops/crow_indices_meta.h>
547
+ #include <ATen/ops/crow_indices_copy_meta.h>
548
+ #include <ATen/ops/ctc_loss_meta.h>
549
+ #include <ATen/ops/cudnn_affine_grid_generator_meta.h>
550
+ #include <ATen/ops/cudnn_affine_grid_generator_backward_meta.h>
551
+ #include <ATen/ops/cudnn_batch_norm_meta.h>
552
+ #include <ATen/ops/cudnn_batch_norm_backward_meta.h>
553
+ #include <ATen/ops/cudnn_convolution_meta.h>
554
+ #include <ATen/ops/cudnn_convolution_add_relu_meta.h>
555
+ #include <ATen/ops/cudnn_convolution_relu_meta.h>
556
+ #include <ATen/ops/cudnn_convolution_transpose_meta.h>
557
+ #include <ATen/ops/cudnn_grid_sampler_meta.h>
558
+ #include <ATen/ops/cudnn_grid_sampler_backward_meta.h>
559
+ #include <ATen/ops/cudnn_is_acceptable_meta.h>
560
+ #include <ATen/ops/cummax_meta.h>
561
+ #include <ATen/ops/cummaxmin_backward_meta.h>
562
+ #include <ATen/ops/cummin_meta.h>
563
+ #include <ATen/ops/cumprod_meta.h>
564
+ #include <ATen/ops/cumprod_backward_meta.h>
565
+ #include <ATen/ops/cumsum_meta.h>
566
+ #include <ATen/ops/cumulative_trapezoid_meta.h>
567
+ #include <ATen/ops/data_meta.h>
568
+ #include <ATen/ops/deg2rad_meta.h>
569
+ #include <ATen/ops/dense_dim_meta.h>
570
+ #include <ATen/ops/dequantize_meta.h>
571
+ #include <ATen/ops/det_meta.h>
572
+ #include <ATen/ops/detach_meta.h>
573
+ #include <ATen/ops/detach_copy_meta.h>
574
+ #include <ATen/ops/diag_meta.h>
575
+ #include <ATen/ops/diag_embed_meta.h>
576
+ #include <ATen/ops/diagflat_meta.h>
577
+ #include <ATen/ops/diagonal_meta.h>
578
+ #include <ATen/ops/diagonal_backward_meta.h>
579
+ #include <ATen/ops/diagonal_copy_meta.h>
580
+ #include <ATen/ops/diagonal_scatter_meta.h>
581
+ #include <ATen/ops/diff_meta.h>
582
+ #include <ATen/ops/digamma_meta.h>
583
+ #include <ATen/ops/dist_meta.h>
584
+ #include <ATen/ops/div_meta.h>
585
+ #include <ATen/ops/divide_meta.h>
586
+ #include <ATen/ops/dot_meta.h>
587
+ #include <ATen/ops/dropout_meta.h>
588
+ #include <ATen/ops/dsplit_meta.h>
589
+ #include <ATen/ops/dstack_meta.h>
590
+ #include <ATen/ops/einsum_meta.h>
591
+ #include <ATen/ops/elu_meta.h>
592
+ #include <ATen/ops/elu_backward_meta.h>
593
+ #include <ATen/ops/embedding_meta.h>
594
+ #include <ATen/ops/embedding_backward_meta.h>
595
+ #include <ATen/ops/embedding_bag_meta.h>
596
+ #include <ATen/ops/embedding_dense_backward_meta.h>
597
+ #include <ATen/ops/embedding_renorm_meta.h>
598
+ #include <ATen/ops/embedding_sparse_backward_meta.h>
599
+ #include <ATen/ops/empty_meta.h>
600
+ #include <ATen/ops/empty_like_meta.h>
601
+ #include <ATen/ops/empty_permuted_meta.h>
602
+ #include <ATen/ops/empty_quantized_meta.h>
603
+ #include <ATen/ops/empty_strided_meta.h>
604
+ #include <ATen/ops/eq_meta.h>
605
+ #include <ATen/ops/equal_meta.h>
606
+ #include <ATen/ops/erf_meta.h>
607
+ #include <ATen/ops/erfc_meta.h>
608
+ #include <ATen/ops/erfinv_meta.h>
609
+ #include <ATen/ops/exp_meta.h>
610
+ #include <ATen/ops/exp2_meta.h>
611
+ #include <ATen/ops/expand_meta.h>
612
+ #include <ATen/ops/expand_as_meta.h>
613
+ #include <ATen/ops/expand_copy_meta.h>
614
+ #include <ATen/ops/expm1_meta.h>
615
+ #include <ATen/ops/exponential_meta.h>
616
+ #include <ATen/ops/eye_meta.h>
617
+ #include <ATen/ops/fake_quantize_per_channel_affine_meta.h>
618
+ #include <ATen/ops/fake_quantize_per_channel_affine_cachemask_meta.h>
619
+ #include <ATen/ops/fake_quantize_per_channel_affine_cachemask_backward_meta.h>
620
+ #include <ATen/ops/fake_quantize_per_tensor_affine_meta.h>
621
+ #include <ATen/ops/fake_quantize_per_tensor_affine_cachemask_meta.h>
622
+ #include <ATen/ops/fake_quantize_per_tensor_affine_cachemask_backward_meta.h>
623
+ #include <ATen/ops/fbgemm_linear_fp16_weight_meta.h>
624
+ #include <ATen/ops/fbgemm_linear_fp16_weight_fp32_activation_meta.h>
625
+ #include <ATen/ops/fbgemm_linear_int8_weight_meta.h>
626
+ #include <ATen/ops/fbgemm_linear_int8_weight_fp32_activation_meta.h>
627
+ #include <ATen/ops/fbgemm_linear_quantize_weight_meta.h>
628
+ #include <ATen/ops/fbgemm_pack_gemm_matrix_fp16_meta.h>
629
+ #include <ATen/ops/fbgemm_pack_quantized_matrix_meta.h>
630
+ #include <ATen/ops/feature_alpha_dropout_meta.h>
631
+ #include <ATen/ops/feature_dropout_meta.h>
632
+ #include <ATen/ops/fft_fft_meta.h>
633
+ #include <ATen/ops/fft_fft2_meta.h>
634
+ #include <ATen/ops/fft_fftfreq_meta.h>
635
+ #include <ATen/ops/fft_fftn_meta.h>
636
+ #include <ATen/ops/fft_fftshift_meta.h>
637
+ #include <ATen/ops/fft_hfft_meta.h>
638
+ #include <ATen/ops/fft_hfft2_meta.h>
639
+ #include <ATen/ops/fft_hfftn_meta.h>
640
+ #include <ATen/ops/fft_ifft_meta.h>
641
+ #include <ATen/ops/fft_ifft2_meta.h>
642
+ #include <ATen/ops/fft_ifftn_meta.h>
643
+ #include <ATen/ops/fft_ifftshift_meta.h>
644
+ #include <ATen/ops/fft_ihfft_meta.h>
645
+ #include <ATen/ops/fft_ihfft2_meta.h>
646
+ #include <ATen/ops/fft_ihfftn_meta.h>
647
+ #include <ATen/ops/fft_irfft_meta.h>
648
+ #include <ATen/ops/fft_irfft2_meta.h>
649
+ #include <ATen/ops/fft_irfftn_meta.h>
650
+ #include <ATen/ops/fft_rfft_meta.h>
651
+ #include <ATen/ops/fft_rfft2_meta.h>
652
+ #include <ATen/ops/fft_rfftfreq_meta.h>
653
+ #include <ATen/ops/fft_rfftn_meta.h>
654
+ #include <ATen/ops/fill_meta.h>
655
+ #include <ATen/ops/fill_diagonal_meta.h>
656
+ #include <ATen/ops/fix_meta.h>
657
+ #include <ATen/ops/flatten_meta.h>
658
+ #include <ATen/ops/flatten_dense_tensors_meta.h>
659
+ #include <ATen/ops/flip_meta.h>
660
+ #include <ATen/ops/fliplr_meta.h>
661
+ #include <ATen/ops/flipud_meta.h>
662
+ #include <ATen/ops/float_power_meta.h>
663
+ #include <ATen/ops/floor_meta.h>
664
+ #include <ATen/ops/floor_divide_meta.h>
665
+ #include <ATen/ops/fmax_meta.h>
666
+ #include <ATen/ops/fmin_meta.h>
667
+ #include <ATen/ops/fmod_meta.h>
668
+ #include <ATen/ops/frac_meta.h>
669
+ #include <ATen/ops/fractional_max_pool2d_meta.h>
670
+ #include <ATen/ops/fractional_max_pool2d_backward_meta.h>
671
+ #include <ATen/ops/fractional_max_pool3d_meta.h>
672
+ #include <ATen/ops/fractional_max_pool3d_backward_meta.h>
673
+ #include <ATen/ops/frexp_meta.h>
674
+ #include <ATen/ops/frobenius_norm_meta.h>
675
+ #include <ATen/ops/from_file_meta.h>
676
+ #include <ATen/ops/full_meta.h>
677
+ #include <ATen/ops/full_like_meta.h>
678
+ #include <ATen/ops/fused_moving_avg_obs_fake_quant_meta.h>
679
+ #include <ATen/ops/gather_meta.h>
680
+ #include <ATen/ops/gather_backward_meta.h>
681
+ #include <ATen/ops/gcd_meta.h>
682
+ #include <ATen/ops/ge_meta.h>
683
+ #include <ATen/ops/gelu_meta.h>
684
+ #include <ATen/ops/gelu_backward_meta.h>
685
+ #include <ATen/ops/geometric_meta.h>
686
+ #include <ATen/ops/geqrf_meta.h>
687
+ #include <ATen/ops/ger_meta.h>
688
+ #include <ATen/ops/glu_meta.h>
689
+ #include <ATen/ops/glu_backward_meta.h>
690
+ #include <ATen/ops/glu_backward_jvp_meta.h>
691
+ #include <ATen/ops/glu_jvp_meta.h>
692
+ #include <ATen/ops/gradient_meta.h>
693
+ #include <ATen/ops/greater_meta.h>
694
+ #include <ATen/ops/greater_equal_meta.h>
695
+ #include <ATen/ops/grid_sampler_meta.h>
696
+ #include <ATen/ops/grid_sampler_2d_meta.h>
697
+ #include <ATen/ops/grid_sampler_2d_backward_meta.h>
698
+ #include <ATen/ops/grid_sampler_3d_meta.h>
699
+ #include <ATen/ops/grid_sampler_3d_backward_meta.h>
700
+ #include <ATen/ops/group_norm_meta.h>
701
+ #include <ATen/ops/gru_meta.h>
702
+ #include <ATen/ops/gru_cell_meta.h>
703
+ #include <ATen/ops/gt_meta.h>
704
+ #include <ATen/ops/hamming_window_meta.h>
705
+ #include <ATen/ops/hann_window_meta.h>
706
+ #include <ATen/ops/hardshrink_meta.h>
707
+ #include <ATen/ops/hardshrink_backward_meta.h>
708
+ #include <ATen/ops/hardsigmoid_meta.h>
709
+ #include <ATen/ops/hardsigmoid_backward_meta.h>
710
+ #include <ATen/ops/hardswish_meta.h>
711
+ #include <ATen/ops/hardswish_backward_meta.h>
712
+ #include <ATen/ops/hardtanh_meta.h>
713
+ #include <ATen/ops/hardtanh_backward_meta.h>
714
+ #include <ATen/ops/heaviside_meta.h>
715
+ #include <ATen/ops/hinge_embedding_loss_meta.h>
716
+ #include <ATen/ops/histc_meta.h>
717
+ #include <ATen/ops/histogram_meta.h>
718
+ #include <ATen/ops/histogramdd_meta.h>
719
+ #include <ATen/ops/hsplit_meta.h>
720
+ #include <ATen/ops/hspmm_meta.h>
721
+ #include <ATen/ops/hstack_meta.h>
722
+ #include <ATen/ops/huber_loss_meta.h>
723
+ #include <ATen/ops/huber_loss_backward_meta.h>
724
+ #include <ATen/ops/hypot_meta.h>
725
+ #include <ATen/ops/i0_meta.h>
726
+ #include <ATen/ops/igamma_meta.h>
727
+ #include <ATen/ops/igammac_meta.h>
728
+ #include <ATen/ops/im2col_meta.h>
729
+ #include <ATen/ops/imag_meta.h>
730
+ #include <ATen/ops/index_meta.h>
731
+ #include <ATen/ops/index_add_meta.h>
732
+ #include <ATen/ops/index_copy_meta.h>
733
+ #include <ATen/ops/index_fill_meta.h>
734
+ #include <ATen/ops/index_put_meta.h>
735
+ #include <ATen/ops/index_reduce_meta.h>
736
+ #include <ATen/ops/index_select_meta.h>
737
+ #include <ATen/ops/index_select_backward_meta.h>
738
+ #include <ATen/ops/indices_meta.h>
739
+ #include <ATen/ops/indices_copy_meta.h>
740
+ #include <ATen/ops/infinitely_differentiable_gelu_backward_meta.h>
741
+ #include <ATen/ops/inner_meta.h>
742
+ #include <ATen/ops/instance_norm_meta.h>
743
+ #include <ATen/ops/int_repr_meta.h>
744
+ #include <ATen/ops/inverse_meta.h>
745
+ #include <ATen/ops/is_coalesced_meta.h>
746
+ #include <ATen/ops/is_complex_meta.h>
747
+ #include <ATen/ops/is_conj_meta.h>
748
+ #include <ATen/ops/is_distributed_meta.h>
749
+ #include <ATen/ops/is_floating_point_meta.h>
750
+ #include <ATen/ops/is_inference_meta.h>
751
+ #include <ATen/ops/is_leaf_meta.h>
752
+ #include <ATen/ops/is_neg_meta.h>
753
+ #include <ATen/ops/is_nonzero_meta.h>
754
+ #include <ATen/ops/is_pinned_meta.h>
755
+ #include <ATen/ops/is_same_size_meta.h>
756
+ #include <ATen/ops/is_set_to_meta.h>
757
+ #include <ATen/ops/is_signed_meta.h>
758
+ #include <ATen/ops/is_vulkan_available_meta.h>
759
+ #include <ATen/ops/isclose_meta.h>
760
+ #include <ATen/ops/isfinite_meta.h>
761
+ #include <ATen/ops/isin_meta.h>
762
+ #include <ATen/ops/isinf_meta.h>
763
+ #include <ATen/ops/isnan_meta.h>
764
+ #include <ATen/ops/isneginf_meta.h>
765
+ #include <ATen/ops/isposinf_meta.h>
766
+ #include <ATen/ops/isreal_meta.h>
767
+ #include <ATen/ops/istft_meta.h>
768
+ #include <ATen/ops/item_meta.h>
769
+ #include <ATen/ops/kaiser_window_meta.h>
770
+ #include <ATen/ops/kl_div_meta.h>
771
+ #include <ATen/ops/kron_meta.h>
772
+ #include <ATen/ops/kthvalue_meta.h>
773
+ #include <ATen/ops/l1_loss_meta.h>
774
+ #include <ATen/ops/layer_norm_meta.h>
775
+ #include <ATen/ops/lcm_meta.h>
776
+ #include <ATen/ops/ldexp_meta.h>
777
+ #include <ATen/ops/le_meta.h>
778
+ #include <ATen/ops/leaky_relu_meta.h>
779
+ #include <ATen/ops/leaky_relu_backward_meta.h>
780
+ #include <ATen/ops/lerp_meta.h>
781
+ #include <ATen/ops/less_meta.h>
782
+ #include <ATen/ops/less_equal_meta.h>
783
+ #include <ATen/ops/lgamma_meta.h>
784
+ #include <ATen/ops/lift_meta.h>
785
+ #include <ATen/ops/lift_fresh_meta.h>
786
+ #include <ATen/ops/lift_fresh_copy_meta.h>
787
+ #include <ATen/ops/linalg_cholesky_meta.h>
788
+ #include <ATen/ops/linalg_cholesky_ex_meta.h>
789
+ #include <ATen/ops/linalg_cond_meta.h>
790
+ #include <ATen/ops/linalg_cross_meta.h>
791
+ #include <ATen/ops/linalg_det_meta.h>
792
+ #include <ATen/ops/linalg_diagonal_meta.h>
793
+ #include <ATen/ops/linalg_eig_meta.h>
794
+ #include <ATen/ops/linalg_eigh_meta.h>
795
+ #include <ATen/ops/linalg_eigvals_meta.h>
796
+ #include <ATen/ops/linalg_eigvalsh_meta.h>
797
+ #include <ATen/ops/linalg_householder_product_meta.h>
798
+ #include <ATen/ops/linalg_inv_meta.h>
799
+ #include <ATen/ops/linalg_inv_ex_meta.h>
800
+ #include <ATen/ops/linalg_ldl_factor_meta.h>
801
+ #include <ATen/ops/linalg_ldl_factor_ex_meta.h>
802
+ #include <ATen/ops/linalg_ldl_solve_meta.h>
803
+ #include <ATen/ops/linalg_lstsq_meta.h>
804
+ #include <ATen/ops/linalg_lu_meta.h>
805
+ #include <ATen/ops/linalg_lu_factor_meta.h>
806
+ #include <ATen/ops/linalg_lu_factor_ex_meta.h>
807
+ #include <ATen/ops/linalg_lu_solve_meta.h>
808
+ #include <ATen/ops/linalg_matmul_meta.h>
809
+ #include <ATen/ops/linalg_matrix_exp_meta.h>
810
+ #include <ATen/ops/linalg_matrix_norm_meta.h>
811
+ #include <ATen/ops/linalg_matrix_power_meta.h>
812
+ #include <ATen/ops/linalg_matrix_rank_meta.h>
813
+ #include <ATen/ops/linalg_multi_dot_meta.h>
814
+ #include <ATen/ops/linalg_norm_meta.h>
815
+ #include <ATen/ops/linalg_pinv_meta.h>
816
+ #include <ATen/ops/linalg_qr_meta.h>
817
+ #include <ATen/ops/linalg_slogdet_meta.h>
818
+ #include <ATen/ops/linalg_solve_meta.h>
819
+ #include <ATen/ops/linalg_solve_ex_meta.h>
820
+ #include <ATen/ops/linalg_solve_triangular_meta.h>
821
+ #include <ATen/ops/linalg_svd_meta.h>
822
+ #include <ATen/ops/linalg_svdvals_meta.h>
823
+ #include <ATen/ops/linalg_tensorinv_meta.h>
824
+ #include <ATen/ops/linalg_tensorsolve_meta.h>
825
+ #include <ATen/ops/linalg_vander_meta.h>
826
+ #include <ATen/ops/linalg_vecdot_meta.h>
827
+ #include <ATen/ops/linalg_vector_norm_meta.h>
828
+ #include <ATen/ops/linear_meta.h>
829
+ #include <ATen/ops/linear_backward_meta.h>
830
+ #include <ATen/ops/linspace_meta.h>
831
+ #include <ATen/ops/log_meta.h>
832
+ #include <ATen/ops/log10_meta.h>
833
+ #include <ATen/ops/log1p_meta.h>
834
+ #include <ATen/ops/log2_meta.h>
835
+ #include <ATen/ops/log_normal_meta.h>
836
+ #include <ATen/ops/log_sigmoid_meta.h>
837
+ #include <ATen/ops/log_sigmoid_backward_meta.h>
838
+ #include <ATen/ops/log_sigmoid_forward_meta.h>
839
+ #include <ATen/ops/log_softmax_meta.h>
840
+ #include <ATen/ops/logaddexp_meta.h>
841
+ #include <ATen/ops/logaddexp2_meta.h>
842
+ #include <ATen/ops/logcumsumexp_meta.h>
843
+ #include <ATen/ops/logdet_meta.h>
844
+ #include <ATen/ops/logical_and_meta.h>
845
+ #include <ATen/ops/logical_not_meta.h>
846
+ #include <ATen/ops/logical_or_meta.h>
847
+ #include <ATen/ops/logical_xor_meta.h>
848
+ #include <ATen/ops/logit_meta.h>
849
+ #include <ATen/ops/logit_backward_meta.h>
850
+ #include <ATen/ops/logspace_meta.h>
851
+ #include <ATen/ops/logsumexp_meta.h>
852
+ #include <ATen/ops/lshift_meta.h>
853
+ #include <ATen/ops/lstm_meta.h>
854
+ #include <ATen/ops/lstm_cell_meta.h>
855
+ #include <ATen/ops/lstm_mps_backward_meta.h>
856
+ #include <ATen/ops/lt_meta.h>
857
+ #include <ATen/ops/lu_solve_meta.h>
858
+ #include <ATen/ops/lu_unpack_meta.h>
859
+ #include <ATen/ops/mH_meta.h>
860
+ #include <ATen/ops/mT_meta.h>
861
+ #include <ATen/ops/margin_ranking_loss_meta.h>
862
+ #include <ATen/ops/masked_fill_meta.h>
863
+ #include <ATen/ops/masked_scatter_meta.h>
864
+ #include <ATen/ops/masked_scatter_backward_meta.h>
865
+ #include <ATen/ops/masked_select_meta.h>
866
+ #include <ATen/ops/masked_select_backward_meta.h>
867
+ #include <ATen/ops/matmul_meta.h>
868
+ #include <ATen/ops/matmul_backward_meta.h>
869
+ #include <ATen/ops/matrix_H_meta.h>
870
+ #include <ATen/ops/matrix_exp_meta.h>
871
+ #include <ATen/ops/matrix_exp_backward_meta.h>
872
+ #include <ATen/ops/matrix_power_meta.h>
873
+ #include <ATen/ops/max_meta.h>
874
+ #include <ATen/ops/max_pool1d_meta.h>
875
+ #include <ATen/ops/max_pool1d_with_indices_meta.h>
876
+ #include <ATen/ops/max_pool2d_meta.h>
877
+ #include <ATen/ops/max_pool2d_backward_meta.h>
878
+ #include <ATen/ops/max_pool2d_with_indices_meta.h>
879
+ #include <ATen/ops/max_pool2d_with_indices_backward_meta.h>
880
+ #include <ATen/ops/max_pool3d_meta.h>
881
+ #include <ATen/ops/max_pool3d_with_indices_meta.h>
882
+ #include <ATen/ops/max_pool3d_with_indices_backward_meta.h>
883
+ #include <ATen/ops/max_unpool2d_meta.h>
884
+ #include <ATen/ops/max_unpool3d_meta.h>
885
+ #include <ATen/ops/maximum_meta.h>
886
+ #include <ATen/ops/mean_meta.h>
887
+ #include <ATen/ops/median_meta.h>
888
+ #include <ATen/ops/meshgrid_meta.h>
889
+ #include <ATen/ops/min_meta.h>
890
+ #include <ATen/ops/minimum_meta.h>
891
+ #include <ATen/ops/miopen_batch_norm_meta.h>
892
+ #include <ATen/ops/miopen_batch_norm_backward_meta.h>
893
+ #include <ATen/ops/miopen_convolution_meta.h>
894
+ #include <ATen/ops/miopen_convolution_add_relu_meta.h>
895
+ #include <ATen/ops/miopen_convolution_relu_meta.h>
896
+ #include <ATen/ops/miopen_convolution_transpose_meta.h>
897
+ #include <ATen/ops/miopen_depthwise_convolution_meta.h>
898
+ #include <ATen/ops/miopen_rnn_meta.h>
899
+ #include <ATen/ops/miopen_rnn_backward_meta.h>
900
+ #include <ATen/ops/mish_meta.h>
901
+ #include <ATen/ops/mish_backward_meta.h>
902
+ #include <ATen/ops/mkldnn_adaptive_avg_pool2d_meta.h>
903
+ #include <ATen/ops/mkldnn_adaptive_avg_pool2d_backward_meta.h>
904
+ #include <ATen/ops/mkldnn_convolution_meta.h>
905
+ #include <ATen/ops/mkldnn_linear_meta.h>
906
+ #include <ATen/ops/mkldnn_linear_backward_meta.h>
907
+ #include <ATen/ops/mkldnn_linear_backward_input_meta.h>
908
+ #include <ATen/ops/mkldnn_linear_backward_weights_meta.h>
909
+ #include <ATen/ops/mkldnn_max_pool2d_meta.h>
910
+ #include <ATen/ops/mkldnn_max_pool2d_backward_meta.h>
911
+ #include <ATen/ops/mkldnn_max_pool3d_meta.h>
912
+ #include <ATen/ops/mkldnn_max_pool3d_backward_meta.h>
913
+ #include <ATen/ops/mkldnn_reorder_conv2d_weight_meta.h>
914
+ #include <ATen/ops/mkldnn_reorder_conv3d_weight_meta.h>
915
+ #include <ATen/ops/mkldnn_rnn_layer_meta.h>
916
+ #include <ATen/ops/mkldnn_rnn_layer_backward_meta.h>
917
+ #include <ATen/ops/mm_meta.h>
918
+ #include <ATen/ops/mode_meta.h>
919
+ #include <ATen/ops/moveaxis_meta.h>
920
+ #include <ATen/ops/movedim_meta.h>
921
+ #include <ATen/ops/mps_convolution_backward_meta.h>
922
+ #include <ATen/ops/mps_convolution_transpose_backward_meta.h>
923
+ #include <ATen/ops/mse_loss_meta.h>
924
+ #include <ATen/ops/mse_loss_backward_meta.h>
925
+ #include <ATen/ops/msort_meta.h>
926
+ #include <ATen/ops/mul_meta.h>
927
+ #include <ATen/ops/multi_margin_loss_meta.h>
928
+ #include <ATen/ops/multi_margin_loss_backward_meta.h>
929
+ #include <ATen/ops/multilabel_margin_loss_meta.h>
930
+ #include <ATen/ops/multilabel_margin_loss_backward_meta.h>
931
+ #include <ATen/ops/multilabel_margin_loss_forward_meta.h>
932
+ #include <ATen/ops/multinomial_meta.h>
933
+ #include <ATen/ops/multiply_meta.h>
934
+ #include <ATen/ops/mv_meta.h>
935
+ #include <ATen/ops/mvlgamma_meta.h>
936
+ #include <ATen/ops/nan_to_num_meta.h>
937
+ #include <ATen/ops/nanmean_meta.h>
938
+ #include <ATen/ops/nanmedian_meta.h>
939
+ #include <ATen/ops/nanquantile_meta.h>
940
+ #include <ATen/ops/nansum_meta.h>
941
+ #include <ATen/ops/narrow_meta.h>
942
+ #include <ATen/ops/narrow_copy_meta.h>
943
+ #include <ATen/ops/native_batch_norm_meta.h>
944
+ #include <ATen/ops/native_batch_norm_backward_meta.h>
945
+ #include <ATen/ops/native_channel_shuffle_meta.h>
946
+ #include <ATen/ops/native_dropout_meta.h>
947
+ #include <ATen/ops/native_dropout_backward_meta.h>
948
+ #include <ATen/ops/native_group_norm_meta.h>
949
+ #include <ATen/ops/native_group_norm_backward_meta.h>
950
+ #include <ATen/ops/native_layer_norm_meta.h>
951
+ #include <ATen/ops/native_layer_norm_backward_meta.h>
952
+ #include <ATen/ops/native_norm_meta.h>
953
+ #include <ATen/ops/ne_meta.h>
954
+ #include <ATen/ops/neg_meta.h>
955
+ #include <ATen/ops/negative_meta.h>
956
+ #include <ATen/ops/nested_to_padded_tensor_meta.h>
957
+ #include <ATen/ops/new_empty_meta.h>
958
+ #include <ATen/ops/new_empty_strided_meta.h>
959
+ #include <ATen/ops/new_full_meta.h>
960
+ #include <ATen/ops/new_ones_meta.h>
961
+ #include <ATen/ops/new_zeros_meta.h>
962
+ #include <ATen/ops/nextafter_meta.h>
963
+ #include <ATen/ops/nll_loss_meta.h>
964
+ #include <ATen/ops/nll_loss2d_meta.h>
965
+ #include <ATen/ops/nll_loss2d_backward_meta.h>
966
+ #include <ATen/ops/nll_loss2d_forward_meta.h>
967
+ #include <ATen/ops/nll_loss_backward_meta.h>
968
+ #include <ATen/ops/nll_loss_forward_meta.h>
969
+ #include <ATen/ops/nll_loss_nd_meta.h>
970
+ #include <ATen/ops/nonzero_meta.h>
971
+ #include <ATen/ops/nonzero_numpy_meta.h>
972
+ #include <ATen/ops/nonzero_static_meta.h>
973
+ #include <ATen/ops/norm_meta.h>
974
+ #include <ATen/ops/norm_except_dim_meta.h>
975
+ #include <ATen/ops/normal_meta.h>
976
+ #include <ATen/ops/not_equal_meta.h>
977
+ #include <ATen/ops/nuclear_norm_meta.h>
978
+ #include <ATen/ops/numpy_T_meta.h>
979
+ #include <ATen/ops/one_hot_meta.h>
980
+ #include <ATen/ops/ones_meta.h>
981
+ #include <ATen/ops/ones_like_meta.h>
982
+ #include <ATen/ops/or_meta.h>
983
+ #include <ATen/ops/orgqr_meta.h>
984
+ #include <ATen/ops/ormqr_meta.h>
985
+ #include <ATen/ops/outer_meta.h>
986
+ #include <ATen/ops/output_nr_meta.h>
987
+ #include <ATen/ops/pad_meta.h>
988
+ #include <ATen/ops/pad_sequence_meta.h>
989
+ #include <ATen/ops/pairwise_distance_meta.h>
990
+ #include <ATen/ops/pdist_meta.h>
991
+ #include <ATen/ops/permute_meta.h>
992
+ #include <ATen/ops/permute_copy_meta.h>
993
+ #include <ATen/ops/pin_memory_meta.h>
994
+ #include <ATen/ops/pinverse_meta.h>
995
+ #include <ATen/ops/pixel_shuffle_meta.h>
996
+ #include <ATen/ops/pixel_unshuffle_meta.h>
997
+ #include <ATen/ops/poisson_meta.h>
998
+ #include <ATen/ops/poisson_nll_loss_meta.h>
999
+ #include <ATen/ops/polar_meta.h>
1000
+ #include <ATen/ops/polygamma_meta.h>
1001
+ #include <ATen/ops/positive_meta.h>
1002
+ #include <ATen/ops/pow_meta.h>
1003
+ #include <ATen/ops/prelu_meta.h>
1004
+ #include <ATen/ops/prod_meta.h>
1005
+ #include <ATen/ops/promote_types_meta.h>
1006
+ #include <ATen/ops/put_meta.h>
1007
+ #include <ATen/ops/q_per_channel_axis_meta.h>
1008
+ #include <ATen/ops/q_per_channel_scales_meta.h>
1009
+ #include <ATen/ops/q_per_channel_zero_points_meta.h>
1010
+ #include <ATen/ops/q_scale_meta.h>
1011
+ #include <ATen/ops/q_zero_point_meta.h>
1012
+ #include <ATen/ops/qr_meta.h>
1013
+ #include <ATen/ops/qscheme_meta.h>
1014
+ #include <ATen/ops/quantile_meta.h>
1015
+ #include <ATen/ops/quantize_per_channel_meta.h>
1016
+ #include <ATen/ops/quantize_per_tensor_meta.h>
1017
+ #include <ATen/ops/quantize_per_tensor_dynamic_meta.h>
1018
+ #include <ATen/ops/quantized_batch_norm_meta.h>
1019
+ #include <ATen/ops/quantized_gru_cell_meta.h>
1020
+ #include <ATen/ops/quantized_lstm_cell_meta.h>
1021
+ #include <ATen/ops/quantized_max_pool1d_meta.h>
1022
+ #include <ATen/ops/quantized_max_pool2d_meta.h>
1023
+ #include <ATen/ops/quantized_max_pool3d_meta.h>
1024
+ #include <ATen/ops/quantized_rnn_relu_cell_meta.h>
1025
+ #include <ATen/ops/quantized_rnn_tanh_cell_meta.h>
1026
+ #include <ATen/ops/rad2deg_meta.h>
1027
+ #include <ATen/ops/rand_meta.h>
1028
+ #include <ATen/ops/rand_like_meta.h>
1029
+ #include <ATen/ops/randint_meta.h>
1030
+ #include <ATen/ops/randint_like_meta.h>
1031
+ #include <ATen/ops/randn_meta.h>
1032
+ #include <ATen/ops/randn_like_meta.h>
1033
+ #include <ATen/ops/random_meta.h>
1034
+ #include <ATen/ops/randperm_meta.h>
1035
+ #include <ATen/ops/range_meta.h>
1036
+ #include <ATen/ops/ravel_meta.h>
1037
+ #include <ATen/ops/real_meta.h>
1038
+ #include <ATen/ops/reciprocal_meta.h>
1039
+ #include <ATen/ops/record_stream_meta.h>
1040
+ #include <ATen/ops/refine_names_meta.h>
1041
+ #include <ATen/ops/reflection_pad1d_meta.h>
1042
+ #include <ATen/ops/reflection_pad1d_backward_meta.h>
1043
+ #include <ATen/ops/reflection_pad2d_meta.h>
1044
+ #include <ATen/ops/reflection_pad2d_backward_meta.h>
1045
+ #include <ATen/ops/reflection_pad3d_meta.h>
1046
+ #include <ATen/ops/reflection_pad3d_backward_meta.h>
1047
+ #include <ATen/ops/relu_meta.h>
1048
+ #include <ATen/ops/relu6_meta.h>
1049
+ #include <ATen/ops/remainder_meta.h>
1050
+ #include <ATen/ops/rename_meta.h>
1051
+ #include <ATen/ops/renorm_meta.h>
1052
+ #include <ATen/ops/repeat_meta.h>
1053
+ #include <ATen/ops/repeat_interleave_meta.h>
1054
+ #include <ATen/ops/replication_pad1d_meta.h>
1055
+ #include <ATen/ops/replication_pad1d_backward_meta.h>
1056
+ #include <ATen/ops/replication_pad2d_meta.h>
1057
+ #include <ATen/ops/replication_pad2d_backward_meta.h>
1058
+ #include <ATen/ops/replication_pad3d_meta.h>
1059
+ #include <ATen/ops/replication_pad3d_backward_meta.h>
1060
+ #include <ATen/ops/requires_grad_meta.h>
1061
+ #include <ATen/ops/reshape_meta.h>
1062
+ #include <ATen/ops/reshape_as_meta.h>
1063
+ #include <ATen/ops/resize_meta.h>
1064
+ #include <ATen/ops/resize_as_meta.h>
1065
+ #include <ATen/ops/resize_as_sparse_meta.h>
1066
+ #include <ATen/ops/resolve_conj_meta.h>
1067
+ #include <ATen/ops/resolve_neg_meta.h>
1068
+ #include <ATen/ops/result_type_meta.h>
1069
+ #include <ATen/ops/retain_grad_meta.h>
1070
+ #include <ATen/ops/retains_grad_meta.h>
1071
+ #include <ATen/ops/rms_norm_meta.h>
1072
+ #include <ATen/ops/rnn_relu_meta.h>
1073
+ #include <ATen/ops/rnn_relu_cell_meta.h>
1074
+ #include <ATen/ops/rnn_tanh_meta.h>
1075
+ #include <ATen/ops/rnn_tanh_cell_meta.h>
1076
+ #include <ATen/ops/roll_meta.h>
1077
+ #include <ATen/ops/rot90_meta.h>
1078
+ #include <ATen/ops/round_meta.h>
1079
+ #include <ATen/ops/row_indices_meta.h>
1080
+ #include <ATen/ops/row_indices_copy_meta.h>
1081
+ #include <ATen/ops/row_stack_meta.h>
1082
+ #include <ATen/ops/rrelu_meta.h>
1083
+ #include <ATen/ops/rrelu_with_noise_meta.h>
1084
+ #include <ATen/ops/rrelu_with_noise_backward_meta.h>
1085
+ #include <ATen/ops/rshift_meta.h>
1086
+ #include <ATen/ops/rsqrt_meta.h>
1087
+ #include <ATen/ops/rsub_meta.h>
1088
+ #include <ATen/ops/scalar_tensor_meta.h>
1089
+ #include <ATen/ops/scaled_dot_product_attention_meta.h>
1090
+ #include <ATen/ops/scatter_meta.h>
1091
+ #include <ATen/ops/scatter_add_meta.h>
1092
+ #include <ATen/ops/scatter_reduce_meta.h>
1093
+ #include <ATen/ops/searchsorted_meta.h>
1094
+ #include <ATen/ops/segment_reduce_meta.h>
1095
+ #include <ATen/ops/select_meta.h>
1096
+ #include <ATen/ops/select_backward_meta.h>
1097
+ #include <ATen/ops/select_copy_meta.h>
1098
+ #include <ATen/ops/select_scatter_meta.h>
1099
+ #include <ATen/ops/selu_meta.h>
1100
+ #include <ATen/ops/set_meta.h>
1101
+ #include <ATen/ops/set_data_meta.h>
1102
+ #include <ATen/ops/sgn_meta.h>
1103
+ #include <ATen/ops/sigmoid_meta.h>
1104
+ #include <ATen/ops/sigmoid_backward_meta.h>
1105
+ #include <ATen/ops/sign_meta.h>
1106
+ #include <ATen/ops/signbit_meta.h>
1107
+ #include <ATen/ops/silu_meta.h>
1108
+ #include <ATen/ops/silu_backward_meta.h>
1109
+ #include <ATen/ops/sin_meta.h>
1110
+ #include <ATen/ops/sinc_meta.h>
1111
+ #include <ATen/ops/sinh_meta.h>
1112
+ #include <ATen/ops/size_meta.h>
1113
+ #include <ATen/ops/slice_meta.h>
1114
+ #include <ATen/ops/slice_backward_meta.h>
1115
+ #include <ATen/ops/slice_copy_meta.h>
1116
+ #include <ATen/ops/slice_inverse_meta.h>
1117
+ #include <ATen/ops/slice_scatter_meta.h>
1118
+ #include <ATen/ops/slogdet_meta.h>
1119
+ #include <ATen/ops/slow_conv3d_meta.h>
1120
+ #include <ATen/ops/slow_conv3d_forward_meta.h>
1121
+ #include <ATen/ops/slow_conv_dilated2d_meta.h>
1122
+ #include <ATen/ops/slow_conv_dilated3d_meta.h>
1123
+ #include <ATen/ops/slow_conv_transpose2d_meta.h>
1124
+ #include <ATen/ops/slow_conv_transpose3d_meta.h>
1125
+ #include <ATen/ops/smm_meta.h>
1126
+ #include <ATen/ops/smooth_l1_loss_meta.h>
1127
+ #include <ATen/ops/smooth_l1_loss_backward_meta.h>
1128
+ #include <ATen/ops/soft_margin_loss_meta.h>
1129
+ #include <ATen/ops/soft_margin_loss_backward_meta.h>
1130
+ #include <ATen/ops/softmax_meta.h>
1131
+ #include <ATen/ops/softplus_meta.h>
1132
+ #include <ATen/ops/softplus_backward_meta.h>
1133
+ #include <ATen/ops/softshrink_meta.h>
1134
+ #include <ATen/ops/softshrink_backward_meta.h>
1135
+ #include <ATen/ops/sort_meta.h>
1136
+ #include <ATen/ops/sparse_bsc_tensor_meta.h>
1137
+ #include <ATen/ops/sparse_bsr_tensor_meta.h>
1138
+ #include <ATen/ops/sparse_compressed_tensor_meta.h>
1139
+ #include <ATen/ops/sparse_coo_tensor_meta.h>
1140
+ #include <ATen/ops/sparse_csc_tensor_meta.h>
1141
+ #include <ATen/ops/sparse_csr_tensor_meta.h>
1142
+ #include <ATen/ops/sparse_dim_meta.h>
1143
+ #include <ATen/ops/sparse_mask_meta.h>
1144
+ #include <ATen/ops/sparse_resize_meta.h>
1145
+ #include <ATen/ops/sparse_resize_and_clear_meta.h>
1146
+ #include <ATen/ops/sparse_sampled_addmm_meta.h>
1147
+ #include <ATen/ops/special_airy_ai_meta.h>
1148
+ #include <ATen/ops/special_bessel_j0_meta.h>
1149
+ #include <ATen/ops/special_bessel_j1_meta.h>
1150
+ #include <ATen/ops/special_bessel_y0_meta.h>
1151
+ #include <ATen/ops/special_bessel_y1_meta.h>
1152
+ #include <ATen/ops/special_chebyshev_polynomial_t_meta.h>
1153
+ #include <ATen/ops/special_chebyshev_polynomial_u_meta.h>
1154
+ #include <ATen/ops/special_chebyshev_polynomial_v_meta.h>
1155
+ #include <ATen/ops/special_chebyshev_polynomial_w_meta.h>
1156
+ #include <ATen/ops/special_digamma_meta.h>
1157
+ #include <ATen/ops/special_entr_meta.h>
1158
+ #include <ATen/ops/special_erf_meta.h>
1159
+ #include <ATen/ops/special_erfc_meta.h>
1160
+ #include <ATen/ops/special_erfcx_meta.h>
1161
+ #include <ATen/ops/special_erfinv_meta.h>
1162
+ #include <ATen/ops/special_exp2_meta.h>
1163
+ #include <ATen/ops/special_expit_meta.h>
1164
+ #include <ATen/ops/special_expm1_meta.h>
1165
+ #include <ATen/ops/special_gammainc_meta.h>
1166
+ #include <ATen/ops/special_gammaincc_meta.h>
1167
+ #include <ATen/ops/special_gammaln_meta.h>
1168
+ #include <ATen/ops/special_hermite_polynomial_h_meta.h>
1169
+ #include <ATen/ops/special_hermite_polynomial_he_meta.h>
1170
+ #include <ATen/ops/special_i0_meta.h>
1171
+ #include <ATen/ops/special_i0e_meta.h>
1172
+ #include <ATen/ops/special_i1_meta.h>
1173
+ #include <ATen/ops/special_i1e_meta.h>
1174
+ #include <ATen/ops/special_laguerre_polynomial_l_meta.h>
1175
+ #include <ATen/ops/special_legendre_polynomial_p_meta.h>
1176
+ #include <ATen/ops/special_log1p_meta.h>
1177
+ #include <ATen/ops/special_log_ndtr_meta.h>
1178
+ #include <ATen/ops/special_log_softmax_meta.h>
1179
+ #include <ATen/ops/special_logit_meta.h>
1180
+ #include <ATen/ops/special_logsumexp_meta.h>
1181
+ #include <ATen/ops/special_modified_bessel_i0_meta.h>
1182
+ #include <ATen/ops/special_modified_bessel_i1_meta.h>
1183
+ #include <ATen/ops/special_modified_bessel_k0_meta.h>
1184
+ #include <ATen/ops/special_modified_bessel_k1_meta.h>
1185
+ #include <ATen/ops/special_multigammaln_meta.h>
1186
+ #include <ATen/ops/special_ndtr_meta.h>
1187
+ #include <ATen/ops/special_ndtri_meta.h>
1188
+ #include <ATen/ops/special_polygamma_meta.h>
1189
+ #include <ATen/ops/special_psi_meta.h>
1190
+ #include <ATen/ops/special_round_meta.h>
1191
+ #include <ATen/ops/special_scaled_modified_bessel_k0_meta.h>
1192
+ #include <ATen/ops/special_scaled_modified_bessel_k1_meta.h>
1193
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_t_meta.h>
1194
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_u_meta.h>
1195
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_v_meta.h>
1196
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_w_meta.h>
1197
+ #include <ATen/ops/special_sinc_meta.h>
1198
+ #include <ATen/ops/special_softmax_meta.h>
1199
+ #include <ATen/ops/special_spherical_bessel_j0_meta.h>
1200
+ #include <ATen/ops/special_xlog1py_meta.h>
1201
+ #include <ATen/ops/special_xlogy_meta.h>
1202
+ #include <ATen/ops/special_zeta_meta.h>
1203
+ #include <ATen/ops/split_meta.h>
1204
+ #include <ATen/ops/split_copy_meta.h>
1205
+ #include <ATen/ops/split_with_sizes_meta.h>
1206
+ #include <ATen/ops/split_with_sizes_copy_meta.h>
1207
+ #include <ATen/ops/sqrt_meta.h>
1208
+ #include <ATen/ops/square_meta.h>
1209
+ #include <ATen/ops/squeeze_meta.h>
1210
+ #include <ATen/ops/squeeze_copy_meta.h>
1211
+ #include <ATen/ops/sspaddmm_meta.h>
1212
+ #include <ATen/ops/stack_meta.h>
1213
+ #include <ATen/ops/std_meta.h>
1214
+ #include <ATen/ops/std_mean_meta.h>
1215
+ #include <ATen/ops/stft_meta.h>
1216
+ #include <ATen/ops/stride_meta.h>
1217
+ #include <ATen/ops/sub_meta.h>
1218
+ #include <ATen/ops/subtract_meta.h>
1219
+ #include <ATen/ops/sum_meta.h>
1220
+ #include <ATen/ops/sum_to_size_meta.h>
1221
+ #include <ATen/ops/svd_meta.h>
1222
+ #include <ATen/ops/swapaxes_meta.h>
1223
+ #include <ATen/ops/swapdims_meta.h>
1224
+ #include <ATen/ops/sym_constrain_range_meta.h>
1225
+ #include <ATen/ops/sym_constrain_range_for_size_meta.h>
1226
+ #include <ATen/ops/sym_numel_meta.h>
1227
+ #include <ATen/ops/sym_size_meta.h>
1228
+ #include <ATen/ops/sym_storage_offset_meta.h>
1229
+ #include <ATen/ops/sym_stride_meta.h>
1230
+ #include <ATen/ops/t_meta.h>
1231
+ #include <ATen/ops/t_copy_meta.h>
1232
+ #include <ATen/ops/take_meta.h>
1233
+ #include <ATen/ops/take_along_dim_meta.h>
1234
+ #include <ATen/ops/tan_meta.h>
1235
+ #include <ATen/ops/tanh_meta.h>
1236
+ #include <ATen/ops/tanh_backward_meta.h>
1237
+ #include <ATen/ops/tensor_split_meta.h>
1238
+ #include <ATen/ops/tensordot_meta.h>
1239
+ #include <ATen/ops/thnn_conv2d_meta.h>
1240
+ #include <ATen/ops/threshold_meta.h>
1241
+ #include <ATen/ops/threshold_backward_meta.h>
1242
+ #include <ATen/ops/tile_meta.h>
1243
+ #include <ATen/ops/to_meta.h>
1244
+ #include <ATen/ops/to_dense_meta.h>
1245
+ #include <ATen/ops/to_dense_backward_meta.h>
1246
+ #include <ATen/ops/to_mkldnn_meta.h>
1247
+ #include <ATen/ops/to_mkldnn_backward_meta.h>
1248
+ #include <ATen/ops/to_padded_tensor_meta.h>
1249
+ #include <ATen/ops/to_sparse_meta.h>
1250
+ #include <ATen/ops/to_sparse_bsc_meta.h>
1251
+ #include <ATen/ops/to_sparse_bsr_meta.h>
1252
+ #include <ATen/ops/to_sparse_csc_meta.h>
1253
+ #include <ATen/ops/to_sparse_csr_meta.h>
1254
+ #include <ATen/ops/topk_meta.h>
1255
+ #include <ATen/ops/trace_meta.h>
1256
+ #include <ATen/ops/trace_backward_meta.h>
1257
+ #include <ATen/ops/transpose_meta.h>
1258
+ #include <ATen/ops/transpose_copy_meta.h>
1259
+ #include <ATen/ops/trapezoid_meta.h>
1260
+ #include <ATen/ops/trapz_meta.h>
1261
+ #include <ATen/ops/triangular_solve_meta.h>
1262
+ #include <ATen/ops/tril_meta.h>
1263
+ #include <ATen/ops/tril_indices_meta.h>
1264
+ #include <ATen/ops/triplet_margin_loss_meta.h>
1265
+ #include <ATen/ops/triu_meta.h>
1266
+ #include <ATen/ops/triu_indices_meta.h>
1267
+ #include <ATen/ops/true_divide_meta.h>
1268
+ #include <ATen/ops/trunc_meta.h>
1269
+ #include <ATen/ops/type_as_meta.h>
1270
+ #include <ATen/ops/unbind_meta.h>
1271
+ #include <ATen/ops/unbind_copy_meta.h>
1272
+ #include <ATen/ops/unflatten_meta.h>
1273
+ #include <ATen/ops/unflatten_dense_tensors_meta.h>
1274
+ #include <ATen/ops/unfold_meta.h>
1275
+ #include <ATen/ops/unfold_backward_meta.h>
1276
+ #include <ATen/ops/unfold_copy_meta.h>
1277
+ #include <ATen/ops/uniform_meta.h>
1278
+ #include <ATen/ops/unique_consecutive_meta.h>
1279
+ #include <ATen/ops/unique_dim_meta.h>
1280
+ #include <ATen/ops/unique_dim_consecutive_meta.h>
1281
+ #include <ATen/ops/unsafe_chunk_meta.h>
1282
+ #include <ATen/ops/unsafe_split_meta.h>
1283
+ #include <ATen/ops/unsafe_split_with_sizes_meta.h>
1284
+ #include <ATen/ops/unsqueeze_meta.h>
1285
+ #include <ATen/ops/unsqueeze_copy_meta.h>
1286
+ #include <ATen/ops/upsample_bicubic2d_meta.h>
1287
+ #include <ATen/ops/upsample_bicubic2d_backward_meta.h>
1288
+ #include <ATen/ops/upsample_bilinear2d_meta.h>
1289
+ #include <ATen/ops/upsample_bilinear2d_backward_meta.h>
1290
+ #include <ATen/ops/upsample_linear1d_meta.h>
1291
+ #include <ATen/ops/upsample_linear1d_backward_meta.h>
1292
+ #include <ATen/ops/upsample_nearest1d_meta.h>
1293
+ #include <ATen/ops/upsample_nearest1d_backward_meta.h>
1294
+ #include <ATen/ops/upsample_nearest2d_meta.h>
1295
+ #include <ATen/ops/upsample_nearest2d_backward_meta.h>
1296
+ #include <ATen/ops/upsample_nearest3d_meta.h>
1297
+ #include <ATen/ops/upsample_nearest3d_backward_meta.h>
1298
+ #include <ATen/ops/upsample_trilinear3d_meta.h>
1299
+ #include <ATen/ops/upsample_trilinear3d_backward_meta.h>
1300
+ #include <ATen/ops/value_selecting_reduction_backward_meta.h>
1301
+ #include <ATen/ops/values_meta.h>
1302
+ #include <ATen/ops/values_copy_meta.h>
1303
+ #include <ATen/ops/vander_meta.h>
1304
+ #include <ATen/ops/var_meta.h>
1305
+ #include <ATen/ops/var_mean_meta.h>
1306
+ #include <ATen/ops/vdot_meta.h>
1307
+ #include <ATen/ops/view_meta.h>
1308
+ #include <ATen/ops/view_as_meta.h>
1309
+ #include <ATen/ops/view_as_complex_meta.h>
1310
+ #include <ATen/ops/view_as_complex_copy_meta.h>
1311
+ #include <ATen/ops/view_as_real_meta.h>
1312
+ #include <ATen/ops/view_as_real_copy_meta.h>
1313
+ #include <ATen/ops/view_copy_meta.h>
1314
+ #include <ATen/ops/vsplit_meta.h>
1315
+ #include <ATen/ops/vstack_meta.h>
1316
+ #include <ATen/ops/where_meta.h>
1317
+ #include <ATen/ops/xlogy_meta.h>
1318
+ #include <ATen/ops/xor_meta.h>
1319
+ #include <ATen/ops/zero_meta.h>
1320
+ #include <ATen/ops/zeros_meta.h>
1321
+ #include <ATen/ops/zeros_like_meta.h>
1322
+
1323
+ namespace at {
1324
+
1325
+ namespace meta {
1326
+
1327
+
1328
+
1329
+ } // namespace meta
1330
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/NumericUtils.h ADDED
@@ -0,0 +1,203 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #ifdef __HIPCC__
4
+ #include <hip/hip_runtime.h>
5
+ #endif
6
+
7
+ #include <c10/macros/Macros.h>
8
+ #include <c10/util/BFloat16.h>
9
+ #include <c10/util/Float8_e4m3fn.h>
10
+ #include <c10/util/Float8_e4m3fnuz.h>
11
+ #include <c10/util/Float8_e5m2.h>
12
+ #include <c10/util/Float8_e5m2fnuz.h>
13
+ #include <c10/util/Half.h>
14
+ #include <c10/util/complex.h>
15
+
16
+ #include <cmath>
17
+ #include <type_traits>
18
+
19
+ namespace at {
20
+
21
+ // std::isnan isn't performant to use on integral types; it will
22
+ // (uselessly) convert to floating point and then do the test.
23
+ // This function is.
24
+
25
+ template <typename T, std::enable_if_t<std::is_integral_v<T>, int> = 0>
26
+ inline C10_HOST_DEVICE bool _isnan(T /*val*/) {
27
+ return false;
28
+ }
29
+
30
+ template <typename T, std::enable_if_t<std::is_floating_point_v<T>, int> = 0>
31
+ inline C10_HOST_DEVICE bool _isnan(T val) {
32
+ #if defined(__CUDACC__) || defined(__HIPCC__)
33
+ return ::isnan(val);
34
+ #else
35
+ return std::isnan(val);
36
+ #endif
37
+ }
38
+
39
+ template <typename T, std::enable_if_t<c10::is_complex<T>::value, int> = 0>
40
+ inline C10_HOST_DEVICE bool _isnan(T val) {
41
+ return std::isnan(val.real()) || std::isnan(val.imag());
42
+ }
43
+
44
+ template <typename T, std::enable_if_t<std::is_same_v<T, at::Half>, int> = 0>
45
+ inline C10_HOST_DEVICE bool _isnan(T val) {
46
+ return at::_isnan(static_cast<float>(val));
47
+ }
48
+
49
+ template <
50
+ typename T,
51
+ std::enable_if_t<std::is_same_v<T, at::BFloat16>, int> = 0>
52
+ inline C10_HOST_DEVICE bool _isnan(at::BFloat16 val) {
53
+ return at::_isnan(static_cast<float>(val));
54
+ }
55
+
56
+ inline C10_HOST_DEVICE bool _isnan(at::BFloat16 val) {
57
+ return at::_isnan(static_cast<float>(val));
58
+ }
59
+
60
+ template <
61
+ typename T,
62
+ std::enable_if_t<std::is_same_v<T, at::Float8_e5m2>, int> = 0>
63
+ inline C10_HOST_DEVICE bool _isnan(T val) {
64
+ return val.isnan();
65
+ }
66
+
67
+ template <
68
+ typename T,
69
+ std::enable_if_t<std::is_same_v<T, at::Float8_e4m3fn>, int> = 0>
70
+ inline C10_HOST_DEVICE bool _isnan(T val) {
71
+ return val.isnan();
72
+ }
73
+
74
+ template <
75
+ typename T,
76
+ std::enable_if_t<std::is_same_v<T, at::Float8_e5m2fnuz>, int> = 0>
77
+ inline C10_HOST_DEVICE bool _isnan(T val) {
78
+ return val.isnan();
79
+ }
80
+
81
+ template <
82
+ typename T,
83
+ std::enable_if_t<std::is_same_v<T, at::Float8_e4m3fnuz>, int> = 0>
84
+ inline C10_HOST_DEVICE bool _isnan(T val) {
85
+ return val.isnan();
86
+ }
87
+
88
+ // std::isinf isn't performant to use on integral types; it will
89
+ // (uselessly) convert to floating point and then do the test.
90
+ // This function is.
91
+
92
+ template <typename T, std::enable_if_t<std::is_integral_v<T>, int> = 0>
93
+ inline C10_HOST_DEVICE bool _isinf(T /*val*/) {
94
+ return false;
95
+ }
96
+
97
+ template <typename T, std::enable_if_t<std::is_floating_point_v<T>, int> = 0>
98
+ inline C10_HOST_DEVICE bool _isinf(T val) {
99
+ #if defined(__CUDACC__) || defined(__HIPCC__)
100
+ return ::isinf(val);
101
+ #else
102
+ return std::isinf(val);
103
+ #endif
104
+ }
105
+
106
+ inline C10_HOST_DEVICE bool _isinf(at::Half val) {
107
+ return at::_isinf(static_cast<float>(val));
108
+ }
109
+
110
+ inline C10_HOST_DEVICE bool _isinf(at::BFloat16 val) {
111
+ return at::_isinf(static_cast<float>(val));
112
+ }
113
+
114
+ inline C10_HOST_DEVICE bool _isinf(at::Float8_e5m2 val) {
115
+ return val.isinf();
116
+ }
117
+
118
+ inline C10_HOST_DEVICE bool _isinf(at::Float8_e4m3fn val [[maybe_unused]]) {
119
+ return false;
120
+ }
121
+
122
+ inline C10_HOST_DEVICE bool _isinf(at::Float8_e5m2fnuz val [[maybe_unused]]) {
123
+ return false;
124
+ }
125
+
126
+ inline C10_HOST_DEVICE bool _isinf(at::Float8_e4m3fnuz val [[maybe_unused]]) {
127
+ return false;
128
+ }
129
+
130
+ template <typename T>
131
+ C10_HOST_DEVICE inline T exp(T x) {
132
+ static_assert(
133
+ !std::is_same_v<T, double>,
134
+ "this template must be used with float or less precise type");
135
+ #if defined(__CUDA_ARCH__) || defined(__HIP_ARCH__)
136
+ // use __expf fast approximation for peak bandwidth
137
+ return __expf(x);
138
+ #else
139
+ return ::exp(x);
140
+ #endif
141
+ }
142
+
143
+ template <>
144
+ C10_HOST_DEVICE inline double exp<double>(double x) {
145
+ return ::exp(x);
146
+ }
147
+
148
+ template <typename T>
149
+ C10_HOST_DEVICE inline T log(T x) {
150
+ static_assert(
151
+ !std::is_same_v<T, double>,
152
+ "this template must be used with float or less precise type");
153
+ #if defined(__CUDA_ARCH__) || defined(__HIP_ARCH__)
154
+ // use __logf fast approximation for peak bandwidth
155
+ return __logf(x);
156
+ #else
157
+ return ::log(x);
158
+ #endif
159
+ }
160
+
161
+ template <>
162
+ C10_HOST_DEVICE inline double log<double>(double x) {
163
+ return ::log(x);
164
+ }
165
+
166
+ template <typename T>
167
+ C10_HOST_DEVICE inline T log1p(T x) {
168
+ static_assert(
169
+ !std::is_same_v<T, double>,
170
+ "this template must be used with float or less precise type");
171
+ #if defined(__CUDA_ARCH__) || defined(__HIP_ARCH__)
172
+ // use __logf fast approximation for peak bandwidth
173
+ // NOTE: There is no __log1pf so unfortunately we lose precision.
174
+ return __logf(1.0f + x);
175
+ #else
176
+ return ::log1p(x);
177
+ #endif
178
+ }
179
+
180
+ template <>
181
+ C10_HOST_DEVICE inline double log1p<double>(double x) {
182
+ return ::log1p(x);
183
+ }
184
+
185
+ template <typename T>
186
+ C10_HOST_DEVICE inline T tan(T x) {
187
+ static_assert(
188
+ !std::is_same_v<T, double>,
189
+ "this template must be used with float or less precise type");
190
+ #if defined(__CUDA_ARCH__) || defined(__HIP_ARCH__)
191
+ // use __tanf fast approximation for peak bandwidth
192
+ return __tanf(x);
193
+ #else
194
+ return ::tan(x);
195
+ #endif
196
+ }
197
+
198
+ template <>
199
+ C10_HOST_DEVICE inline double tan<double>(double x) {
200
+ return ::tan(x);
201
+ }
202
+
203
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/OpaqueTensorImpl.h ADDED
@@ -0,0 +1,187 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <c10/core/MemoryFormat.h>
4
+ #include <c10/core/SymIntArrayRef.h>
5
+ #include <c10/core/TensorImpl.h>
6
+ #include <c10/util/Exception.h>
7
+
8
+ namespace at {
9
+
10
+ // An "Opaque" TensorImpl -- there are no strides and (for now)
11
+ // even data() is not supported (thus no pointer arithmetic).
12
+
13
+ // NOTE: We could allow data() in the future, but would have to ensure pointer
14
+ // arithmetic code is properly guarded.
15
+ //
16
+ // NOTE: This does not support resize_ (and other metadata-changing ops) because
17
+ // of `shallow_copy_and_detach`. We would need to define an interface to
18
+ // "shallow copy" in order to add support.
19
+
20
+ template <typename OpaqueHandle>
21
+ struct TORCH_API OpaqueTensorImpl : public TensorImpl {
22
+ // public constructor for now...
23
+ OpaqueTensorImpl(
24
+ at::DispatchKeySet key_set,
25
+ const caffe2::TypeMeta data_type,
26
+ c10::Device device,
27
+ OpaqueHandle opaque_handle,
28
+ c10::IntArrayRef sizes,
29
+ bool is_non_overlapping_and_dense = true)
30
+ : TensorImpl(key_set, data_type, device),
31
+ opaque_handle_(std::move(opaque_handle)) {
32
+ set_storage_access_should_throw();
33
+ set_custom_sizes_strides(SizesStridesPolicy::CustomStrides);
34
+ sizes_and_strides_.set_sizes(sizes);
35
+ refresh_numel();
36
+ // NOLINTNEXTLINE(cppcoreguidelines-prefer-member-initializer)
37
+ is_non_overlapping_and_dense_ = is_non_overlapping_and_dense;
38
+ }
39
+
40
+ // Destructor doesn't call release_resources because it's
41
+ // unnecessary; don't forget to change that if needed!
42
+ void release_resources() override {
43
+ TensorImpl::release_resources();
44
+ opaque_handle_ = {};
45
+ }
46
+
47
+ void set_size(int64_t dim, int64_t new_size) override {
48
+ AT_ERROR("opaque tensors do not have set_size");
49
+ }
50
+
51
+ void set_stride(int64_t dim, int64_t new_stride) override {
52
+ AT_ERROR("opaque tensors do not have set_stride");
53
+ }
54
+
55
+ void set_storage_offset(int64_t storage_offset) override {
56
+ AT_ERROR("opaque tensors do not have set_storage_offset");
57
+ }
58
+
59
+ #ifdef DEBUG
60
+ bool has_storage() const override {
61
+ TORCH_INTERNAL_ASSERT_DEBUG_ONLY(
62
+ !storage_, "OpaqueTensorImpl assumes that storage_ is never set");
63
+ return false;
64
+ }
65
+ #endif
66
+
67
+ /**
68
+ * Return a TensorImpl that is a shallow-copy of this TensorImpl.
69
+ *
70
+ * For usage of `version_counter` and `allow_tensor_metadata_change`,
71
+ * see NOTE [ TensorImpl Shallow-Copying ].
72
+ */
73
+ c10::intrusive_ptr<TensorImpl> shallow_copy_and_detach(
74
+ const c10::VariableVersion& version_counter,
75
+ bool allow_tensor_metadata_change) const override {
76
+ auto impl = c10::make_intrusive<OpaqueTensorImpl<OpaqueHandle>>(
77
+ key_set(),
78
+ dtype(),
79
+ device(),
80
+ opaque_handle_,
81
+ sizes_and_strides_.sizes_arrayref());
82
+ copy_tensor_metadata(
83
+ /*src_opaque_impl=*/this,
84
+ /*dest_opaque_impl=*/impl.get(),
85
+ /*version_counter=*/version_counter,
86
+ /*allow_tensor_metadata_change=*/allow_tensor_metadata_change);
87
+ impl->refresh_numel();
88
+ return impl;
89
+ }
90
+
91
+ /**
92
+ * Return a TensorImpl that is a shallow-copy of this TensorImpl.
93
+ *
94
+ * For usage of `version_counter` and `allow_tensor_metadata_change`,
95
+ * see NOTE [ TensorImpl Shallow-Copying ].
96
+ */
97
+ c10::intrusive_ptr<TensorImpl> shallow_copy_and_detach(
98
+ c10::VariableVersion&& version_counter,
99
+ bool allow_tensor_metadata_change) const override {
100
+ auto impl = c10::make_intrusive<OpaqueTensorImpl<OpaqueHandle>>(
101
+ key_set(),
102
+ dtype(),
103
+ device(),
104
+ opaque_handle_,
105
+ sizes_and_strides_.sizes_arrayref());
106
+ copy_tensor_metadata(
107
+ /*src_opaque_impl=*/this,
108
+ /*dest_opaque_impl=*/impl.get(),
109
+ /*version_counter=*/std::move(version_counter),
110
+ /*allow_tensor_metadata_change=*/allow_tensor_metadata_change);
111
+ impl->refresh_numel();
112
+ return impl;
113
+ }
114
+
115
+ /**
116
+ * Shallow-copies data from another TensorImpl into this TensorImpl.
117
+ *
118
+ * For why this function doesn't check this TensorImpl's
119
+ * `allow_tensor_metadata_change_`, see NOTE [ TensorImpl Shallow-Copying ].
120
+ */
121
+ void shallow_copy_from(const c10::intrusive_ptr<TensorImpl>& impl) override {
122
+ AT_ASSERT(has_compatible_shallow_copy_type(impl->key_set()));
123
+ auto opaque_impl =
124
+ static_cast<const OpaqueTensorImpl<OpaqueHandle>*>(impl.get());
125
+ copy_tensor_metadata(
126
+ /*src_impl=*/opaque_impl,
127
+ /*dest_impl=*/this,
128
+ /*version_counter=*/version_counter(),
129
+ /*allow_tensor_metadata_change=*/allow_tensor_metadata_change());
130
+ refresh_numel();
131
+ }
132
+
133
+ const OpaqueHandle& opaque_handle() const {
134
+ return opaque_handle_;
135
+ }
136
+
137
+ OpaqueHandle& unsafe_opaque_handle() {
138
+ return opaque_handle_;
139
+ }
140
+
141
+ protected:
142
+ /**
143
+ * Copy the tensor metadata fields (e.g. sizes / strides / storage pointer /
144
+ * storage_offset) from one TensorImpl to another TensorImpl.
145
+ *
146
+ * For usage of `version_counter` and `allow_tensor_metadata_change`, see NOTE
147
+ * [ TensorImpl Shallow-Copying ].
148
+ */
149
+ static void copy_tensor_metadata(
150
+ const OpaqueTensorImpl<OpaqueHandle>* src_opaque_impl,
151
+ OpaqueTensorImpl<OpaqueHandle>* dest_opaque_impl,
152
+ const c10::VariableVersion& version_counter,
153
+ bool allow_tensor_metadata_change) {
154
+ TensorImpl::copy_tensor_metadata(
155
+ src_opaque_impl,
156
+ dest_opaque_impl,
157
+ version_counter,
158
+ allow_tensor_metadata_change);
159
+
160
+ // OpaqueTensorImpl-specific fields.
161
+ dest_opaque_impl->opaque_handle_ = src_opaque_impl->opaque_handle_;
162
+ }
163
+
164
+ static void copy_tensor_metadata(
165
+ const OpaqueTensorImpl<OpaqueHandle>* src_opaque_impl,
166
+ OpaqueTensorImpl<OpaqueHandle>* dest_opaque_impl,
167
+ c10::VariableVersion&& version_counter,
168
+ bool allow_tensor_metadata_change) {
169
+ TensorImpl::copy_tensor_metadata(
170
+ src_opaque_impl,
171
+ dest_opaque_impl,
172
+ std::move(version_counter),
173
+ allow_tensor_metadata_change);
174
+
175
+ // OpaqueTensorImpl-specific fields.
176
+ dest_opaque_impl->opaque_handle_ = src_opaque_impl->opaque_handle_;
177
+ }
178
+
179
+ private:
180
+ const char* tensorimpl_type_name() const override {
181
+ return "OpaqueTensorImpl";
182
+ }
183
+
184
+ OpaqueHandle opaque_handle_;
185
+ };
186
+
187
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/Operators.h ADDED
@@ -0,0 +1,1385 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ // @generated by torchgen/gen.py from Operators.h
4
+
5
+ #ifdef TORCH_ASSERT_NO_OPERATORS
6
+ #error This change adds a dependency on native_functions.yaml, \
7
+ meaning the file will need to be re-compiled every time an operator \
8
+ is changed or added. Consider if your change would be better placed in \
9
+ another file, or if a more specific header might achieve the same goal. \
10
+ See NOTE: [Tensor vs. TensorBase]
11
+ #endif
12
+
13
+ #if defined(AT_PER_OPERATOR_HEADERS) && defined(TORCH_ASSERT_ONLY_METHOD_OPERATORS)
14
+ #error This change adds a dependency on all pytorch operators, meaning the \
15
+ file will need to be re-compiled every time an operator is changed or added. \
16
+ Consider including a specific operator from <ATen/ops/{my_operator}_ops.h> \
17
+ and see NOTE [TORCH_ASSERT_ONLY_METHOD_OPERATORS].
18
+ #endif
19
+
20
+ #include <c10/core/SymInt.h>
21
+ #include <c10/core/SymIntArrayRef.h>
22
+ #include <c10/core/Scalar.h>
23
+ #include <c10/core/TensorOptions.h>
24
+ #include <c10/core/QScheme.h>
25
+ #include <c10/util/OptionalArrayRef.h>
26
+ #include <tuple>
27
+ #include <vector>
28
+
29
+ #include <ATen/ops/_adaptive_avg_pool2d_ops.h>
30
+ #include <ATen/ops/_adaptive_avg_pool2d_backward_ops.h>
31
+ #include <ATen/ops/_adaptive_avg_pool3d_ops.h>
32
+ #include <ATen/ops/_adaptive_avg_pool3d_backward_ops.h>
33
+ #include <ATen/ops/_add_batch_dim_ops.h>
34
+ #include <ATen/ops/_add_relu_ops.h>
35
+ #include <ATen/ops/_addmm_activation_ops.h>
36
+ #include <ATen/ops/_aminmax_ops.h>
37
+ #include <ATen/ops/_amp_foreach_non_finite_check_and_unscale_ops.h>
38
+ #include <ATen/ops/_amp_update_scale_ops.h>
39
+ #include <ATen/ops/_assert_async_ops.h>
40
+ #include <ATen/ops/_assert_scalar_ops.h>
41
+ #include <ATen/ops/_assert_tensor_metadata_ops.h>
42
+ #include <ATen/ops/_autocast_to_full_precision_ops.h>
43
+ #include <ATen/ops/_autocast_to_reduced_precision_ops.h>
44
+ #include <ATen/ops/_backward_ops.h>
45
+ #include <ATen/ops/_batch_norm_impl_index_ops.h>
46
+ #include <ATen/ops/_batch_norm_impl_index_backward_ops.h>
47
+ #include <ATen/ops/_batch_norm_no_update_ops.h>
48
+ #include <ATen/ops/_batch_norm_with_update_ops.h>
49
+ #include <ATen/ops/_cast_Byte_ops.h>
50
+ #include <ATen/ops/_cast_Char_ops.h>
51
+ #include <ATen/ops/_cast_Double_ops.h>
52
+ #include <ATen/ops/_cast_Float_ops.h>
53
+ #include <ATen/ops/_cast_Half_ops.h>
54
+ #include <ATen/ops/_cast_Int_ops.h>
55
+ #include <ATen/ops/_cast_Long_ops.h>
56
+ #include <ATen/ops/_cast_Short_ops.h>
57
+ #include <ATen/ops/_cdist_backward_ops.h>
58
+ #include <ATen/ops/_cdist_forward_ops.h>
59
+ #include <ATen/ops/_cholesky_solve_helper_ops.h>
60
+ #include <ATen/ops/_choose_qparams_per_tensor_ops.h>
61
+ #include <ATen/ops/_chunk_cat_ops.h>
62
+ #include <ATen/ops/_coalesce_ops.h>
63
+ #include <ATen/ops/_coalesced_ops.h>
64
+ #include <ATen/ops/_compute_linear_combination_ops.h>
65
+ #include <ATen/ops/_conj_ops.h>
66
+ #include <ATen/ops/_conj_copy_ops.h>
67
+ #include <ATen/ops/_conj_physical_ops.h>
68
+ #include <ATen/ops/_conv_depthwise2d_ops.h>
69
+ #include <ATen/ops/_convert_indices_from_coo_to_csr_ops.h>
70
+ #include <ATen/ops/_convert_indices_from_csr_to_coo_ops.h>
71
+ #include <ATen/ops/_convert_weight_to_int4pack_ops.h>
72
+ #include <ATen/ops/_convolution_ops.h>
73
+ #include <ATen/ops/_convolution_double_backward_ops.h>
74
+ #include <ATen/ops/_convolution_mode_ops.h>
75
+ #include <ATen/ops/_copy_from_ops.h>
76
+ #include <ATen/ops/_copy_from_and_resize_ops.h>
77
+ #include <ATen/ops/_cslt_compress_ops.h>
78
+ #include <ATen/ops/_cslt_sparse_mm_ops.h>
79
+ #include <ATen/ops/_cslt_sparse_mm_search_ops.h>
80
+ #include <ATen/ops/_ctc_loss_ops.h>
81
+ #include <ATen/ops/_ctc_loss_backward_ops.h>
82
+ #include <ATen/ops/_cudnn_ctc_loss_ops.h>
83
+ #include <ATen/ops/_cudnn_init_dropout_state_ops.h>
84
+ #include <ATen/ops/_cudnn_rnn_ops.h>
85
+ #include <ATen/ops/_cudnn_rnn_backward_ops.h>
86
+ #include <ATen/ops/_cudnn_rnn_flatten_weight_ops.h>
87
+ #include <ATen/ops/_cufft_clear_plan_cache_ops.h>
88
+ #include <ATen/ops/_cufft_get_plan_cache_max_size_ops.h>
89
+ #include <ATen/ops/_cufft_get_plan_cache_size_ops.h>
90
+ #include <ATen/ops/_cufft_set_plan_cache_max_size_ops.h>
91
+ #include <ATen/ops/_cummax_helper_ops.h>
92
+ #include <ATen/ops/_cummin_helper_ops.h>
93
+ #include <ATen/ops/_debug_has_internal_overlap_ops.h>
94
+ #include <ATen/ops/_dimI_ops.h>
95
+ #include <ATen/ops/_dimV_ops.h>
96
+ #include <ATen/ops/_dim_arange_ops.h>
97
+ #include <ATen/ops/_dirichlet_grad_ops.h>
98
+ #include <ATen/ops/_efficient_attention_backward_ops.h>
99
+ #include <ATen/ops/_efficient_attention_forward_ops.h>
100
+ #include <ATen/ops/_efficientzerotensor_ops.h>
101
+ #include <ATen/ops/_embedding_bag_ops.h>
102
+ #include <ATen/ops/_embedding_bag_backward_ops.h>
103
+ #include <ATen/ops/_embedding_bag_dense_backward_ops.h>
104
+ #include <ATen/ops/_embedding_bag_forward_only_ops.h>
105
+ #include <ATen/ops/_embedding_bag_per_sample_weights_backward_ops.h>
106
+ #include <ATen/ops/_embedding_bag_sparse_backward_ops.h>
107
+ #include <ATen/ops/_empty_affine_quantized_ops.h>
108
+ #include <ATen/ops/_empty_per_channel_affine_quantized_ops.h>
109
+ #include <ATen/ops/_euclidean_dist_ops.h>
110
+ #include <ATen/ops/_fake_quantize_learnable_per_channel_affine_ops.h>
111
+ #include <ATen/ops/_fake_quantize_learnable_per_channel_affine_backward_ops.h>
112
+ #include <ATen/ops/_fake_quantize_learnable_per_tensor_affine_ops.h>
113
+ #include <ATen/ops/_fake_quantize_learnable_per_tensor_affine_backward_ops.h>
114
+ #include <ATen/ops/_fake_quantize_per_tensor_affine_cachemask_tensor_qparams_ops.h>
115
+ #include <ATen/ops/_fft_c2c_ops.h>
116
+ #include <ATen/ops/_fft_c2r_ops.h>
117
+ #include <ATen/ops/_fft_r2c_ops.h>
118
+ #include <ATen/ops/_fill_mem_eff_dropout_mask_ops.h>
119
+ #include <ATen/ops/_flash_attention_backward_ops.h>
120
+ #include <ATen/ops/_flash_attention_forward_ops.h>
121
+ #include <ATen/ops/_foobar_ops.h>
122
+ #include <ATen/ops/_foreach_abs_ops.h>
123
+ #include <ATen/ops/_foreach_acos_ops.h>
124
+ #include <ATen/ops/_foreach_add_ops.h>
125
+ #include <ATen/ops/_foreach_addcdiv_ops.h>
126
+ #include <ATen/ops/_foreach_addcmul_ops.h>
127
+ #include <ATen/ops/_foreach_asin_ops.h>
128
+ #include <ATen/ops/_foreach_atan_ops.h>
129
+ #include <ATen/ops/_foreach_ceil_ops.h>
130
+ #include <ATen/ops/_foreach_clamp_max_ops.h>
131
+ #include <ATen/ops/_foreach_clamp_min_ops.h>
132
+ #include <ATen/ops/_foreach_copy_ops.h>
133
+ #include <ATen/ops/_foreach_cos_ops.h>
134
+ #include <ATen/ops/_foreach_cosh_ops.h>
135
+ #include <ATen/ops/_foreach_div_ops.h>
136
+ #include <ATen/ops/_foreach_erf_ops.h>
137
+ #include <ATen/ops/_foreach_erfc_ops.h>
138
+ #include <ATen/ops/_foreach_exp_ops.h>
139
+ #include <ATen/ops/_foreach_expm1_ops.h>
140
+ #include <ATen/ops/_foreach_floor_ops.h>
141
+ #include <ATen/ops/_foreach_frac_ops.h>
142
+ #include <ATen/ops/_foreach_lerp_ops.h>
143
+ #include <ATen/ops/_foreach_lgamma_ops.h>
144
+ #include <ATen/ops/_foreach_log_ops.h>
145
+ #include <ATen/ops/_foreach_log10_ops.h>
146
+ #include <ATen/ops/_foreach_log1p_ops.h>
147
+ #include <ATen/ops/_foreach_log2_ops.h>
148
+ #include <ATen/ops/_foreach_max_ops.h>
149
+ #include <ATen/ops/_foreach_maximum_ops.h>
150
+ #include <ATen/ops/_foreach_minimum_ops.h>
151
+ #include <ATen/ops/_foreach_mul_ops.h>
152
+ #include <ATen/ops/_foreach_neg_ops.h>
153
+ #include <ATen/ops/_foreach_norm_ops.h>
154
+ #include <ATen/ops/_foreach_pow_ops.h>
155
+ #include <ATen/ops/_foreach_reciprocal_ops.h>
156
+ #include <ATen/ops/_foreach_round_ops.h>
157
+ #include <ATen/ops/_foreach_sigmoid_ops.h>
158
+ #include <ATen/ops/_foreach_sign_ops.h>
159
+ #include <ATen/ops/_foreach_sin_ops.h>
160
+ #include <ATen/ops/_foreach_sinh_ops.h>
161
+ #include <ATen/ops/_foreach_sqrt_ops.h>
162
+ #include <ATen/ops/_foreach_sub_ops.h>
163
+ #include <ATen/ops/_foreach_tan_ops.h>
164
+ #include <ATen/ops/_foreach_tanh_ops.h>
165
+ #include <ATen/ops/_foreach_trunc_ops.h>
166
+ #include <ATen/ops/_foreach_zero_ops.h>
167
+ #include <ATen/ops/_functional_assert_async_ops.h>
168
+ #include <ATen/ops/_functional_assert_scalar_ops.h>
169
+ #include <ATen/ops/_functional_sym_constrain_range_ops.h>
170
+ #include <ATen/ops/_functional_sym_constrain_range_for_size_ops.h>
171
+ #include <ATen/ops/_fused_adagrad_ops.h>
172
+ #include <ATen/ops/_fused_adam_ops.h>
173
+ #include <ATen/ops/_fused_adamw_ops.h>
174
+ #include <ATen/ops/_fused_dropout_ops.h>
175
+ #include <ATen/ops/_fused_moving_avg_obs_fq_helper_ops.h>
176
+ #include <ATen/ops/_fused_sdp_choice_ops.h>
177
+ #include <ATen/ops/_fused_sgd_ops.h>
178
+ #include <ATen/ops/_fw_primal_ops.h>
179
+ #include <ATen/ops/_fw_primal_copy_ops.h>
180
+ #include <ATen/ops/_gather_sparse_backward_ops.h>
181
+ #include <ATen/ops/_grid_sampler_2d_cpu_fallback_ops.h>
182
+ #include <ATen/ops/_grid_sampler_2d_cpu_fallback_backward_ops.h>
183
+ #include <ATen/ops/_has_compatible_shallow_copy_type_ops.h>
184
+ #include <ATen/ops/_has_same_storage_numel_ops.h>
185
+ #include <ATen/ops/_histogramdd_bin_edges_ops.h>
186
+ #include <ATen/ops/_histogramdd_from_bin_cts_ops.h>
187
+ #include <ATen/ops/_histogramdd_from_bin_tensors_ops.h>
188
+ #include <ATen/ops/_index_put_impl_ops.h>
189
+ #include <ATen/ops/_indices_ops.h>
190
+ #include <ATen/ops/_indices_copy_ops.h>
191
+ #include <ATen/ops/_int_mm_ops.h>
192
+ #include <ATen/ops/_is_all_true_ops.h>
193
+ #include <ATen/ops/_is_any_true_ops.h>
194
+ #include <ATen/ops/_is_zerotensor_ops.h>
195
+ #include <ATen/ops/_jagged_to_padded_dense_forward_ops.h>
196
+ #include <ATen/ops/_lazy_clone_ops.h>
197
+ #include <ATen/ops/_linalg_check_errors_ops.h>
198
+ #include <ATen/ops/_linalg_det_ops.h>
199
+ #include <ATen/ops/_linalg_eigh_ops.h>
200
+ #include <ATen/ops/_linalg_eigvals_ops.h>
201
+ #include <ATen/ops/_linalg_slogdet_ops.h>
202
+ #include <ATen/ops/_linalg_solve_ex_ops.h>
203
+ #include <ATen/ops/_linalg_svd_ops.h>
204
+ #include <ATen/ops/_local_scalar_dense_ops.h>
205
+ #include <ATen/ops/_log_softmax_ops.h>
206
+ #include <ATen/ops/_log_softmax_backward_data_ops.h>
207
+ #include <ATen/ops/_logcumsumexp_ops.h>
208
+ #include <ATen/ops/_lstm_mps_ops.h>
209
+ #include <ATen/ops/_lu_with_info_ops.h>
210
+ #include <ATen/ops/_make_dep_token_ops.h>
211
+ #include <ATen/ops/_make_dual_ops.h>
212
+ #include <ATen/ops/_make_dual_copy_ops.h>
213
+ #include <ATen/ops/_make_per_channel_quantized_tensor_ops.h>
214
+ #include <ATen/ops/_make_per_tensor_quantized_tensor_ops.h>
215
+ #include <ATen/ops/_masked_scale_ops.h>
216
+ #include <ATen/ops/_masked_softmax_ops.h>
217
+ #include <ATen/ops/_masked_softmax_backward_ops.h>
218
+ #include <ATen/ops/_mixed_dtypes_linear_ops.h>
219
+ #include <ATen/ops/_mkldnn_reshape_ops.h>
220
+ #include <ATen/ops/_mkldnn_transpose_ops.h>
221
+ #include <ATen/ops/_mps_convolution_ops.h>
222
+ #include <ATen/ops/_mps_convolution_transpose_ops.h>
223
+ #include <ATen/ops/_native_batch_norm_legit_ops.h>
224
+ #include <ATen/ops/_native_batch_norm_legit_no_training_ops.h>
225
+ #include <ATen/ops/_native_multi_head_attention_ops.h>
226
+ #include <ATen/ops/_neg_view_ops.h>
227
+ #include <ATen/ops/_neg_view_copy_ops.h>
228
+ #include <ATen/ops/_nested_compute_contiguous_strides_offsets_ops.h>
229
+ #include <ATen/ops/_nested_from_padded_ops.h>
230
+ #include <ATen/ops/_nested_from_padded_and_nested_example_ops.h>
231
+ #include <ATen/ops/_nested_get_jagged_dummy_ops.h>
232
+ #include <ATen/ops/_nested_get_lengths_ops.h>
233
+ #include <ATen/ops/_nested_get_max_seqlen_ops.h>
234
+ #include <ATen/ops/_nested_get_min_seqlen_ops.h>
235
+ #include <ATen/ops/_nested_get_offsets_ops.h>
236
+ #include <ATen/ops/_nested_get_ragged_idx_ops.h>
237
+ #include <ATen/ops/_nested_get_values_ops.h>
238
+ #include <ATen/ops/_nested_get_values_copy_ops.h>
239
+ #include <ATen/ops/_nested_select_backward_ops.h>
240
+ #include <ATen/ops/_nested_sum_backward_ops.h>
241
+ #include <ATen/ops/_nested_tensor_from_mask_ops.h>
242
+ #include <ATen/ops/_nested_tensor_from_mask_left_aligned_ops.h>
243
+ #include <ATen/ops/_nested_tensor_from_tensor_list_ops.h>
244
+ #include <ATen/ops/_nested_tensor_size_ops.h>
245
+ #include <ATen/ops/_nested_tensor_softmax_with_shape_ops.h>
246
+ #include <ATen/ops/_nested_tensor_storage_offsets_ops.h>
247
+ #include <ATen/ops/_nested_tensor_strides_ops.h>
248
+ #include <ATen/ops/_nested_view_from_buffer_ops.h>
249
+ #include <ATen/ops/_nested_view_from_buffer_copy_ops.h>
250
+ #include <ATen/ops/_nested_view_from_jagged_ops.h>
251
+ #include <ATen/ops/_nested_view_from_jagged_copy_ops.h>
252
+ #include <ATen/ops/_new_zeros_with_same_feature_meta_ops.h>
253
+ #include <ATen/ops/_nnpack_available_ops.h>
254
+ #include <ATen/ops/_nnpack_spatial_convolution_ops.h>
255
+ #include <ATen/ops/_nnz_ops.h>
256
+ #include <ATen/ops/_pack_padded_sequence_ops.h>
257
+ #include <ATen/ops/_pack_padded_sequence_backward_ops.h>
258
+ #include <ATen/ops/_pad_circular_ops.h>
259
+ #include <ATen/ops/_pad_enum_ops.h>
260
+ #include <ATen/ops/_pad_packed_sequence_ops.h>
261
+ #include <ATen/ops/_padded_dense_to_jagged_forward_ops.h>
262
+ #include <ATen/ops/_pdist_backward_ops.h>
263
+ #include <ATen/ops/_pdist_forward_ops.h>
264
+ #include <ATen/ops/_pin_memory_ops.h>
265
+ #include <ATen/ops/_prelu_kernel_ops.h>
266
+ #include <ATen/ops/_prelu_kernel_backward_ops.h>
267
+ #include <ATen/ops/_print_ops.h>
268
+ #include <ATen/ops/_propagate_xla_data_ops.h>
269
+ #include <ATen/ops/_remove_batch_dim_ops.h>
270
+ #include <ATen/ops/_reshape_alias_ops.h>
271
+ #include <ATen/ops/_reshape_alias_copy_ops.h>
272
+ #include <ATen/ops/_reshape_copy_ops.h>
273
+ #include <ATen/ops/_reshape_from_tensor_ops.h>
274
+ #include <ATen/ops/_resize_output_ops.h>
275
+ #include <ATen/ops/_rowwise_prune_ops.h>
276
+ #include <ATen/ops/_safe_softmax_ops.h>
277
+ #include <ATen/ops/_sample_dirichlet_ops.h>
278
+ #include <ATen/ops/_saturate_weight_to_fp16_ops.h>
279
+ #include <ATen/ops/_scaled_dot_product_attention_math_ops.h>
280
+ #include <ATen/ops/_scaled_dot_product_attention_math_for_mps_ops.h>
281
+ #include <ATen/ops/_scaled_dot_product_cudnn_attention_ops.h>
282
+ #include <ATen/ops/_scaled_dot_product_cudnn_attention_backward_ops.h>
283
+ #include <ATen/ops/_scaled_dot_product_efficient_attention_ops.h>
284
+ #include <ATen/ops/_scaled_dot_product_efficient_attention_backward_ops.h>
285
+ #include <ATen/ops/_scaled_dot_product_flash_attention_ops.h>
286
+ #include <ATen/ops/_scaled_dot_product_flash_attention_backward_ops.h>
287
+ #include <ATen/ops/_scaled_dot_product_flash_attention_for_cpu_ops.h>
288
+ #include <ATen/ops/_scaled_dot_product_flash_attention_for_cpu_backward_ops.h>
289
+ #include <ATen/ops/_scaled_dot_product_fused_attention_overrideable_ops.h>
290
+ #include <ATen/ops/_scaled_dot_product_fused_attention_overrideable_backward_ops.h>
291
+ #include <ATen/ops/_scaled_mm_ops.h>
292
+ #include <ATen/ops/_segment_reduce_backward_ops.h>
293
+ #include <ATen/ops/_shape_as_tensor_ops.h>
294
+ #include <ATen/ops/_slow_conv2d_backward_ops.h>
295
+ #include <ATen/ops/_slow_conv2d_forward_ops.h>
296
+ #include <ATen/ops/_sobol_engine_draw_ops.h>
297
+ #include <ATen/ops/_sobol_engine_ff_ops.h>
298
+ #include <ATen/ops/_sobol_engine_initialize_state_ops.h>
299
+ #include <ATen/ops/_sobol_engine_scramble_ops.h>
300
+ #include <ATen/ops/_softmax_ops.h>
301
+ #include <ATen/ops/_softmax_backward_data_ops.h>
302
+ #include <ATen/ops/_sparse_addmm_ops.h>
303
+ #include <ATen/ops/_sparse_broadcast_to_ops.h>
304
+ #include <ATen/ops/_sparse_broadcast_to_copy_ops.h>
305
+ #include <ATen/ops/_sparse_bsc_tensor_unsafe_ops.h>
306
+ #include <ATen/ops/_sparse_bsr_tensor_unsafe_ops.h>
307
+ #include <ATen/ops/_sparse_compressed_tensor_unsafe_ops.h>
308
+ #include <ATen/ops/_sparse_compressed_tensor_with_dims_ops.h>
309
+ #include <ATen/ops/_sparse_coo_tensor_unsafe_ops.h>
310
+ #include <ATen/ops/_sparse_coo_tensor_with_dims_ops.h>
311
+ #include <ATen/ops/_sparse_coo_tensor_with_dims_and_tensors_ops.h>
312
+ #include <ATen/ops/_sparse_csc_tensor_unsafe_ops.h>
313
+ #include <ATen/ops/_sparse_csr_prod_ops.h>
314
+ #include <ATen/ops/_sparse_csr_sum_ops.h>
315
+ #include <ATen/ops/_sparse_csr_tensor_unsafe_ops.h>
316
+ #include <ATen/ops/_sparse_log_softmax_ops.h>
317
+ #include <ATen/ops/_sparse_log_softmax_backward_data_ops.h>
318
+ #include <ATen/ops/_sparse_mask_projection_ops.h>
319
+ #include <ATen/ops/_sparse_mm_ops.h>
320
+ #include <ATen/ops/_sparse_mm_reduce_impl_ops.h>
321
+ #include <ATen/ops/_sparse_mm_reduce_impl_backward_ops.h>
322
+ #include <ATen/ops/_sparse_semi_structured_addmm_ops.h>
323
+ #include <ATen/ops/_sparse_semi_structured_apply_ops.h>
324
+ #include <ATen/ops/_sparse_semi_structured_apply_dense_ops.h>
325
+ #include <ATen/ops/_sparse_semi_structured_linear_ops.h>
326
+ #include <ATen/ops/_sparse_semi_structured_mm_ops.h>
327
+ #include <ATen/ops/_sparse_semi_structured_tile_ops.h>
328
+ #include <ATen/ops/_sparse_softmax_ops.h>
329
+ #include <ATen/ops/_sparse_softmax_backward_data_ops.h>
330
+ #include <ATen/ops/_sparse_sparse_matmul_ops.h>
331
+ #include <ATen/ops/_sparse_sum_ops.h>
332
+ #include <ATen/ops/_sparse_sum_backward_ops.h>
333
+ #include <ATen/ops/_spdiags_ops.h>
334
+ #include <ATen/ops/_spsolve_ops.h>
335
+ #include <ATen/ops/_stack_ops.h>
336
+ #include <ATen/ops/_standard_gamma_ops.h>
337
+ #include <ATen/ops/_standard_gamma_grad_ops.h>
338
+ #include <ATen/ops/_test_ambiguous_defaults_ops.h>
339
+ #include <ATen/ops/_test_autograd_multiple_dispatch_ops.h>
340
+ #include <ATen/ops/_test_autograd_multiple_dispatch_view_ops.h>
341
+ #include <ATen/ops/_test_autograd_multiple_dispatch_view_copy_ops.h>
342
+ #include <ATen/ops/_test_check_tensor_ops.h>
343
+ #include <ATen/ops/_test_functorch_fallback_ops.h>
344
+ #include <ATen/ops/_test_optional_filled_intlist_ops.h>
345
+ #include <ATen/ops/_test_optional_floatlist_ops.h>
346
+ #include <ATen/ops/_test_optional_intlist_ops.h>
347
+ #include <ATen/ops/_test_parallel_materialize_ops.h>
348
+ #include <ATen/ops/_test_serialization_subcmul_ops.h>
349
+ #include <ATen/ops/_test_string_default_ops.h>
350
+ #include <ATen/ops/_test_warn_in_autograd_ops.h>
351
+ #include <ATen/ops/_thnn_differentiable_gru_cell_backward_ops.h>
352
+ #include <ATen/ops/_thnn_differentiable_lstm_cell_backward_ops.h>
353
+ #include <ATen/ops/_thnn_fused_gru_cell_ops.h>
354
+ #include <ATen/ops/_thnn_fused_gru_cell_backward_ops.h>
355
+ #include <ATen/ops/_thnn_fused_lstm_cell_ops.h>
356
+ #include <ATen/ops/_thnn_fused_lstm_cell_backward_ops.h>
357
+ #include <ATen/ops/_thnn_fused_lstm_cell_backward_impl_ops.h>
358
+ #include <ATen/ops/_to_copy_ops.h>
359
+ #include <ATen/ops/_to_cpu_ops.h>
360
+ #include <ATen/ops/_to_dense_ops.h>
361
+ #include <ATen/ops/_to_sparse_ops.h>
362
+ #include <ATen/ops/_to_sparse_bsc_ops.h>
363
+ #include <ATen/ops/_to_sparse_bsr_ops.h>
364
+ #include <ATen/ops/_to_sparse_csc_ops.h>
365
+ #include <ATen/ops/_to_sparse_csr_ops.h>
366
+ #include <ATen/ops/_to_sparse_semi_structured_ops.h>
367
+ #include <ATen/ops/_transform_bias_rescale_qkv_ops.h>
368
+ #include <ATen/ops/_transformer_encoder_layer_fwd_ops.h>
369
+ #include <ATen/ops/_trilinear_ops.h>
370
+ #include <ATen/ops/_triton_multi_head_attention_ops.h>
371
+ #include <ATen/ops/_triton_scaled_dot_attention_ops.h>
372
+ #include <ATen/ops/_unique_ops.h>
373
+ #include <ATen/ops/_unique2_ops.h>
374
+ #include <ATen/ops/_unpack_dual_ops.h>
375
+ #include <ATen/ops/_unsafe_index_ops.h>
376
+ #include <ATen/ops/_unsafe_index_put_ops.h>
377
+ #include <ATen/ops/_unsafe_masked_index_ops.h>
378
+ #include <ATen/ops/_unsafe_masked_index_put_accumulate_ops.h>
379
+ #include <ATen/ops/_unsafe_view_ops.h>
380
+ #include <ATen/ops/_upsample_bicubic2d_aa_ops.h>
381
+ #include <ATen/ops/_upsample_bicubic2d_aa_backward_ops.h>
382
+ #include <ATen/ops/_upsample_bilinear2d_aa_ops.h>
383
+ #include <ATen/ops/_upsample_bilinear2d_aa_backward_ops.h>
384
+ #include <ATen/ops/_upsample_nearest_exact1d_ops.h>
385
+ #include <ATen/ops/_upsample_nearest_exact1d_backward_ops.h>
386
+ #include <ATen/ops/_upsample_nearest_exact2d_ops.h>
387
+ #include <ATen/ops/_upsample_nearest_exact2d_backward_ops.h>
388
+ #include <ATen/ops/_upsample_nearest_exact3d_ops.h>
389
+ #include <ATen/ops/_upsample_nearest_exact3d_backward_ops.h>
390
+ #include <ATen/ops/_use_cudnn_ctc_loss_ops.h>
391
+ #include <ATen/ops/_use_cudnn_rnn_flatten_weight_ops.h>
392
+ #include <ATen/ops/_validate_compressed_sparse_indices_ops.h>
393
+ #include <ATen/ops/_validate_sparse_bsc_tensor_args_ops.h>
394
+ #include <ATen/ops/_validate_sparse_bsr_tensor_args_ops.h>
395
+ #include <ATen/ops/_validate_sparse_compressed_tensor_args_ops.h>
396
+ #include <ATen/ops/_validate_sparse_coo_tensor_args_ops.h>
397
+ #include <ATen/ops/_validate_sparse_csc_tensor_args_ops.h>
398
+ #include <ATen/ops/_validate_sparse_csr_tensor_args_ops.h>
399
+ #include <ATen/ops/_values_ops.h>
400
+ #include <ATen/ops/_values_copy_ops.h>
401
+ #include <ATen/ops/_version_ops.h>
402
+ #include <ATen/ops/_weight_int4pack_mm_ops.h>
403
+ #include <ATen/ops/_weight_int8pack_mm_ops.h>
404
+ #include <ATen/ops/_weight_norm_ops.h>
405
+ #include <ATen/ops/_weight_norm_differentiable_backward_ops.h>
406
+ #include <ATen/ops/_weight_norm_interface_ops.h>
407
+ #include <ATen/ops/_weight_norm_interface_backward_ops.h>
408
+ #include <ATen/ops/_wrapped_linear_prepack_ops.h>
409
+ #include <ATen/ops/_wrapped_quantized_linear_prepacked_ops.h>
410
+ #include <ATen/ops/abs_ops.h>
411
+ #include <ATen/ops/absolute_ops.h>
412
+ #include <ATen/ops/acos_ops.h>
413
+ #include <ATen/ops/acosh_ops.h>
414
+ #include <ATen/ops/adaptive_avg_pool1d_ops.h>
415
+ #include <ATen/ops/adaptive_avg_pool2d_ops.h>
416
+ #include <ATen/ops/adaptive_avg_pool3d_ops.h>
417
+ #include <ATen/ops/adaptive_avg_pool3d_backward_ops.h>
418
+ #include <ATen/ops/adaptive_max_pool1d_ops.h>
419
+ #include <ATen/ops/adaptive_max_pool2d_ops.h>
420
+ #include <ATen/ops/adaptive_max_pool2d_backward_ops.h>
421
+ #include <ATen/ops/adaptive_max_pool3d_ops.h>
422
+ #include <ATen/ops/adaptive_max_pool3d_backward_ops.h>
423
+ #include <ATen/ops/add_ops.h>
424
+ #include <ATen/ops/addbmm_ops.h>
425
+ #include <ATen/ops/addcdiv_ops.h>
426
+ #include <ATen/ops/addcmul_ops.h>
427
+ #include <ATen/ops/addmm_ops.h>
428
+ #include <ATen/ops/addmv_ops.h>
429
+ #include <ATen/ops/addr_ops.h>
430
+ #include <ATen/ops/adjoint_ops.h>
431
+ #include <ATen/ops/affine_grid_generator_ops.h>
432
+ #include <ATen/ops/affine_grid_generator_backward_ops.h>
433
+ #include <ATen/ops/alias_ops.h>
434
+ #include <ATen/ops/alias_copy_ops.h>
435
+ #include <ATen/ops/align_as_ops.h>
436
+ #include <ATen/ops/align_tensors_ops.h>
437
+ #include <ATen/ops/align_to_ops.h>
438
+ #include <ATen/ops/all_ops.h>
439
+ #include <ATen/ops/allclose_ops.h>
440
+ #include <ATen/ops/alpha_dropout_ops.h>
441
+ #include <ATen/ops/amax_ops.h>
442
+ #include <ATen/ops/amin_ops.h>
443
+ #include <ATen/ops/aminmax_ops.h>
444
+ #include <ATen/ops/and_ops.h>
445
+ #include <ATen/ops/angle_ops.h>
446
+ #include <ATen/ops/any_ops.h>
447
+ #include <ATen/ops/arange_ops.h>
448
+ #include <ATen/ops/arccos_ops.h>
449
+ #include <ATen/ops/arccosh_ops.h>
450
+ #include <ATen/ops/arcsin_ops.h>
451
+ #include <ATen/ops/arcsinh_ops.h>
452
+ #include <ATen/ops/arctan_ops.h>
453
+ #include <ATen/ops/arctan2_ops.h>
454
+ #include <ATen/ops/arctanh_ops.h>
455
+ #include <ATen/ops/argmax_ops.h>
456
+ #include <ATen/ops/argmin_ops.h>
457
+ #include <ATen/ops/argsort_ops.h>
458
+ #include <ATen/ops/argwhere_ops.h>
459
+ #include <ATen/ops/as_strided_ops.h>
460
+ #include <ATen/ops/as_strided_copy_ops.h>
461
+ #include <ATen/ops/as_strided_scatter_ops.h>
462
+ #include <ATen/ops/asin_ops.h>
463
+ #include <ATen/ops/asinh_ops.h>
464
+ #include <ATen/ops/atan_ops.h>
465
+ #include <ATen/ops/atan2_ops.h>
466
+ #include <ATen/ops/atanh_ops.h>
467
+ #include <ATen/ops/atleast_1d_ops.h>
468
+ #include <ATen/ops/atleast_2d_ops.h>
469
+ #include <ATen/ops/atleast_3d_ops.h>
470
+ #include <ATen/ops/avg_pool1d_ops.h>
471
+ #include <ATen/ops/avg_pool2d_ops.h>
472
+ #include <ATen/ops/avg_pool2d_backward_ops.h>
473
+ #include <ATen/ops/avg_pool3d_ops.h>
474
+ #include <ATen/ops/avg_pool3d_backward_ops.h>
475
+ #include <ATen/ops/baddbmm_ops.h>
476
+ #include <ATen/ops/bartlett_window_ops.h>
477
+ #include <ATen/ops/batch_norm_ops.h>
478
+ #include <ATen/ops/batch_norm_backward_ops.h>
479
+ #include <ATen/ops/batch_norm_backward_elemt_ops.h>
480
+ #include <ATen/ops/batch_norm_backward_reduce_ops.h>
481
+ #include <ATen/ops/batch_norm_elemt_ops.h>
482
+ #include <ATen/ops/batch_norm_gather_stats_ops.h>
483
+ #include <ATen/ops/batch_norm_gather_stats_with_counts_ops.h>
484
+ #include <ATen/ops/batch_norm_stats_ops.h>
485
+ #include <ATen/ops/batch_norm_update_stats_ops.h>
486
+ #include <ATen/ops/bernoulli_ops.h>
487
+ #include <ATen/ops/bilinear_ops.h>
488
+ #include <ATen/ops/binary_cross_entropy_ops.h>
489
+ #include <ATen/ops/binary_cross_entropy_backward_ops.h>
490
+ #include <ATen/ops/binary_cross_entropy_with_logits_ops.h>
491
+ #include <ATen/ops/bincount_ops.h>
492
+ #include <ATen/ops/binomial_ops.h>
493
+ #include <ATen/ops/bitwise_and_ops.h>
494
+ #include <ATen/ops/bitwise_left_shift_ops.h>
495
+ #include <ATen/ops/bitwise_not_ops.h>
496
+ #include <ATen/ops/bitwise_or_ops.h>
497
+ #include <ATen/ops/bitwise_right_shift_ops.h>
498
+ #include <ATen/ops/bitwise_xor_ops.h>
499
+ #include <ATen/ops/blackman_window_ops.h>
500
+ #include <ATen/ops/block_diag_ops.h>
501
+ #include <ATen/ops/bmm_ops.h>
502
+ #include <ATen/ops/broadcast_tensors_ops.h>
503
+ #include <ATen/ops/broadcast_to_ops.h>
504
+ #include <ATen/ops/bucketize_ops.h>
505
+ #include <ATen/ops/can_cast_ops.h>
506
+ #include <ATen/ops/cartesian_prod_ops.h>
507
+ #include <ATen/ops/cat_ops.h>
508
+ #include <ATen/ops/cauchy_ops.h>
509
+ #include <ATen/ops/ccol_indices_ops.h>
510
+ #include <ATen/ops/ccol_indices_copy_ops.h>
511
+ #include <ATen/ops/cdist_ops.h>
512
+ #include <ATen/ops/ceil_ops.h>
513
+ #include <ATen/ops/celu_ops.h>
514
+ #include <ATen/ops/chain_matmul_ops.h>
515
+ #include <ATen/ops/chalf_ops.h>
516
+ #include <ATen/ops/channel_shuffle_ops.h>
517
+ #include <ATen/ops/cholesky_ops.h>
518
+ #include <ATen/ops/cholesky_inverse_ops.h>
519
+ #include <ATen/ops/cholesky_solve_ops.h>
520
+ #include <ATen/ops/choose_qparams_optimized_ops.h>
521
+ #include <ATen/ops/chunk_ops.h>
522
+ #include <ATen/ops/clamp_ops.h>
523
+ #include <ATen/ops/clamp_max_ops.h>
524
+ #include <ATen/ops/clamp_min_ops.h>
525
+ #include <ATen/ops/clip_ops.h>
526
+ #include <ATen/ops/clone_ops.h>
527
+ #include <ATen/ops/coalesce_ops.h>
528
+ #include <ATen/ops/col2im_ops.h>
529
+ #include <ATen/ops/col_indices_ops.h>
530
+ #include <ATen/ops/col_indices_copy_ops.h>
531
+ #include <ATen/ops/column_stack_ops.h>
532
+ #include <ATen/ops/combinations_ops.h>
533
+ #include <ATen/ops/complex_ops.h>
534
+ #include <ATen/ops/concat_ops.h>
535
+ #include <ATen/ops/concatenate_ops.h>
536
+ #include <ATen/ops/conj_ops.h>
537
+ #include <ATen/ops/conj_physical_ops.h>
538
+ #include <ATen/ops/constant_pad_nd_ops.h>
539
+ #include <ATen/ops/contiguous_ops.h>
540
+ #include <ATen/ops/conv1d_ops.h>
541
+ #include <ATen/ops/conv2d_ops.h>
542
+ #include <ATen/ops/conv3d_ops.h>
543
+ #include <ATen/ops/conv_depthwise3d_ops.h>
544
+ #include <ATen/ops/conv_tbc_ops.h>
545
+ #include <ATen/ops/conv_tbc_backward_ops.h>
546
+ #include <ATen/ops/conv_transpose1d_ops.h>
547
+ #include <ATen/ops/conv_transpose2d_ops.h>
548
+ #include <ATen/ops/conv_transpose3d_ops.h>
549
+ #include <ATen/ops/convolution_ops.h>
550
+ #include <ATen/ops/convolution_backward_ops.h>
551
+ #include <ATen/ops/convolution_backward_overrideable_ops.h>
552
+ #include <ATen/ops/convolution_overrideable_ops.h>
553
+ #include <ATen/ops/copy_ops.h>
554
+ #include <ATen/ops/copy_sparse_to_sparse_ops.h>
555
+ #include <ATen/ops/copysign_ops.h>
556
+ #include <ATen/ops/corrcoef_ops.h>
557
+ #include <ATen/ops/cos_ops.h>
558
+ #include <ATen/ops/cosh_ops.h>
559
+ #include <ATen/ops/cosine_embedding_loss_ops.h>
560
+ #include <ATen/ops/cosine_similarity_ops.h>
561
+ #include <ATen/ops/count_nonzero_ops.h>
562
+ #include <ATen/ops/cov_ops.h>
563
+ #include <ATen/ops/cross_ops.h>
564
+ #include <ATen/ops/cross_entropy_loss_ops.h>
565
+ #include <ATen/ops/crow_indices_ops.h>
566
+ #include <ATen/ops/crow_indices_copy_ops.h>
567
+ #include <ATen/ops/ctc_loss_ops.h>
568
+ #include <ATen/ops/cudnn_affine_grid_generator_ops.h>
569
+ #include <ATen/ops/cudnn_affine_grid_generator_backward_ops.h>
570
+ #include <ATen/ops/cudnn_batch_norm_ops.h>
571
+ #include <ATen/ops/cudnn_batch_norm_backward_ops.h>
572
+ #include <ATen/ops/cudnn_convolution_ops.h>
573
+ #include <ATen/ops/cudnn_convolution_add_relu_ops.h>
574
+ #include <ATen/ops/cudnn_convolution_relu_ops.h>
575
+ #include <ATen/ops/cudnn_convolution_transpose_ops.h>
576
+ #include <ATen/ops/cudnn_grid_sampler_ops.h>
577
+ #include <ATen/ops/cudnn_grid_sampler_backward_ops.h>
578
+ #include <ATen/ops/cudnn_is_acceptable_ops.h>
579
+ #include <ATen/ops/cummax_ops.h>
580
+ #include <ATen/ops/cummaxmin_backward_ops.h>
581
+ #include <ATen/ops/cummin_ops.h>
582
+ #include <ATen/ops/cumprod_ops.h>
583
+ #include <ATen/ops/cumprod_backward_ops.h>
584
+ #include <ATen/ops/cumsum_ops.h>
585
+ #include <ATen/ops/cumulative_trapezoid_ops.h>
586
+ #include <ATen/ops/data_ops.h>
587
+ #include <ATen/ops/deg2rad_ops.h>
588
+ #include <ATen/ops/dense_dim_ops.h>
589
+ #include <ATen/ops/dequantize_ops.h>
590
+ #include <ATen/ops/det_ops.h>
591
+ #include <ATen/ops/detach_ops.h>
592
+ #include <ATen/ops/detach_copy_ops.h>
593
+ #include <ATen/ops/diag_ops.h>
594
+ #include <ATen/ops/diag_embed_ops.h>
595
+ #include <ATen/ops/diagflat_ops.h>
596
+ #include <ATen/ops/diagonal_ops.h>
597
+ #include <ATen/ops/diagonal_backward_ops.h>
598
+ #include <ATen/ops/diagonal_copy_ops.h>
599
+ #include <ATen/ops/diagonal_scatter_ops.h>
600
+ #include <ATen/ops/diff_ops.h>
601
+ #include <ATen/ops/digamma_ops.h>
602
+ #include <ATen/ops/dist_ops.h>
603
+ #include <ATen/ops/div_ops.h>
604
+ #include <ATen/ops/divide_ops.h>
605
+ #include <ATen/ops/dot_ops.h>
606
+ #include <ATen/ops/dropout_ops.h>
607
+ #include <ATen/ops/dsplit_ops.h>
608
+ #include <ATen/ops/dstack_ops.h>
609
+ #include <ATen/ops/einsum_ops.h>
610
+ #include <ATen/ops/elu_ops.h>
611
+ #include <ATen/ops/elu_backward_ops.h>
612
+ #include <ATen/ops/embedding_ops.h>
613
+ #include <ATen/ops/embedding_backward_ops.h>
614
+ #include <ATen/ops/embedding_bag_ops.h>
615
+ #include <ATen/ops/embedding_dense_backward_ops.h>
616
+ #include <ATen/ops/embedding_renorm_ops.h>
617
+ #include <ATen/ops/embedding_sparse_backward_ops.h>
618
+ #include <ATen/ops/empty_ops.h>
619
+ #include <ATen/ops/empty_like_ops.h>
620
+ #include <ATen/ops/empty_permuted_ops.h>
621
+ #include <ATen/ops/empty_quantized_ops.h>
622
+ #include <ATen/ops/empty_strided_ops.h>
623
+ #include <ATen/ops/eq_ops.h>
624
+ #include <ATen/ops/equal_ops.h>
625
+ #include <ATen/ops/erf_ops.h>
626
+ #include <ATen/ops/erfc_ops.h>
627
+ #include <ATen/ops/erfinv_ops.h>
628
+ #include <ATen/ops/exp_ops.h>
629
+ #include <ATen/ops/exp2_ops.h>
630
+ #include <ATen/ops/expand_ops.h>
631
+ #include <ATen/ops/expand_as_ops.h>
632
+ #include <ATen/ops/expand_copy_ops.h>
633
+ #include <ATen/ops/expm1_ops.h>
634
+ #include <ATen/ops/exponential_ops.h>
635
+ #include <ATen/ops/eye_ops.h>
636
+ #include <ATen/ops/fake_quantize_per_channel_affine_ops.h>
637
+ #include <ATen/ops/fake_quantize_per_channel_affine_cachemask_ops.h>
638
+ #include <ATen/ops/fake_quantize_per_channel_affine_cachemask_backward_ops.h>
639
+ #include <ATen/ops/fake_quantize_per_tensor_affine_ops.h>
640
+ #include <ATen/ops/fake_quantize_per_tensor_affine_cachemask_ops.h>
641
+ #include <ATen/ops/fake_quantize_per_tensor_affine_cachemask_backward_ops.h>
642
+ #include <ATen/ops/fbgemm_linear_fp16_weight_ops.h>
643
+ #include <ATen/ops/fbgemm_linear_fp16_weight_fp32_activation_ops.h>
644
+ #include <ATen/ops/fbgemm_linear_int8_weight_ops.h>
645
+ #include <ATen/ops/fbgemm_linear_int8_weight_fp32_activation_ops.h>
646
+ #include <ATen/ops/fbgemm_linear_quantize_weight_ops.h>
647
+ #include <ATen/ops/fbgemm_pack_gemm_matrix_fp16_ops.h>
648
+ #include <ATen/ops/fbgemm_pack_quantized_matrix_ops.h>
649
+ #include <ATen/ops/feature_alpha_dropout_ops.h>
650
+ #include <ATen/ops/feature_dropout_ops.h>
651
+ #include <ATen/ops/fft_fft_ops.h>
652
+ #include <ATen/ops/fft_fft2_ops.h>
653
+ #include <ATen/ops/fft_fftfreq_ops.h>
654
+ #include <ATen/ops/fft_fftn_ops.h>
655
+ #include <ATen/ops/fft_fftshift_ops.h>
656
+ #include <ATen/ops/fft_hfft_ops.h>
657
+ #include <ATen/ops/fft_hfft2_ops.h>
658
+ #include <ATen/ops/fft_hfftn_ops.h>
659
+ #include <ATen/ops/fft_ifft_ops.h>
660
+ #include <ATen/ops/fft_ifft2_ops.h>
661
+ #include <ATen/ops/fft_ifftn_ops.h>
662
+ #include <ATen/ops/fft_ifftshift_ops.h>
663
+ #include <ATen/ops/fft_ihfft_ops.h>
664
+ #include <ATen/ops/fft_ihfft2_ops.h>
665
+ #include <ATen/ops/fft_ihfftn_ops.h>
666
+ #include <ATen/ops/fft_irfft_ops.h>
667
+ #include <ATen/ops/fft_irfft2_ops.h>
668
+ #include <ATen/ops/fft_irfftn_ops.h>
669
+ #include <ATen/ops/fft_rfft_ops.h>
670
+ #include <ATen/ops/fft_rfft2_ops.h>
671
+ #include <ATen/ops/fft_rfftfreq_ops.h>
672
+ #include <ATen/ops/fft_rfftn_ops.h>
673
+ #include <ATen/ops/fill_ops.h>
674
+ #include <ATen/ops/fill_diagonal_ops.h>
675
+ #include <ATen/ops/fix_ops.h>
676
+ #include <ATen/ops/flatten_ops.h>
677
+ #include <ATen/ops/flatten_dense_tensors_ops.h>
678
+ #include <ATen/ops/flip_ops.h>
679
+ #include <ATen/ops/fliplr_ops.h>
680
+ #include <ATen/ops/flipud_ops.h>
681
+ #include <ATen/ops/float_power_ops.h>
682
+ #include <ATen/ops/floor_ops.h>
683
+ #include <ATen/ops/floor_divide_ops.h>
684
+ #include <ATen/ops/fmax_ops.h>
685
+ #include <ATen/ops/fmin_ops.h>
686
+ #include <ATen/ops/fmod_ops.h>
687
+ #include <ATen/ops/frac_ops.h>
688
+ #include <ATen/ops/fractional_max_pool2d_ops.h>
689
+ #include <ATen/ops/fractional_max_pool2d_backward_ops.h>
690
+ #include <ATen/ops/fractional_max_pool3d_ops.h>
691
+ #include <ATen/ops/fractional_max_pool3d_backward_ops.h>
692
+ #include <ATen/ops/frexp_ops.h>
693
+ #include <ATen/ops/frobenius_norm_ops.h>
694
+ #include <ATen/ops/from_file_ops.h>
695
+ #include <ATen/ops/full_ops.h>
696
+ #include <ATen/ops/full_like_ops.h>
697
+ #include <ATen/ops/fused_moving_avg_obs_fake_quant_ops.h>
698
+ #include <ATen/ops/gather_ops.h>
699
+ #include <ATen/ops/gather_backward_ops.h>
700
+ #include <ATen/ops/gcd_ops.h>
701
+ #include <ATen/ops/ge_ops.h>
702
+ #include <ATen/ops/gelu_ops.h>
703
+ #include <ATen/ops/gelu_backward_ops.h>
704
+ #include <ATen/ops/geometric_ops.h>
705
+ #include <ATen/ops/geqrf_ops.h>
706
+ #include <ATen/ops/ger_ops.h>
707
+ #include <ATen/ops/glu_ops.h>
708
+ #include <ATen/ops/glu_backward_ops.h>
709
+ #include <ATen/ops/glu_backward_jvp_ops.h>
710
+ #include <ATen/ops/glu_jvp_ops.h>
711
+ #include <ATen/ops/gradient_ops.h>
712
+ #include <ATen/ops/greater_ops.h>
713
+ #include <ATen/ops/greater_equal_ops.h>
714
+ #include <ATen/ops/grid_sampler_ops.h>
715
+ #include <ATen/ops/grid_sampler_2d_ops.h>
716
+ #include <ATen/ops/grid_sampler_2d_backward_ops.h>
717
+ #include <ATen/ops/grid_sampler_3d_ops.h>
718
+ #include <ATen/ops/grid_sampler_3d_backward_ops.h>
719
+ #include <ATen/ops/group_norm_ops.h>
720
+ #include <ATen/ops/gru_ops.h>
721
+ #include <ATen/ops/gru_cell_ops.h>
722
+ #include <ATen/ops/gt_ops.h>
723
+ #include <ATen/ops/hamming_window_ops.h>
724
+ #include <ATen/ops/hann_window_ops.h>
725
+ #include <ATen/ops/hardshrink_ops.h>
726
+ #include <ATen/ops/hardshrink_backward_ops.h>
727
+ #include <ATen/ops/hardsigmoid_ops.h>
728
+ #include <ATen/ops/hardsigmoid_backward_ops.h>
729
+ #include <ATen/ops/hardswish_ops.h>
730
+ #include <ATen/ops/hardswish_backward_ops.h>
731
+ #include <ATen/ops/hardtanh_ops.h>
732
+ #include <ATen/ops/hardtanh_backward_ops.h>
733
+ #include <ATen/ops/heaviside_ops.h>
734
+ #include <ATen/ops/hinge_embedding_loss_ops.h>
735
+ #include <ATen/ops/histc_ops.h>
736
+ #include <ATen/ops/histogram_ops.h>
737
+ #include <ATen/ops/histogramdd_ops.h>
738
+ #include <ATen/ops/hsplit_ops.h>
739
+ #include <ATen/ops/hspmm_ops.h>
740
+ #include <ATen/ops/hstack_ops.h>
741
+ #include <ATen/ops/huber_loss_ops.h>
742
+ #include <ATen/ops/huber_loss_backward_ops.h>
743
+ #include <ATen/ops/hypot_ops.h>
744
+ #include <ATen/ops/i0_ops.h>
745
+ #include <ATen/ops/igamma_ops.h>
746
+ #include <ATen/ops/igammac_ops.h>
747
+ #include <ATen/ops/im2col_ops.h>
748
+ #include <ATen/ops/imag_ops.h>
749
+ #include <ATen/ops/index_ops.h>
750
+ #include <ATen/ops/index_add_ops.h>
751
+ #include <ATen/ops/index_copy_ops.h>
752
+ #include <ATen/ops/index_fill_ops.h>
753
+ #include <ATen/ops/index_put_ops.h>
754
+ #include <ATen/ops/index_reduce_ops.h>
755
+ #include <ATen/ops/index_select_ops.h>
756
+ #include <ATen/ops/index_select_backward_ops.h>
757
+ #include <ATen/ops/indices_ops.h>
758
+ #include <ATen/ops/indices_copy_ops.h>
759
+ #include <ATen/ops/infinitely_differentiable_gelu_backward_ops.h>
760
+ #include <ATen/ops/inner_ops.h>
761
+ #include <ATen/ops/instance_norm_ops.h>
762
+ #include <ATen/ops/int_repr_ops.h>
763
+ #include <ATen/ops/inverse_ops.h>
764
+ #include <ATen/ops/is_coalesced_ops.h>
765
+ #include <ATen/ops/is_complex_ops.h>
766
+ #include <ATen/ops/is_conj_ops.h>
767
+ #include <ATen/ops/is_distributed_ops.h>
768
+ #include <ATen/ops/is_floating_point_ops.h>
769
+ #include <ATen/ops/is_inference_ops.h>
770
+ #include <ATen/ops/is_leaf_ops.h>
771
+ #include <ATen/ops/is_neg_ops.h>
772
+ #include <ATen/ops/is_nonzero_ops.h>
773
+ #include <ATen/ops/is_pinned_ops.h>
774
+ #include <ATen/ops/is_same_size_ops.h>
775
+ #include <ATen/ops/is_set_to_ops.h>
776
+ #include <ATen/ops/is_signed_ops.h>
777
+ #include <ATen/ops/is_vulkan_available_ops.h>
778
+ #include <ATen/ops/isclose_ops.h>
779
+ #include <ATen/ops/isfinite_ops.h>
780
+ #include <ATen/ops/isin_ops.h>
781
+ #include <ATen/ops/isinf_ops.h>
782
+ #include <ATen/ops/isnan_ops.h>
783
+ #include <ATen/ops/isneginf_ops.h>
784
+ #include <ATen/ops/isposinf_ops.h>
785
+ #include <ATen/ops/isreal_ops.h>
786
+ #include <ATen/ops/istft_ops.h>
787
+ #include <ATen/ops/item_ops.h>
788
+ #include <ATen/ops/kaiser_window_ops.h>
789
+ #include <ATen/ops/kl_div_ops.h>
790
+ #include <ATen/ops/kron_ops.h>
791
+ #include <ATen/ops/kthvalue_ops.h>
792
+ #include <ATen/ops/l1_loss_ops.h>
793
+ #include <ATen/ops/layer_norm_ops.h>
794
+ #include <ATen/ops/lcm_ops.h>
795
+ #include <ATen/ops/ldexp_ops.h>
796
+ #include <ATen/ops/le_ops.h>
797
+ #include <ATen/ops/leaky_relu_ops.h>
798
+ #include <ATen/ops/leaky_relu_backward_ops.h>
799
+ #include <ATen/ops/lerp_ops.h>
800
+ #include <ATen/ops/less_ops.h>
801
+ #include <ATen/ops/less_equal_ops.h>
802
+ #include <ATen/ops/lgamma_ops.h>
803
+ #include <ATen/ops/lift_ops.h>
804
+ #include <ATen/ops/lift_fresh_ops.h>
805
+ #include <ATen/ops/lift_fresh_copy_ops.h>
806
+ #include <ATen/ops/linalg_cholesky_ops.h>
807
+ #include <ATen/ops/linalg_cholesky_ex_ops.h>
808
+ #include <ATen/ops/linalg_cond_ops.h>
809
+ #include <ATen/ops/linalg_cross_ops.h>
810
+ #include <ATen/ops/linalg_det_ops.h>
811
+ #include <ATen/ops/linalg_diagonal_ops.h>
812
+ #include <ATen/ops/linalg_eig_ops.h>
813
+ #include <ATen/ops/linalg_eigh_ops.h>
814
+ #include <ATen/ops/linalg_eigvals_ops.h>
815
+ #include <ATen/ops/linalg_eigvalsh_ops.h>
816
+ #include <ATen/ops/linalg_householder_product_ops.h>
817
+ #include <ATen/ops/linalg_inv_ops.h>
818
+ #include <ATen/ops/linalg_inv_ex_ops.h>
819
+ #include <ATen/ops/linalg_ldl_factor_ops.h>
820
+ #include <ATen/ops/linalg_ldl_factor_ex_ops.h>
821
+ #include <ATen/ops/linalg_ldl_solve_ops.h>
822
+ #include <ATen/ops/linalg_lstsq_ops.h>
823
+ #include <ATen/ops/linalg_lu_ops.h>
824
+ #include <ATen/ops/linalg_lu_factor_ops.h>
825
+ #include <ATen/ops/linalg_lu_factor_ex_ops.h>
826
+ #include <ATen/ops/linalg_lu_solve_ops.h>
827
+ #include <ATen/ops/linalg_matmul_ops.h>
828
+ #include <ATen/ops/linalg_matrix_exp_ops.h>
829
+ #include <ATen/ops/linalg_matrix_norm_ops.h>
830
+ #include <ATen/ops/linalg_matrix_power_ops.h>
831
+ #include <ATen/ops/linalg_matrix_rank_ops.h>
832
+ #include <ATen/ops/linalg_multi_dot_ops.h>
833
+ #include <ATen/ops/linalg_norm_ops.h>
834
+ #include <ATen/ops/linalg_pinv_ops.h>
835
+ #include <ATen/ops/linalg_qr_ops.h>
836
+ #include <ATen/ops/linalg_slogdet_ops.h>
837
+ #include <ATen/ops/linalg_solve_ops.h>
838
+ #include <ATen/ops/linalg_solve_ex_ops.h>
839
+ #include <ATen/ops/linalg_solve_triangular_ops.h>
840
+ #include <ATen/ops/linalg_svd_ops.h>
841
+ #include <ATen/ops/linalg_svdvals_ops.h>
842
+ #include <ATen/ops/linalg_tensorinv_ops.h>
843
+ #include <ATen/ops/linalg_tensorsolve_ops.h>
844
+ #include <ATen/ops/linalg_vander_ops.h>
845
+ #include <ATen/ops/linalg_vecdot_ops.h>
846
+ #include <ATen/ops/linalg_vector_norm_ops.h>
847
+ #include <ATen/ops/linear_ops.h>
848
+ #include <ATen/ops/linear_backward_ops.h>
849
+ #include <ATen/ops/linspace_ops.h>
850
+ #include <ATen/ops/log_ops.h>
851
+ #include <ATen/ops/log10_ops.h>
852
+ #include <ATen/ops/log1p_ops.h>
853
+ #include <ATen/ops/log2_ops.h>
854
+ #include <ATen/ops/log_normal_ops.h>
855
+ #include <ATen/ops/log_sigmoid_ops.h>
856
+ #include <ATen/ops/log_sigmoid_backward_ops.h>
857
+ #include <ATen/ops/log_sigmoid_forward_ops.h>
858
+ #include <ATen/ops/log_softmax_ops.h>
859
+ #include <ATen/ops/logaddexp_ops.h>
860
+ #include <ATen/ops/logaddexp2_ops.h>
861
+ #include <ATen/ops/logcumsumexp_ops.h>
862
+ #include <ATen/ops/logdet_ops.h>
863
+ #include <ATen/ops/logical_and_ops.h>
864
+ #include <ATen/ops/logical_not_ops.h>
865
+ #include <ATen/ops/logical_or_ops.h>
866
+ #include <ATen/ops/logical_xor_ops.h>
867
+ #include <ATen/ops/logit_ops.h>
868
+ #include <ATen/ops/logit_backward_ops.h>
869
+ #include <ATen/ops/logspace_ops.h>
870
+ #include <ATen/ops/logsumexp_ops.h>
871
+ #include <ATen/ops/lshift_ops.h>
872
+ #include <ATen/ops/lstm_ops.h>
873
+ #include <ATen/ops/lstm_cell_ops.h>
874
+ #include <ATen/ops/lstm_mps_backward_ops.h>
875
+ #include <ATen/ops/lt_ops.h>
876
+ #include <ATen/ops/lu_solve_ops.h>
877
+ #include <ATen/ops/lu_unpack_ops.h>
878
+ #include <ATen/ops/mH_ops.h>
879
+ #include <ATen/ops/mT_ops.h>
880
+ #include <ATen/ops/margin_ranking_loss_ops.h>
881
+ #include <ATen/ops/masked_fill_ops.h>
882
+ #include <ATen/ops/masked_scatter_ops.h>
883
+ #include <ATen/ops/masked_scatter_backward_ops.h>
884
+ #include <ATen/ops/masked_select_ops.h>
885
+ #include <ATen/ops/masked_select_backward_ops.h>
886
+ #include <ATen/ops/matmul_ops.h>
887
+ #include <ATen/ops/matmul_backward_ops.h>
888
+ #include <ATen/ops/matrix_H_ops.h>
889
+ #include <ATen/ops/matrix_exp_ops.h>
890
+ #include <ATen/ops/matrix_exp_backward_ops.h>
891
+ #include <ATen/ops/matrix_power_ops.h>
892
+ #include <ATen/ops/max_ops.h>
893
+ #include <ATen/ops/max_pool1d_ops.h>
894
+ #include <ATen/ops/max_pool1d_with_indices_ops.h>
895
+ #include <ATen/ops/max_pool2d_ops.h>
896
+ #include <ATen/ops/max_pool2d_backward_ops.h>
897
+ #include <ATen/ops/max_pool2d_with_indices_ops.h>
898
+ #include <ATen/ops/max_pool2d_with_indices_backward_ops.h>
899
+ #include <ATen/ops/max_pool3d_ops.h>
900
+ #include <ATen/ops/max_pool3d_with_indices_ops.h>
901
+ #include <ATen/ops/max_pool3d_with_indices_backward_ops.h>
902
+ #include <ATen/ops/max_unpool2d_ops.h>
903
+ #include <ATen/ops/max_unpool3d_ops.h>
904
+ #include <ATen/ops/maximum_ops.h>
905
+ #include <ATen/ops/mean_ops.h>
906
+ #include <ATen/ops/median_ops.h>
907
+ #include <ATen/ops/meshgrid_ops.h>
908
+ #include <ATen/ops/min_ops.h>
909
+ #include <ATen/ops/minimum_ops.h>
910
+ #include <ATen/ops/miopen_batch_norm_ops.h>
911
+ #include <ATen/ops/miopen_batch_norm_backward_ops.h>
912
+ #include <ATen/ops/miopen_convolution_ops.h>
913
+ #include <ATen/ops/miopen_convolution_add_relu_ops.h>
914
+ #include <ATen/ops/miopen_convolution_relu_ops.h>
915
+ #include <ATen/ops/miopen_convolution_transpose_ops.h>
916
+ #include <ATen/ops/miopen_depthwise_convolution_ops.h>
917
+ #include <ATen/ops/miopen_rnn_ops.h>
918
+ #include <ATen/ops/miopen_rnn_backward_ops.h>
919
+ #include <ATen/ops/mish_ops.h>
920
+ #include <ATen/ops/mish_backward_ops.h>
921
+ #include <ATen/ops/mkldnn_adaptive_avg_pool2d_ops.h>
922
+ #include <ATen/ops/mkldnn_adaptive_avg_pool2d_backward_ops.h>
923
+ #include <ATen/ops/mkldnn_convolution_ops.h>
924
+ #include <ATen/ops/mkldnn_linear_ops.h>
925
+ #include <ATen/ops/mkldnn_linear_backward_ops.h>
926
+ #include <ATen/ops/mkldnn_linear_backward_input_ops.h>
927
+ #include <ATen/ops/mkldnn_linear_backward_weights_ops.h>
928
+ #include <ATen/ops/mkldnn_max_pool2d_ops.h>
929
+ #include <ATen/ops/mkldnn_max_pool2d_backward_ops.h>
930
+ #include <ATen/ops/mkldnn_max_pool3d_ops.h>
931
+ #include <ATen/ops/mkldnn_max_pool3d_backward_ops.h>
932
+ #include <ATen/ops/mkldnn_reorder_conv2d_weight_ops.h>
933
+ #include <ATen/ops/mkldnn_reorder_conv3d_weight_ops.h>
934
+ #include <ATen/ops/mkldnn_rnn_layer_ops.h>
935
+ #include <ATen/ops/mkldnn_rnn_layer_backward_ops.h>
936
+ #include <ATen/ops/mm_ops.h>
937
+ #include <ATen/ops/mode_ops.h>
938
+ #include <ATen/ops/moveaxis_ops.h>
939
+ #include <ATen/ops/movedim_ops.h>
940
+ #include <ATen/ops/mps_convolution_backward_ops.h>
941
+ #include <ATen/ops/mps_convolution_transpose_backward_ops.h>
942
+ #include <ATen/ops/mse_loss_ops.h>
943
+ #include <ATen/ops/mse_loss_backward_ops.h>
944
+ #include <ATen/ops/msort_ops.h>
945
+ #include <ATen/ops/mul_ops.h>
946
+ #include <ATen/ops/multi_margin_loss_ops.h>
947
+ #include <ATen/ops/multi_margin_loss_backward_ops.h>
948
+ #include <ATen/ops/multilabel_margin_loss_ops.h>
949
+ #include <ATen/ops/multilabel_margin_loss_backward_ops.h>
950
+ #include <ATen/ops/multilabel_margin_loss_forward_ops.h>
951
+ #include <ATen/ops/multinomial_ops.h>
952
+ #include <ATen/ops/multiply_ops.h>
953
+ #include <ATen/ops/mv_ops.h>
954
+ #include <ATen/ops/mvlgamma_ops.h>
955
+ #include <ATen/ops/nan_to_num_ops.h>
956
+ #include <ATen/ops/nanmean_ops.h>
957
+ #include <ATen/ops/nanmedian_ops.h>
958
+ #include <ATen/ops/nanquantile_ops.h>
959
+ #include <ATen/ops/nansum_ops.h>
960
+ #include <ATen/ops/narrow_ops.h>
961
+ #include <ATen/ops/narrow_copy_ops.h>
962
+ #include <ATen/ops/native_batch_norm_ops.h>
963
+ #include <ATen/ops/native_batch_norm_backward_ops.h>
964
+ #include <ATen/ops/native_channel_shuffle_ops.h>
965
+ #include <ATen/ops/native_dropout_ops.h>
966
+ #include <ATen/ops/native_dropout_backward_ops.h>
967
+ #include <ATen/ops/native_group_norm_ops.h>
968
+ #include <ATen/ops/native_group_norm_backward_ops.h>
969
+ #include <ATen/ops/native_layer_norm_ops.h>
970
+ #include <ATen/ops/native_layer_norm_backward_ops.h>
971
+ #include <ATen/ops/native_norm_ops.h>
972
+ #include <ATen/ops/ne_ops.h>
973
+ #include <ATen/ops/neg_ops.h>
974
+ #include <ATen/ops/negative_ops.h>
975
+ #include <ATen/ops/nested_to_padded_tensor_ops.h>
976
+ #include <ATen/ops/new_empty_ops.h>
977
+ #include <ATen/ops/new_empty_strided_ops.h>
978
+ #include <ATen/ops/new_full_ops.h>
979
+ #include <ATen/ops/new_ones_ops.h>
980
+ #include <ATen/ops/new_zeros_ops.h>
981
+ #include <ATen/ops/nextafter_ops.h>
982
+ #include <ATen/ops/nll_loss_ops.h>
983
+ #include <ATen/ops/nll_loss2d_ops.h>
984
+ #include <ATen/ops/nll_loss2d_backward_ops.h>
985
+ #include <ATen/ops/nll_loss2d_forward_ops.h>
986
+ #include <ATen/ops/nll_loss_backward_ops.h>
987
+ #include <ATen/ops/nll_loss_forward_ops.h>
988
+ #include <ATen/ops/nll_loss_nd_ops.h>
989
+ #include <ATen/ops/nonzero_ops.h>
990
+ #include <ATen/ops/nonzero_numpy_ops.h>
991
+ #include <ATen/ops/nonzero_static_ops.h>
992
+ #include <ATen/ops/norm_ops.h>
993
+ #include <ATen/ops/norm_except_dim_ops.h>
994
+ #include <ATen/ops/normal_ops.h>
995
+ #include <ATen/ops/not_equal_ops.h>
996
+ #include <ATen/ops/nuclear_norm_ops.h>
997
+ #include <ATen/ops/numpy_T_ops.h>
998
+ #include <ATen/ops/one_hot_ops.h>
999
+ #include <ATen/ops/ones_ops.h>
1000
+ #include <ATen/ops/ones_like_ops.h>
1001
+ #include <ATen/ops/or_ops.h>
1002
+ #include <ATen/ops/orgqr_ops.h>
1003
+ #include <ATen/ops/ormqr_ops.h>
1004
+ #include <ATen/ops/outer_ops.h>
1005
+ #include <ATen/ops/output_nr_ops.h>
1006
+ #include <ATen/ops/pad_ops.h>
1007
+ #include <ATen/ops/pad_sequence_ops.h>
1008
+ #include <ATen/ops/pairwise_distance_ops.h>
1009
+ #include <ATen/ops/pdist_ops.h>
1010
+ #include <ATen/ops/permute_ops.h>
1011
+ #include <ATen/ops/permute_copy_ops.h>
1012
+ #include <ATen/ops/pin_memory_ops.h>
1013
+ #include <ATen/ops/pinverse_ops.h>
1014
+ #include <ATen/ops/pixel_shuffle_ops.h>
1015
+ #include <ATen/ops/pixel_unshuffle_ops.h>
1016
+ #include <ATen/ops/poisson_ops.h>
1017
+ #include <ATen/ops/poisson_nll_loss_ops.h>
1018
+ #include <ATen/ops/polar_ops.h>
1019
+ #include <ATen/ops/polygamma_ops.h>
1020
+ #include <ATen/ops/positive_ops.h>
1021
+ #include <ATen/ops/pow_ops.h>
1022
+ #include <ATen/ops/prelu_ops.h>
1023
+ #include <ATen/ops/prod_ops.h>
1024
+ #include <ATen/ops/promote_types_ops.h>
1025
+ #include <ATen/ops/put_ops.h>
1026
+ #include <ATen/ops/q_per_channel_axis_ops.h>
1027
+ #include <ATen/ops/q_per_channel_scales_ops.h>
1028
+ #include <ATen/ops/q_per_channel_zero_points_ops.h>
1029
+ #include <ATen/ops/q_scale_ops.h>
1030
+ #include <ATen/ops/q_zero_point_ops.h>
1031
+ #include <ATen/ops/qr_ops.h>
1032
+ #include <ATen/ops/qscheme_ops.h>
1033
+ #include <ATen/ops/quantile_ops.h>
1034
+ #include <ATen/ops/quantize_per_channel_ops.h>
1035
+ #include <ATen/ops/quantize_per_tensor_ops.h>
1036
+ #include <ATen/ops/quantize_per_tensor_dynamic_ops.h>
1037
+ #include <ATen/ops/quantized_batch_norm_ops.h>
1038
+ #include <ATen/ops/quantized_gru_cell_ops.h>
1039
+ #include <ATen/ops/quantized_lstm_cell_ops.h>
1040
+ #include <ATen/ops/quantized_max_pool1d_ops.h>
1041
+ #include <ATen/ops/quantized_max_pool2d_ops.h>
1042
+ #include <ATen/ops/quantized_max_pool3d_ops.h>
1043
+ #include <ATen/ops/quantized_rnn_relu_cell_ops.h>
1044
+ #include <ATen/ops/quantized_rnn_tanh_cell_ops.h>
1045
+ #include <ATen/ops/rad2deg_ops.h>
1046
+ #include <ATen/ops/rand_ops.h>
1047
+ #include <ATen/ops/rand_like_ops.h>
1048
+ #include <ATen/ops/randint_ops.h>
1049
+ #include <ATen/ops/randint_like_ops.h>
1050
+ #include <ATen/ops/randn_ops.h>
1051
+ #include <ATen/ops/randn_like_ops.h>
1052
+ #include <ATen/ops/random_ops.h>
1053
+ #include <ATen/ops/randperm_ops.h>
1054
+ #include <ATen/ops/range_ops.h>
1055
+ #include <ATen/ops/ravel_ops.h>
1056
+ #include <ATen/ops/real_ops.h>
1057
+ #include <ATen/ops/reciprocal_ops.h>
1058
+ #include <ATen/ops/record_stream_ops.h>
1059
+ #include <ATen/ops/refine_names_ops.h>
1060
+ #include <ATen/ops/reflection_pad1d_ops.h>
1061
+ #include <ATen/ops/reflection_pad1d_backward_ops.h>
1062
+ #include <ATen/ops/reflection_pad2d_ops.h>
1063
+ #include <ATen/ops/reflection_pad2d_backward_ops.h>
1064
+ #include <ATen/ops/reflection_pad3d_ops.h>
1065
+ #include <ATen/ops/reflection_pad3d_backward_ops.h>
1066
+ #include <ATen/ops/relu_ops.h>
1067
+ #include <ATen/ops/relu6_ops.h>
1068
+ #include <ATen/ops/remainder_ops.h>
1069
+ #include <ATen/ops/rename_ops.h>
1070
+ #include <ATen/ops/renorm_ops.h>
1071
+ #include <ATen/ops/repeat_ops.h>
1072
+ #include <ATen/ops/repeat_interleave_ops.h>
1073
+ #include <ATen/ops/replication_pad1d_ops.h>
1074
+ #include <ATen/ops/replication_pad1d_backward_ops.h>
1075
+ #include <ATen/ops/replication_pad2d_ops.h>
1076
+ #include <ATen/ops/replication_pad2d_backward_ops.h>
1077
+ #include <ATen/ops/replication_pad3d_ops.h>
1078
+ #include <ATen/ops/replication_pad3d_backward_ops.h>
1079
+ #include <ATen/ops/requires_grad_ops.h>
1080
+ #include <ATen/ops/reshape_ops.h>
1081
+ #include <ATen/ops/reshape_as_ops.h>
1082
+ #include <ATen/ops/resize_ops.h>
1083
+ #include <ATen/ops/resize_as_ops.h>
1084
+ #include <ATen/ops/resize_as_sparse_ops.h>
1085
+ #include <ATen/ops/resolve_conj_ops.h>
1086
+ #include <ATen/ops/resolve_neg_ops.h>
1087
+ #include <ATen/ops/result_type_ops.h>
1088
+ #include <ATen/ops/retain_grad_ops.h>
1089
+ #include <ATen/ops/retains_grad_ops.h>
1090
+ #include <ATen/ops/rms_norm_ops.h>
1091
+ #include <ATen/ops/rnn_relu_ops.h>
1092
+ #include <ATen/ops/rnn_relu_cell_ops.h>
1093
+ #include <ATen/ops/rnn_tanh_ops.h>
1094
+ #include <ATen/ops/rnn_tanh_cell_ops.h>
1095
+ #include <ATen/ops/roll_ops.h>
1096
+ #include <ATen/ops/rot90_ops.h>
1097
+ #include <ATen/ops/round_ops.h>
1098
+ #include <ATen/ops/row_indices_ops.h>
1099
+ #include <ATen/ops/row_indices_copy_ops.h>
1100
+ #include <ATen/ops/row_stack_ops.h>
1101
+ #include <ATen/ops/rrelu_ops.h>
1102
+ #include <ATen/ops/rrelu_with_noise_ops.h>
1103
+ #include <ATen/ops/rrelu_with_noise_backward_ops.h>
1104
+ #include <ATen/ops/rshift_ops.h>
1105
+ #include <ATen/ops/rsqrt_ops.h>
1106
+ #include <ATen/ops/rsub_ops.h>
1107
+ #include <ATen/ops/scalar_tensor_ops.h>
1108
+ #include <ATen/ops/scaled_dot_product_attention_ops.h>
1109
+ #include <ATen/ops/scatter_ops.h>
1110
+ #include <ATen/ops/scatter_add_ops.h>
1111
+ #include <ATen/ops/scatter_reduce_ops.h>
1112
+ #include <ATen/ops/searchsorted_ops.h>
1113
+ #include <ATen/ops/segment_reduce_ops.h>
1114
+ #include <ATen/ops/select_ops.h>
1115
+ #include <ATen/ops/select_backward_ops.h>
1116
+ #include <ATen/ops/select_copy_ops.h>
1117
+ #include <ATen/ops/select_scatter_ops.h>
1118
+ #include <ATen/ops/selu_ops.h>
1119
+ #include <ATen/ops/set_ops.h>
1120
+ #include <ATen/ops/set_data_ops.h>
1121
+ #include <ATen/ops/sgn_ops.h>
1122
+ #include <ATen/ops/sigmoid_ops.h>
1123
+ #include <ATen/ops/sigmoid_backward_ops.h>
1124
+ #include <ATen/ops/sign_ops.h>
1125
+ #include <ATen/ops/signbit_ops.h>
1126
+ #include <ATen/ops/silu_ops.h>
1127
+ #include <ATen/ops/silu_backward_ops.h>
1128
+ #include <ATen/ops/sin_ops.h>
1129
+ #include <ATen/ops/sinc_ops.h>
1130
+ #include <ATen/ops/sinh_ops.h>
1131
+ #include <ATen/ops/size_ops.h>
1132
+ #include <ATen/ops/slice_ops.h>
1133
+ #include <ATen/ops/slice_backward_ops.h>
1134
+ #include <ATen/ops/slice_copy_ops.h>
1135
+ #include <ATen/ops/slice_inverse_ops.h>
1136
+ #include <ATen/ops/slice_scatter_ops.h>
1137
+ #include <ATen/ops/slogdet_ops.h>
1138
+ #include <ATen/ops/slow_conv3d_ops.h>
1139
+ #include <ATen/ops/slow_conv3d_forward_ops.h>
1140
+ #include <ATen/ops/slow_conv_dilated2d_ops.h>
1141
+ #include <ATen/ops/slow_conv_dilated3d_ops.h>
1142
+ #include <ATen/ops/slow_conv_transpose2d_ops.h>
1143
+ #include <ATen/ops/slow_conv_transpose3d_ops.h>
1144
+ #include <ATen/ops/smm_ops.h>
1145
+ #include <ATen/ops/smooth_l1_loss_ops.h>
1146
+ #include <ATen/ops/smooth_l1_loss_backward_ops.h>
1147
+ #include <ATen/ops/soft_margin_loss_ops.h>
1148
+ #include <ATen/ops/soft_margin_loss_backward_ops.h>
1149
+ #include <ATen/ops/softmax_ops.h>
1150
+ #include <ATen/ops/softplus_ops.h>
1151
+ #include <ATen/ops/softplus_backward_ops.h>
1152
+ #include <ATen/ops/softshrink_ops.h>
1153
+ #include <ATen/ops/softshrink_backward_ops.h>
1154
+ #include <ATen/ops/sort_ops.h>
1155
+ #include <ATen/ops/sparse_bsc_tensor_ops.h>
1156
+ #include <ATen/ops/sparse_bsr_tensor_ops.h>
1157
+ #include <ATen/ops/sparse_compressed_tensor_ops.h>
1158
+ #include <ATen/ops/sparse_coo_tensor_ops.h>
1159
+ #include <ATen/ops/sparse_csc_tensor_ops.h>
1160
+ #include <ATen/ops/sparse_csr_tensor_ops.h>
1161
+ #include <ATen/ops/sparse_dim_ops.h>
1162
+ #include <ATen/ops/sparse_mask_ops.h>
1163
+ #include <ATen/ops/sparse_resize_ops.h>
1164
+ #include <ATen/ops/sparse_resize_and_clear_ops.h>
1165
+ #include <ATen/ops/sparse_sampled_addmm_ops.h>
1166
+ #include <ATen/ops/special_airy_ai_ops.h>
1167
+ #include <ATen/ops/special_bessel_j0_ops.h>
1168
+ #include <ATen/ops/special_bessel_j1_ops.h>
1169
+ #include <ATen/ops/special_bessel_y0_ops.h>
1170
+ #include <ATen/ops/special_bessel_y1_ops.h>
1171
+ #include <ATen/ops/special_chebyshev_polynomial_t_ops.h>
1172
+ #include <ATen/ops/special_chebyshev_polynomial_u_ops.h>
1173
+ #include <ATen/ops/special_chebyshev_polynomial_v_ops.h>
1174
+ #include <ATen/ops/special_chebyshev_polynomial_w_ops.h>
1175
+ #include <ATen/ops/special_digamma_ops.h>
1176
+ #include <ATen/ops/special_entr_ops.h>
1177
+ #include <ATen/ops/special_erf_ops.h>
1178
+ #include <ATen/ops/special_erfc_ops.h>
1179
+ #include <ATen/ops/special_erfcx_ops.h>
1180
+ #include <ATen/ops/special_erfinv_ops.h>
1181
+ #include <ATen/ops/special_exp2_ops.h>
1182
+ #include <ATen/ops/special_expit_ops.h>
1183
+ #include <ATen/ops/special_expm1_ops.h>
1184
+ #include <ATen/ops/special_gammainc_ops.h>
1185
+ #include <ATen/ops/special_gammaincc_ops.h>
1186
+ #include <ATen/ops/special_gammaln_ops.h>
1187
+ #include <ATen/ops/special_hermite_polynomial_h_ops.h>
1188
+ #include <ATen/ops/special_hermite_polynomial_he_ops.h>
1189
+ #include <ATen/ops/special_i0_ops.h>
1190
+ #include <ATen/ops/special_i0e_ops.h>
1191
+ #include <ATen/ops/special_i1_ops.h>
1192
+ #include <ATen/ops/special_i1e_ops.h>
1193
+ #include <ATen/ops/special_laguerre_polynomial_l_ops.h>
1194
+ #include <ATen/ops/special_legendre_polynomial_p_ops.h>
1195
+ #include <ATen/ops/special_log1p_ops.h>
1196
+ #include <ATen/ops/special_log_ndtr_ops.h>
1197
+ #include <ATen/ops/special_log_softmax_ops.h>
1198
+ #include <ATen/ops/special_logit_ops.h>
1199
+ #include <ATen/ops/special_logsumexp_ops.h>
1200
+ #include <ATen/ops/special_modified_bessel_i0_ops.h>
1201
+ #include <ATen/ops/special_modified_bessel_i1_ops.h>
1202
+ #include <ATen/ops/special_modified_bessel_k0_ops.h>
1203
+ #include <ATen/ops/special_modified_bessel_k1_ops.h>
1204
+ #include <ATen/ops/special_multigammaln_ops.h>
1205
+ #include <ATen/ops/special_ndtr_ops.h>
1206
+ #include <ATen/ops/special_ndtri_ops.h>
1207
+ #include <ATen/ops/special_polygamma_ops.h>
1208
+ #include <ATen/ops/special_psi_ops.h>
1209
+ #include <ATen/ops/special_round_ops.h>
1210
+ #include <ATen/ops/special_scaled_modified_bessel_k0_ops.h>
1211
+ #include <ATen/ops/special_scaled_modified_bessel_k1_ops.h>
1212
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_t_ops.h>
1213
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_u_ops.h>
1214
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_v_ops.h>
1215
+ #include <ATen/ops/special_shifted_chebyshev_polynomial_w_ops.h>
1216
+ #include <ATen/ops/special_sinc_ops.h>
1217
+ #include <ATen/ops/special_softmax_ops.h>
1218
+ #include <ATen/ops/special_spherical_bessel_j0_ops.h>
1219
+ #include <ATen/ops/special_xlog1py_ops.h>
1220
+ #include <ATen/ops/special_xlogy_ops.h>
1221
+ #include <ATen/ops/special_zeta_ops.h>
1222
+ #include <ATen/ops/split_ops.h>
1223
+ #include <ATen/ops/split_copy_ops.h>
1224
+ #include <ATen/ops/split_with_sizes_ops.h>
1225
+ #include <ATen/ops/split_with_sizes_copy_ops.h>
1226
+ #include <ATen/ops/sqrt_ops.h>
1227
+ #include <ATen/ops/square_ops.h>
1228
+ #include <ATen/ops/squeeze_ops.h>
1229
+ #include <ATen/ops/squeeze_copy_ops.h>
1230
+ #include <ATen/ops/sspaddmm_ops.h>
1231
+ #include <ATen/ops/stack_ops.h>
1232
+ #include <ATen/ops/std_ops.h>
1233
+ #include <ATen/ops/std_mean_ops.h>
1234
+ #include <ATen/ops/stft_ops.h>
1235
+ #include <ATen/ops/stride_ops.h>
1236
+ #include <ATen/ops/sub_ops.h>
1237
+ #include <ATen/ops/subtract_ops.h>
1238
+ #include <ATen/ops/sum_ops.h>
1239
+ #include <ATen/ops/sum_to_size_ops.h>
1240
+ #include <ATen/ops/svd_ops.h>
1241
+ #include <ATen/ops/swapaxes_ops.h>
1242
+ #include <ATen/ops/swapdims_ops.h>
1243
+ #include <ATen/ops/sym_constrain_range_ops.h>
1244
+ #include <ATen/ops/sym_constrain_range_for_size_ops.h>
1245
+ #include <ATen/ops/sym_numel_ops.h>
1246
+ #include <ATen/ops/sym_size_ops.h>
1247
+ #include <ATen/ops/sym_storage_offset_ops.h>
1248
+ #include <ATen/ops/sym_stride_ops.h>
1249
+ #include <ATen/ops/t_ops.h>
1250
+ #include <ATen/ops/t_copy_ops.h>
1251
+ #include <ATen/ops/take_ops.h>
1252
+ #include <ATen/ops/take_along_dim_ops.h>
1253
+ #include <ATen/ops/tan_ops.h>
1254
+ #include <ATen/ops/tanh_ops.h>
1255
+ #include <ATen/ops/tanh_backward_ops.h>
1256
+ #include <ATen/ops/tensor_split_ops.h>
1257
+ #include <ATen/ops/tensordot_ops.h>
1258
+ #include <ATen/ops/thnn_conv2d_ops.h>
1259
+ #include <ATen/ops/threshold_ops.h>
1260
+ #include <ATen/ops/threshold_backward_ops.h>
1261
+ #include <ATen/ops/tile_ops.h>
1262
+ #include <ATen/ops/to_ops.h>
1263
+ #include <ATen/ops/to_dense_ops.h>
1264
+ #include <ATen/ops/to_dense_backward_ops.h>
1265
+ #include <ATen/ops/to_mkldnn_ops.h>
1266
+ #include <ATen/ops/to_mkldnn_backward_ops.h>
1267
+ #include <ATen/ops/to_padded_tensor_ops.h>
1268
+ #include <ATen/ops/to_sparse_ops.h>
1269
+ #include <ATen/ops/to_sparse_bsc_ops.h>
1270
+ #include <ATen/ops/to_sparse_bsr_ops.h>
1271
+ #include <ATen/ops/to_sparse_csc_ops.h>
1272
+ #include <ATen/ops/to_sparse_csr_ops.h>
1273
+ #include <ATen/ops/topk_ops.h>
1274
+ #include <ATen/ops/trace_ops.h>
1275
+ #include <ATen/ops/trace_backward_ops.h>
1276
+ #include <ATen/ops/transpose_ops.h>
1277
+ #include <ATen/ops/transpose_copy_ops.h>
1278
+ #include <ATen/ops/trapezoid_ops.h>
1279
+ #include <ATen/ops/trapz_ops.h>
1280
+ #include <ATen/ops/triangular_solve_ops.h>
1281
+ #include <ATen/ops/tril_ops.h>
1282
+ #include <ATen/ops/tril_indices_ops.h>
1283
+ #include <ATen/ops/triplet_margin_loss_ops.h>
1284
+ #include <ATen/ops/triu_ops.h>
1285
+ #include <ATen/ops/triu_indices_ops.h>
1286
+ #include <ATen/ops/true_divide_ops.h>
1287
+ #include <ATen/ops/trunc_ops.h>
1288
+ #include <ATen/ops/type_as_ops.h>
1289
+ #include <ATen/ops/unbind_ops.h>
1290
+ #include <ATen/ops/unbind_copy_ops.h>
1291
+ #include <ATen/ops/unflatten_ops.h>
1292
+ #include <ATen/ops/unflatten_dense_tensors_ops.h>
1293
+ #include <ATen/ops/unfold_ops.h>
1294
+ #include <ATen/ops/unfold_backward_ops.h>
1295
+ #include <ATen/ops/unfold_copy_ops.h>
1296
+ #include <ATen/ops/uniform_ops.h>
1297
+ #include <ATen/ops/unique_consecutive_ops.h>
1298
+ #include <ATen/ops/unique_dim_ops.h>
1299
+ #include <ATen/ops/unique_dim_consecutive_ops.h>
1300
+ #include <ATen/ops/unsafe_chunk_ops.h>
1301
+ #include <ATen/ops/unsafe_split_ops.h>
1302
+ #include <ATen/ops/unsafe_split_with_sizes_ops.h>
1303
+ #include <ATen/ops/unsqueeze_ops.h>
1304
+ #include <ATen/ops/unsqueeze_copy_ops.h>
1305
+ #include <ATen/ops/upsample_bicubic2d_ops.h>
1306
+ #include <ATen/ops/upsample_bicubic2d_backward_ops.h>
1307
+ #include <ATen/ops/upsample_bilinear2d_ops.h>
1308
+ #include <ATen/ops/upsample_bilinear2d_backward_ops.h>
1309
+ #include <ATen/ops/upsample_linear1d_ops.h>
1310
+ #include <ATen/ops/upsample_linear1d_backward_ops.h>
1311
+ #include <ATen/ops/upsample_nearest1d_ops.h>
1312
+ #include <ATen/ops/upsample_nearest1d_backward_ops.h>
1313
+ #include <ATen/ops/upsample_nearest2d_ops.h>
1314
+ #include <ATen/ops/upsample_nearest2d_backward_ops.h>
1315
+ #include <ATen/ops/upsample_nearest3d_ops.h>
1316
+ #include <ATen/ops/upsample_nearest3d_backward_ops.h>
1317
+ #include <ATen/ops/upsample_trilinear3d_ops.h>
1318
+ #include <ATen/ops/upsample_trilinear3d_backward_ops.h>
1319
+ #include <ATen/ops/value_selecting_reduction_backward_ops.h>
1320
+ #include <ATen/ops/values_ops.h>
1321
+ #include <ATen/ops/values_copy_ops.h>
1322
+ #include <ATen/ops/vander_ops.h>
1323
+ #include <ATen/ops/var_ops.h>
1324
+ #include <ATen/ops/var_mean_ops.h>
1325
+ #include <ATen/ops/vdot_ops.h>
1326
+ #include <ATen/ops/view_ops.h>
1327
+ #include <ATen/ops/view_as_ops.h>
1328
+ #include <ATen/ops/view_as_complex_ops.h>
1329
+ #include <ATen/ops/view_as_complex_copy_ops.h>
1330
+ #include <ATen/ops/view_as_real_ops.h>
1331
+ #include <ATen/ops/view_as_real_copy_ops.h>
1332
+ #include <ATen/ops/view_copy_ops.h>
1333
+ #include <ATen/ops/vsplit_ops.h>
1334
+ #include <ATen/ops/vstack_ops.h>
1335
+ #include <ATen/ops/where_ops.h>
1336
+ #include <ATen/ops/xlogy_ops.h>
1337
+ #include <ATen/ops/xor_ops.h>
1338
+ #include <ATen/ops/zero_ops.h>
1339
+ #include <ATen/ops/zeros_ops.h>
1340
+ #include <ATen/ops/zeros_like_ops.h>
1341
+
1342
+ // Extension writers: do you write wrapper functions? Are you frustrated with
1343
+ // resolving overloads of operators? Are you frustrated with dealing with
1344
+ // pointer-to-methods and resolving overloads of pointer-to-methods?? Look no
1345
+ // further, this is the utility for you.
1346
+ //
1347
+ // Given an operator schema: aten::op.overload(...
1348
+ //
1349
+ // Use ATEN_FN2(op, overload) to get a *function* version of the operator
1350
+ // that is guaranteed to not be overloaded. This means that you can safely
1351
+ // decltype(&ATEN_FN2(op, overload)) it. NB: the 2 means this macro takes 2 args.
1352
+ //
1353
+ // Given an operator schema without an overload name: aten::op(...
1354
+ //
1355
+ // Use ATEN_FN(op) to get an unambiguous *function* version of the operator.
1356
+ //
1357
+ // There is some interesting behavior for out= operations.
1358
+ // ATEN_FN2(sin, out) gives a function that is *faithful* to the schema;
1359
+ // that is, the order of arguments is exactly what it looks like in the schema.
1360
+
1361
+ #define ATEN_FN2(op_name, overload) at::_ops::op_name##_##overload::call
1362
+ #define ATEN_FN(op_name) at::_ops::op_name::call
1363
+
1364
+ // Separately, ATEN_OP(op) and ATEN_OP2(op, overload) define a class containing compile-time
1365
+ // metadata about a given aten operator.
1366
+ // Notable data on the class includes:
1367
+ // - ATEN_OP2(add, Tensor)::name // returns the string name: "add"
1368
+ // - ATEN_OP2(add, Tensor)::overload_name // returns the string overload name: "Tensor"
1369
+ // - ATEN_OP2(add, Tensor)::schema // returns the C++ schema type: at::Tensor (const at::Tensor &, const at::Tensor &, const at::Scalar &)
1370
+ // - ATEN_OP2(add, Tensor)::schema_str // returns the string jit type: "add.Tensor(Tensor self, Tensor other, *, Scalar alpha=1) -> Tensor"
1371
+
1372
+ #define ATEN_OP2(op_name, overload) at::_ops::op_name##_##overload
1373
+ #define ATEN_OP(op_name) at::_ops::op_name
1374
+
1375
+ // WARNING: Please do not call any of the ops in the _ops namespace directly.
1376
+ // Use the ATEN_FN macros. We do not guarantee stability of the naming
1377
+ // scheme for the functions in at::_ops
1378
+
1379
+ // See Note [The ATen Operators API] for details of the at::_ops namespace
1380
+
1381
+ namespace at {
1382
+ namespace _ops {
1383
+
1384
+ } // namespace _ops
1385
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/Parallel-inl.h ADDED
@@ -0,0 +1,93 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <c10/util/Exception.h>
4
+ #include <c10/util/ParallelGuard.h>
5
+ #include <c10/util/SmallVector.h>
6
+
7
+ namespace at {
8
+
9
+ template <class F>
10
+ inline void parallel_for(
11
+ const int64_t begin,
12
+ const int64_t end,
13
+ const int64_t grain_size,
14
+ const F& f) {
15
+ TORCH_INTERNAL_ASSERT_DEBUG_ONLY(grain_size >= 0);
16
+ if (begin >= end) {
17
+ return;
18
+ }
19
+
20
+ #ifdef INTRA_OP_PARALLEL
21
+ at::internal::lazy_init_num_threads();
22
+ const auto numiter = end - begin;
23
+ const bool use_parallel =
24
+ (numiter > grain_size && numiter > 1 && !at::in_parallel_region() &&
25
+ at::get_num_threads() > 1);
26
+ if (!use_parallel) {
27
+ internal::ThreadIdGuard tid_guard(0);
28
+ c10::ParallelGuard guard(true);
29
+ f(begin, end);
30
+ return;
31
+ }
32
+
33
+ internal::invoke_parallel(
34
+ begin, end, grain_size, [&](int64_t begin, int64_t end) {
35
+ c10::ParallelGuard guard(true);
36
+ f(begin, end);
37
+ });
38
+ #else
39
+ internal::ThreadIdGuard tid_guard(0);
40
+ c10::ParallelGuard guard(true);
41
+ f(begin, end);
42
+ #endif
43
+ }
44
+
45
+ template <class scalar_t, class F, class SF>
46
+ inline scalar_t parallel_reduce(
47
+ const int64_t begin,
48
+ const int64_t end,
49
+ const int64_t grain_size,
50
+ const scalar_t ident,
51
+ const F& f,
52
+ const SF& sf) {
53
+ TORCH_CHECK(grain_size >= 0);
54
+ if (begin >= end) {
55
+ return ident;
56
+ }
57
+
58
+ #ifdef INTRA_OP_PARALLEL
59
+ at::internal::lazy_init_num_threads();
60
+ const auto max_threads = at::get_num_threads();
61
+ const bool use_parallel =
62
+ ((end - begin) > grain_size && !at::in_parallel_region() &&
63
+ max_threads > 1);
64
+ if (!use_parallel) {
65
+ internal::ThreadIdGuard tid_guard(0);
66
+ c10::ParallelGuard guard(true);
67
+ return f(begin, end, ident);
68
+ }
69
+
70
+ c10::SmallVector<scalar_t, 64> results(max_threads, ident);
71
+ internal::invoke_parallel(
72
+ begin,
73
+ end,
74
+ grain_size,
75
+ [&](const int64_t my_begin, const int64_t my_end) {
76
+ const auto tid = at::get_thread_num();
77
+ c10::ParallelGuard guard(true);
78
+ results[tid] = f(my_begin, my_end, ident);
79
+ });
80
+
81
+ scalar_t result = ident;
82
+ for (auto partial_result : results) {
83
+ result = sf(result, partial_result);
84
+ }
85
+ return result;
86
+ #else
87
+ internal::ThreadIdGuard tid_guard(0);
88
+ c10::ParallelGuard guard(true);
89
+ return f(begin, end, ident);
90
+ #endif
91
+ }
92
+
93
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/ParallelNative.h ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <c10/util/Exception.h>
4
+
5
+ #define INTRA_OP_PARALLEL
6
+
7
+ namespace at::internal {
8
+
9
+ TORCH_API void invoke_parallel(
10
+ const int64_t begin,
11
+ const int64_t end,
12
+ const int64_t grain_size,
13
+ const std::function<void(int64_t, int64_t)>& f);
14
+
15
+ } // namespace at::internal
.venv/lib/python3.11/site-packages/torch/include/ATen/ParallelOpenMP.h ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <algorithm>
4
+ #include <atomic>
5
+ #include <cstddef>
6
+ #include <exception>
7
+
8
+ #ifdef _OPENMP
9
+ #define INTRA_OP_PARALLEL
10
+
11
+ #include <omp.h>
12
+ #endif
13
+
14
+ #ifdef _OPENMP
15
+ namespace at::internal {
16
+ template <typename F>
17
+ inline void invoke_parallel(
18
+ int64_t begin,
19
+ int64_t end,
20
+ int64_t grain_size,
21
+ const F& f) {
22
+ std::atomic_flag err_flag = ATOMIC_FLAG_INIT;
23
+ std::exception_ptr eptr;
24
+
25
+ #pragma omp parallel
26
+ {
27
+ // choose number of tasks based on grain size and number of threads
28
+ // can't use num_threads clause due to bugs in GOMP's thread pool (See
29
+ // #32008)
30
+ int64_t num_threads = omp_get_num_threads();
31
+ if (grain_size > 0) {
32
+ num_threads = std::min(num_threads, divup((end - begin), grain_size));
33
+ }
34
+
35
+ int64_t tid = omp_get_thread_num();
36
+ int64_t chunk_size = divup((end - begin), num_threads);
37
+ int64_t begin_tid = begin + tid * chunk_size;
38
+ if (begin_tid < end) {
39
+ try {
40
+ internal::ThreadIdGuard tid_guard(tid);
41
+ f(begin_tid, std::min(end, chunk_size + begin_tid));
42
+ } catch (...) {
43
+ if (!err_flag.test_and_set()) {
44
+ eptr = std::current_exception();
45
+ }
46
+ }
47
+ }
48
+ }
49
+ if (eptr) {
50
+ std::rethrow_exception(eptr);
51
+ }
52
+ }
53
+ } // namespace at::internal
54
+ #endif // _OPENMP
.venv/lib/python3.11/site-packages/torch/include/ATen/PythonTorchFunctionTLS.h ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <c10/core/SafePyObject.h>
4
+ #include <c10/macros/Macros.h>
5
+
6
+ namespace at::impl {
7
+
8
+ enum TorchFunctionDisabledState { ENABLED, SUBCLASSES_DISABLED, ALL_DISABLED };
9
+
10
+ struct TORCH_API PythonTorchFunctionTLS {
11
+ static void set_disabled_state(TorchFunctionDisabledState disabled_state_);
12
+ static TorchFunctionDisabledState get_disabled_state();
13
+
14
+ static void push_onto_stack(std::shared_ptr<SafePyObject> mode);
15
+ static const std::shared_ptr<SafePyObject> pop_stack();
16
+ static const std::shared_ptr<SafePyObject>& get_stack_at(int64_t idx);
17
+ static int64_t stack_len();
18
+
19
+ static const PythonTorchFunctionTLS& get_state();
20
+ static void set_state(const PythonTorchFunctionTLS& state);
21
+
22
+ private:
23
+ // The mode TLS is split into
24
+ // - disabled_state, which says which part of torch function are disabled
25
+ // - stack_, which is a vector of modes representing the stack of user
26
+ // defined modes
27
+ TorchFunctionDisabledState disabled_state_ =
28
+ TorchFunctionDisabledState::ENABLED;
29
+ std::vector<std::shared_ptr<c10::SafePyObject>> stack_;
30
+ };
31
+
32
+ TORCH_API bool torch_function_mode_enabled();
33
+
34
+ TORCH_API bool torch_function_all_disabled();
35
+
36
+ } // namespace at::impl
.venv/lib/python3.11/site-packages/torch/include/ATen/RedispatchFunctions.h ADDED
The diff for this file is too large to render. See raw diff
 
.venv/lib/python3.11/site-packages/torch/include/ATen/SmallVector.h ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ #pragma once
2
+ #include <c10/util/SmallVector.h>
.venv/lib/python3.11/site-packages/torch/include/ATen/SparseTensorImpl.h ADDED
@@ -0,0 +1,421 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <ATen/Tensor.h>
4
+ #include <c10/core/TensorImpl.h>
5
+ #include <c10/core/impl/TorchDispatchModeTLS.h>
6
+ #include <c10/util/Exception.h>
7
+ #include <c10/util/irange.h>
8
+
9
+ #ifndef AT_PER_OPERATOR_HEADERS
10
+ #include <ATen/Functions.h>
11
+ #else
12
+ #include <ATen/ops/empty.h>
13
+ #include <ATen/ops/resize.h>
14
+ #endif
15
+
16
+ namespace at {
17
+ struct TORCH_API SparseTensorImpl : public TensorImpl {
18
+ // Stored in COO format, indices + values.
19
+
20
+ // INVARIANTS:
21
+ // sparse_dim: range [0, len(shape)]; sparse_dim + dense_dim = len(shape)
22
+ // dense_dim : range [0, len(shape)]; sparse_dim + dense_dim = len(shape)
23
+ // _indices.shape: dimensionality: 2, shape: (sparse_dim, nnz)
24
+ // _values.shape: dimensionality: 1 + dense_dim. shape: (nnz,
25
+ // shape[sparse_dim:])
26
+
27
+ int64_t sparse_dim_ = 0; // number of sparse dimensions
28
+ int64_t dense_dim_ = 0; // number of dense dimensions
29
+
30
+ Tensor indices_; // always a LongTensor
31
+ Tensor values_;
32
+
33
+ // A sparse tensor is 'coalesced' if every index occurs at most once in
34
+ // the indices tensor, and the indices are in sorted order. (This means
35
+ // that it is very easy to convert a coalesced tensor to CSR format: you
36
+ // need only compute CSR format indices.)
37
+ //
38
+ // Most math operations can only be performed on coalesced sparse tensors,
39
+ // because many algorithms proceed by merging two sorted lists (of indices).
40
+ bool coalesced_ = false;
41
+
42
+ // compute_numel with integer multiplication overflow check, see gh-57542
43
+ void refresh_numel() {
44
+ TensorImpl::safe_refresh_numel();
45
+ }
46
+
47
+ public:
48
+ // Public for now...
49
+ explicit SparseTensorImpl(at::DispatchKeySet, const caffe2::TypeMeta);
50
+
51
+ void release_resources() override;
52
+
53
+ int64_t nnz() const {
54
+ return values_.size(0);
55
+ }
56
+
57
+ c10::SymInt sym_nnz() const {
58
+ return values_.sym_size(0);
59
+ }
60
+ int64_t sparse_dim() const {
61
+ return sparse_dim_;
62
+ }
63
+ int64_t dense_dim() const {
64
+ return dense_dim_;
65
+ }
66
+ bool coalesced() const {
67
+ return coalesced_;
68
+ }
69
+ Tensor indices() const {
70
+ return indices_;
71
+ }
72
+ Tensor values() const {
73
+ return values_;
74
+ }
75
+
76
+ void set_size(int64_t dim, int64_t new_size) override;
77
+ void set_stride(int64_t dim, int64_t new_stride) override;
78
+ void set_storage_offset(int64_t storage_offset) override;
79
+
80
+ #ifdef DEBUG
81
+ bool has_storage() const override;
82
+ #endif
83
+
84
+ // WARNING: This function does NOT preserve invariants of sparse_dim/dense_dim
85
+ // with respect to indices and values
86
+ void raw_resize_(int64_t sparse_dim, int64_t dense_dim, IntArrayRef size) {
87
+ TORCH_CHECK(
88
+ allow_tensor_metadata_change(),
89
+ "raw_resize_ ",
90
+ err_msg_tensor_metadata_change_not_allowed);
91
+ TORCH_CHECK(
92
+ !has_symbolic_sizes_strides_,
93
+ "raw_resize_ called on tensor with symbolic shape")
94
+ set_sizes_and_strides(size, std::vector<int64_t>(size.size()));
95
+ sparse_dim_ = sparse_dim;
96
+ dense_dim_ = dense_dim;
97
+ refresh_numel();
98
+ }
99
+
100
+ // NOTE: This function preserves invariants of sparse_dim/dense_dim with
101
+ // respect to indices and values.
102
+ //
103
+ // NOTE: This function supports the following cases:
104
+ // 1. When we keep the number of dense dimensions unchanged, and NOT shrinking
105
+ // the size of any of the dense dimensions.
106
+ // 2. When we keep the number of sparse dimensions unchanged, and NOT
107
+ // shrinking the size of any of the sparse dimensions.
108
+ // 3. When the sparse tensor has zero nnz, in which case we are free to change
109
+ // the shapes of both its sparse and dense dimensions.
110
+ //
111
+ // This function DOESN'T support (and will throw an error) the following
112
+ // cases:
113
+ // 1. When we attempt to change the number of sparse dimensions on a non-empty
114
+ // sparse tensor (such an operation will invalidate the indices stored).
115
+ // 2. When we attempt to change the number of dense dimensions on a non-empty
116
+ // sparse tensor (such an operation will behave differently from an equivalent
117
+ // dense tensor's resize method, and for API consistency we don't support it).
118
+ // 3. When we attempt to shrink the size of any of the dense dimensions on a
119
+ // non-empty sparse tensor (such an operation will behave differently from an
120
+ // equivalent dense tensor's resize method, and for API consistency we don't
121
+ // support it).
122
+ // 4. When we attempt to shrink the size of any of the sparse dimensions on a
123
+ // non-empty sparse tensor (this could make some of the stored indices
124
+ // out-of-bound and thus unsafe).
125
+ template <typename T>
126
+ void _resize_(int64_t sparse_dim, int64_t dense_dim, ArrayRef<T> size) {
127
+ TORCH_CHECK(
128
+ allow_tensor_metadata_change(),
129
+ "resize_ ",
130
+ err_msg_tensor_metadata_change_not_allowed);
131
+ TORCH_CHECK(
132
+ !has_symbolic_sizes_strides_,
133
+ "resize_ called on tensor with symbolic shape")
134
+ TORCH_CHECK(
135
+ sparse_dim + dense_dim == static_cast<int64_t>(size.size()),
136
+ "number of dimensions must be sparse_dim (",
137
+ sparse_dim,
138
+ ") + dense_dim (",
139
+ dense_dim,
140
+ "), but got ",
141
+ size.size());
142
+ if (nnz() > 0) {
143
+ [[maybe_unused]] auto constexpr alt_options_msg =
144
+ "You could try the following options:\n\
145
+ 1. If you need an empty sparse tensor of this size, call `x = torch.sparse_coo_tensor(size)`.\n\
146
+ 2. If you need to resize this tensor, you have the following options:\n\
147
+ 1. For both sparse and dense dimensions, keep the number of them constant and the size of them non-shrinking, and then try the same call again.\n\
148
+ 2. Or, create a new sparse tensor with the correct indices and values from this sparse tensor.";
149
+
150
+ TORCH_CHECK(
151
+ sparse_dim == sparse_dim_,
152
+ "changing the number of sparse dimensions (from ",
153
+ sparse_dim_,
154
+ " to ",
155
+ sparse_dim,
156
+ ") on a non-empty sparse tensor is not supported.\n",
157
+ alt_options_msg);
158
+
159
+ TORCH_CHECK(
160
+ dense_dim == dense_dim_,
161
+ "changing the number of dense dimensions (from ",
162
+ dense_dim_,
163
+ " to ",
164
+ dense_dim,
165
+ ") on a non-empty sparse tensor is not supported.\n",
166
+ alt_options_msg);
167
+
168
+ bool shrinking_sparse_dims = false;
169
+ bool shrinking_dense_dim = false;
170
+ auto sparse_size_original = generic_sizes<T>().slice(0, sparse_dim);
171
+ auto sparse_size_new = size.slice(0, sparse_dim);
172
+ for (const auto i : c10::irange(sparse_dim)) {
173
+ if (sparse_size_new[i] < sparse_size_original[i]) {
174
+ shrinking_sparse_dims = true;
175
+ break;
176
+ }
177
+ }
178
+ auto dense_size_original = generic_sizes<T>().slice(sparse_dim);
179
+ auto dense_size_new = size.slice(sparse_dim);
180
+ for (const auto i : c10::irange(dense_dim)) {
181
+ if (dense_size_new[i] < dense_size_original[i]) {
182
+ shrinking_dense_dim = true;
183
+ break;
184
+ }
185
+ }
186
+
187
+ TORCH_CHECK(
188
+ !shrinking_sparse_dims,
189
+ "shrinking the size of sparse dimensions (from ",
190
+ sparse_size_original,
191
+ " to ",
192
+ sparse_size_new,
193
+ ") on a non-empty sparse tensor is not supported.\n",
194
+ alt_options_msg);
195
+
196
+ TORCH_CHECK(
197
+ !shrinking_dense_dim,
198
+ "shrinking the size of dense dimensions (from ",
199
+ dense_size_original,
200
+ " to ",
201
+ dense_size_new,
202
+ ") on a non-empty sparse tensor is not supported.\n",
203
+ alt_options_msg);
204
+ }
205
+
206
+ auto sizes_and_strides = generic_sizes<T>();
207
+ const bool size_equals_sizes = std::equal(
208
+ size.begin(),
209
+ size.end(),
210
+ sizes_and_strides.begin(),
211
+ sizes_and_strides.end());
212
+ if ((!size_equals_sizes) || (sparse_dim != sparse_dim_) ||
213
+ (dense_dim != dense_dim_)) {
214
+ auto nnz = at::symint::sizes<T>(values())[0];
215
+ std::vector<T> values_size = {nnz};
216
+ auto dense_size = size.slice(sparse_dim);
217
+ values_size.insert(
218
+ values_size.end(), dense_size.begin(), dense_size.end());
219
+ at::symint::resize_<T>(values_, values_size);
220
+ at::symint::resize_<T>(indices_, {T(sparse_dim), nnz});
221
+ }
222
+
223
+ if (!size_equals_sizes) {
224
+ set_sizes_and_strides(size, std::vector<T>(size.size()));
225
+ }
226
+ sparse_dim_ = sparse_dim;
227
+ dense_dim_ = dense_dim;
228
+ refresh_numel();
229
+ }
230
+
231
+ void resize_(int64_t sparse_dim, int64_t dense_dim, ArrayRef<int64_t> size) {
232
+ return _resize_(sparse_dim, dense_dim, size);
233
+ }
234
+
235
+ void resize_(
236
+ int64_t sparse_dim,
237
+ int64_t dense_dim,
238
+ ArrayRef<c10::SymInt> size) {
239
+ return _resize_(sparse_dim, dense_dim, size);
240
+ }
241
+
242
+ // NOTE: this function will resize the sparse tensor and also set `indices`
243
+ // and `values` to empty.
244
+ void resize_and_clear_(
245
+ int64_t sparse_dim,
246
+ int64_t dense_dim,
247
+ IntArrayRef size) {
248
+ TORCH_CHECK(
249
+ allow_tensor_metadata_change(),
250
+ "resize_and_clear_ ",
251
+ err_msg_tensor_metadata_change_not_allowed);
252
+ TORCH_CHECK(
253
+ !has_symbolic_sizes_strides_,
254
+ "resize_and_clear_ called on tensor with symbolic shape")
255
+ TORCH_CHECK(
256
+ sparse_dim + dense_dim == static_cast<int64_t>(size.size()),
257
+ "number of dimensions must be sparse_dim (",
258
+ sparse_dim,
259
+ ") + dense_dim (",
260
+ dense_dim,
261
+ "), but got ",
262
+ size.size());
263
+
264
+ set_sizes_and_strides(size, std::vector<int64_t>(size.size()));
265
+ sparse_dim_ = sparse_dim;
266
+ dense_dim_ = dense_dim;
267
+
268
+ auto empty_indices = at::empty({sparse_dim, 0}, indices().options());
269
+ std::vector<int64_t> values_size = {0};
270
+ auto dense_size = sizes().slice(sparse_dim);
271
+ values_size.insert(values_size.end(), dense_size.begin(), dense_size.end());
272
+ auto empty_values = at::empty(values_size, values().options());
273
+ set_indices_and_values_unsafe(empty_indices, empty_values);
274
+ refresh_numel();
275
+ }
276
+
277
+ void set_coalesced(bool coalesced) {
278
+ TORCH_CHECK(
279
+ allow_tensor_metadata_change(),
280
+ "set_coalesced ",
281
+ err_msg_tensor_metadata_change_not_allowed);
282
+ coalesced_ = coalesced;
283
+ }
284
+
285
+ // NOTE: this function is only used internally and not exposed to Python
286
+ // frontend
287
+ void set_nnz_and_narrow(int64_t new_nnz) {
288
+ TORCH_CHECK(
289
+ allow_tensor_metadata_change(),
290
+ "set_nnz_and_narrow ",
291
+ err_msg_tensor_metadata_change_not_allowed);
292
+ AT_ASSERT(new_nnz <= nnz());
293
+ indices_ = indices_.narrow(1, 0, new_nnz);
294
+ values_ = values_.narrow(0, 0, new_nnz);
295
+ if (new_nnz < 2) {
296
+ coalesced_ = true;
297
+ }
298
+ }
299
+
300
+ // Takes indices and values and directly puts them into the sparse tensor, no
301
+ // copy. NOTE: this function is unsafe because it doesn't check whether any
302
+ // indices are out of boundaries of `sizes`, so it should ONLY be used where
303
+ // we know that the indices are guaranteed to be within bounds. This used to
304
+ // be called THSTensor_(_move) NB: This used to be able to avoid a refcount
305
+ // bump, but I was too lazy to make it happen
306
+ void set_indices_and_values_unsafe(
307
+ const Tensor& indices,
308
+ const Tensor& values);
309
+
310
+ template <typename VariableVersion>
311
+ c10::intrusive_ptr<TensorImpl> shallow_copy_and_detach_core(
312
+ VariableVersion&& version_counter,
313
+ bool allow_tensor_metadata_change) const {
314
+ const auto mode_stack_len = c10::impl::TorchDispatchModeTLS::stack_len();
315
+ c10::impl::PyInterpreter&& interpreter = nullptr;
316
+ if (mode_stack_len > 0 &&
317
+ !c10::impl::tls_is_dispatch_key_excluded(DispatchKey::Python)) {
318
+ const auto& cur_torch_dispatch_mode_state =
319
+ c10::impl::TorchDispatchModeTLS::get_stack_at(mode_stack_len - 1);
320
+ interpreter = cur_torch_dispatch_mode_state->pyinterpreter();
321
+ } else if (
322
+ key_set_.has(DispatchKey::Python) &&
323
+ !c10::impl::tls_is_dispatch_key_excluded(DispatchKey::Python)) {
324
+ interpreter = pyobj_slot_.load_pyobj_interpreter();
325
+ } else {
326
+ // otherwise just copy the SparseTensorImpl and not the PyObject.
327
+ auto impl = c10::make_intrusive<SparseTensorImpl>(key_set(), dtype());
328
+ copy_tensor_metadata(
329
+ /*src_sparse_impl=*/this,
330
+ /*dest_sparse_impl=*/impl.get(),
331
+ /*version_counter=*/version_counter,
332
+ /*allow_tensor_metadata_change=*/allow_tensor_metadata_change);
333
+ impl->refresh_numel();
334
+ return impl;
335
+ }
336
+ auto r = interpreter->detach(this);
337
+ r->set_version_counter(std::forward<VariableVersion>(version_counter));
338
+ r->set_allow_tensor_metadata_change(allow_tensor_metadata_change);
339
+ return r;
340
+ }
341
+
342
+ /**
343
+ * Return a TensorImpl that is a shallow-copy of this TensorImpl.
344
+ *
345
+ * For usage of `version_counter` and `allow_tensor_metadata_change`,
346
+ * see NOTE [ TensorImpl Shallow-Copying ].
347
+ */
348
+ c10::intrusive_ptr<TensorImpl> shallow_copy_and_detach(
349
+ const c10::VariableVersion& version_counter,
350
+ bool allow_tensor_metadata_change) const override {
351
+ return shallow_copy_and_detach_core(
352
+ version_counter, allow_tensor_metadata_change);
353
+ }
354
+
355
+ /**
356
+ * Return a TensorImpl that is a shallow-copy of this TensorImpl.
357
+ *
358
+ * For usage of `version_counter` and `allow_tensor_metadata_change`,
359
+ * see NOTE [ TensorImpl Shallow-Copying ].
360
+ */
361
+ c10::intrusive_ptr<TensorImpl> shallow_copy_and_detach(
362
+ c10::VariableVersion&& version_counter,
363
+ bool allow_tensor_metadata_change) const override {
364
+ return shallow_copy_and_detach_core(
365
+ std::move(version_counter), allow_tensor_metadata_change);
366
+ }
367
+
368
+ /**
369
+ * Shallow-copies data from another TensorImpl into this TensorImpl.
370
+ *
371
+ * For why this function doesn't check this TensorImpl's
372
+ * `allow_tensor_metadata_change_`, see NOTE [ TensorImpl Shallow-Copying ].
373
+ */
374
+ void shallow_copy_from(const c10::intrusive_ptr<TensorImpl>& impl) override {
375
+ AT_ASSERT(has_compatible_shallow_copy_type(impl->key_set()));
376
+ auto sparse_impl = static_cast<const SparseTensorImpl*>(impl.get());
377
+ copy_tensor_metadata(
378
+ /*src_sparse_impl=*/sparse_impl,
379
+ /*dest_sparse_impl=*/this,
380
+ /*version_counter=*/version_counter(),
381
+ /*allow_tensor_metadata_change=*/allow_tensor_metadata_change());
382
+ refresh_numel();
383
+ }
384
+
385
+ private:
386
+ explicit SparseTensorImpl(
387
+ at::DispatchKeySet,
388
+ const caffe2::TypeMeta,
389
+ at::Tensor indices,
390
+ at::Tensor values);
391
+
392
+ /**
393
+ * Copy the tensor metadata fields (e.g. sizes / strides / storage pointer /
394
+ * storage_offset) from one TensorImpl to another TensorImpl.
395
+ *
396
+ * For usage of `version_counter` and `allow_tensor_metadata_change`, see NOTE
397
+ * [ TensorImpl Shallow-Copying ].
398
+ */
399
+ static void copy_tensor_metadata(
400
+ const SparseTensorImpl* src_sparse_impl,
401
+ SparseTensorImpl* dest_sparse_impl,
402
+ c10::VariableVersion version_counter,
403
+ bool allow_tensor_metadata_change) {
404
+ TensorImpl::copy_tensor_metadata(
405
+ src_sparse_impl,
406
+ dest_sparse_impl,
407
+ std::move(version_counter),
408
+ allow_tensor_metadata_change);
409
+
410
+ // Sparse-specific fields
411
+ dest_sparse_impl->sparse_dim_ = src_sparse_impl->sparse_dim();
412
+ dest_sparse_impl->dense_dim_ = src_sparse_impl->dense_dim();
413
+ dest_sparse_impl->indices_ = src_sparse_impl->indices();
414
+ dest_sparse_impl->values_ = src_sparse_impl->values();
415
+ dest_sparse_impl->coalesced_ = src_sparse_impl->coalesced();
416
+ }
417
+
418
+ const char* tensorimpl_type_name() const override;
419
+ };
420
+
421
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/StorageUtils.h ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <c10/core/Storage.h>
4
+ #include <c10/core/StorageImpl.h>
5
+ #include <c10/util/intrusive_ptr.h>
6
+
7
+ namespace at {
8
+
9
+ class TensorBase;
10
+
11
+ // Here we define a series of utils to create/manipulate ATen backed
12
+ // c10 storage implementations.
13
+
14
+ /**
15
+ * Create a new shared memory storage impl managed by file descriptor
16
+ *
17
+ * @param size size in bytes
18
+ */
19
+ C10_EXPORT c10::intrusive_ptr<c10::StorageImpl> new_shm_fd_storage(size_t size);
20
+
21
+ /**
22
+ * Copy src to dst
23
+ * Caller must guarantee the validness of the storage objects
24
+ * during the entire copy process, esp. when it's async.
25
+ *
26
+ * This can probably live in c10 namespace later if needed,
27
+ * but for now keep it in at to keep implementation simple.
28
+ *
29
+ * @param dst dst tensor
30
+ * @param src src tensor
31
+ * @param non_blocking (default false) whether this operation blocks caller
32
+ */
33
+ C10_EXPORT void storage_copy(
34
+ c10::Storage& dst,
35
+ const c10::Storage& src,
36
+ bool non_blocking = false);
37
+
38
+ /**
39
+ * In place change the storage to shm based.
40
+ *
41
+ * This is only applicable to CPU tensors not already shared.
42
+ * Otherwise, it's a no op to mirror the THP tensor behavior:
43
+ * https://pytorch.org/docs/stable/generated/torch.Tensor.share_memory_.html
44
+ *
45
+ * @param t a tensor
46
+ */
47
+ C10_EXPORT void share_memory_(TensorBase& t);
48
+
49
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/TensorAccessor.h ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ #pragma once
2
+ #include <ATen/core/TensorAccessor.h>
.venv/lib/python3.11/site-packages/torch/include/ATen/TensorIndexing.h ADDED
@@ -0,0 +1,737 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <ATen/ExpandUtils.h>
4
+ #include <ATen/ScalarOps.h>
5
+ #include <ATen/core/Tensor.h>
6
+ #include <ATen/core/TensorBody.h>
7
+ #include <c10/core/SymInt.h>
8
+ #include <c10/util/irange.h>
9
+ #include <optional>
10
+
11
+ #ifndef AT_PER_OPERATOR_HEADERS
12
+ #include <ATen/Functions.h>
13
+ #include <ATen/NativeFunctions.h>
14
+ #else
15
+ #include <ATen/ops/alias.h>
16
+ #include <ATen/ops/empty.h>
17
+ #include <ATen/ops/scalar_tensor.h>
18
+ #include <ATen/ops/zeros.h>
19
+ #endif
20
+
21
+ #include <ATen/core/List.h>
22
+
23
+ #include <utility>
24
+
25
+ namespace at::indexing {
26
+
27
+ constexpr int64_t INDEX_MIN = c10::SymInt::min_representable_int();
28
+ constexpr int64_t INDEX_MAX = -(INDEX_MIN + 1);
29
+
30
+ enum class TensorIndexType { None, Ellipsis, SymInt, Boolean, Slice, Tensor };
31
+
32
+ constexpr std::nullopt_t None = std::nullopt;
33
+
34
+ struct TORCH_API EllipsisIndexType final {
35
+ EllipsisIndexType() = default;
36
+ };
37
+ TORCH_API extern const EllipsisIndexType Ellipsis;
38
+
39
+ struct TORCH_API Slice final {
40
+ public:
41
+ Slice(
42
+ std::optional<c10::SymInt> start_index = std::nullopt,
43
+ std::optional<c10::SymInt> stop_index = std::nullopt,
44
+ std::optional<c10::SymInt> step_index = std::nullopt) {
45
+ if (!step_index.has_value()) {
46
+ step_ = c10::SymInt(1);
47
+ } else {
48
+ step_ = std::move(step_index).value();
49
+ }
50
+
51
+ TORCH_CHECK_VALUE(
52
+ step_.sym_ne(0).expect_true(__FILE__, __LINE__),
53
+ "slice step cannot be zero");
54
+
55
+ if (!start_index.has_value()) {
56
+ start_ = c10::SymInt(step_ < 0 ? INDEX_MAX : 0);
57
+ } else {
58
+ start_ = std::move(start_index).value();
59
+ }
60
+
61
+ if (!stop_index.has_value()) {
62
+ stop_ = c10::SymInt(step_ < 0 ? INDEX_MIN : INDEX_MAX);
63
+ } else {
64
+ stop_ = std::move(stop_index).value();
65
+ }
66
+ }
67
+
68
+ inline c10::SymInt start() const {
69
+ return start_;
70
+ }
71
+
72
+ inline c10::SymInt stop() const {
73
+ return stop_;
74
+ }
75
+
76
+ inline c10::SymInt step() const {
77
+ return step_;
78
+ }
79
+
80
+ private:
81
+ c10::SymInt start_;
82
+ c10::SymInt stop_;
83
+ c10::SymInt step_;
84
+ };
85
+
86
+ TORCH_API std::ostream& operator<<(std::ostream& stream, const Slice& slice);
87
+
88
+ // `at::indexing::TensorIndex` is used for converting C++ tensor indices such as
89
+ // `{None, "...", Ellipsis, 0, true, Slice(1, None, 2), torch::tensor({1, 2})}`
90
+ // into its equivalent `std::vector<TensorIndex>`, so that further tensor
91
+ // indexing operations can be performed using the supplied indices.
92
+ //
93
+ // There is one-to-one correspondence between Python and C++ tensor index types:
94
+ // Python | C++
95
+ // -----------------------------------------------------
96
+ // `None` | `at::indexing::None`
97
+ // `Ellipsis` | `at::indexing::Ellipsis`
98
+ // `...` | `"..."`
99
+ // `123` | `123`
100
+ // `True` / `False` | `true` / `false`
101
+ // `:` | `Slice()` / `Slice(None, None)`
102
+ // `::` | `Slice()` / `Slice(None, None, None)`
103
+ // `1:` | `Slice(1, None)`
104
+ // `1::` | `Slice(1, None, None)`
105
+ // `:3` | `Slice(None, 3)`
106
+ // `:3:` | `Slice(None, 3, None)`
107
+ // `::2` | `Slice(None, None, 2)`
108
+ // `1:3` | `Slice(1, 3)`
109
+ // `1::2` | `Slice(1, None, 2)`
110
+ // `:3:2` | `Slice(None, 3, 2)`
111
+ // `1:3:2` | `Slice(1, 3, 2)`
112
+ // `torch.tensor([1, 2])`) | `torch::tensor({1, 2})`
113
+ struct TORCH_API TensorIndex final {
114
+ // Case 1: `at::indexing::None`
115
+ TensorIndex(std::nullopt_t) : type_(TensorIndexType::None) {}
116
+
117
+ // Case 2: "..." / `at::indexing::Ellipsis`
118
+ TensorIndex(at::indexing::EllipsisIndexType)
119
+ : type_(TensorIndexType::Ellipsis) {}
120
+ TensorIndex(const char* str) : TensorIndex(at::indexing::Ellipsis) {
121
+ TORCH_CHECK_VALUE(
122
+ strcmp(str, "...") == 0,
123
+ "Expected \"...\" to represent an ellipsis index, but got \"",
124
+ str,
125
+ "\"");
126
+ }
127
+
128
+ // Case 3: (Sym) Integer value
129
+ TensorIndex(SymInt integer)
130
+ : integer_(std::move(integer)), type_(TensorIndexType::SymInt) {}
131
+ TensorIndex(int64_t integer) : TensorIndex(SymInt(integer)) {}
132
+ TensorIndex(int integer) : TensorIndex(SymInt(integer)) {}
133
+
134
+ // Case 4: Boolean value
135
+ template <class T, class = std::enable_if_t<std::is_same_v<bool, T>>>
136
+ TensorIndex(T boolean) : boolean_(boolean), type_(TensorIndexType::Boolean) {}
137
+
138
+ // Case 5: Slice represented in `at::indexing::Slice` form
139
+ TensorIndex(Slice slice)
140
+ : slice_(std::move(slice)), type_(TensorIndexType::Slice) {}
141
+
142
+ // Case 6: Tensor value
143
+ TensorIndex(Tensor tensor)
144
+ : tensor_(std::move(tensor)), type_(TensorIndexType::Tensor) {}
145
+
146
+ inline bool is_none() const {
147
+ return type_ == TensorIndexType::None;
148
+ }
149
+
150
+ inline bool is_ellipsis() const {
151
+ return type_ == TensorIndexType::Ellipsis;
152
+ }
153
+
154
+ inline bool is_integer() const {
155
+ return type_ == TensorIndexType::SymInt;
156
+ }
157
+
158
+ inline SymInt integer() const {
159
+ return integer_;
160
+ }
161
+
162
+ inline bool is_boolean() const {
163
+ return type_ == TensorIndexType::Boolean;
164
+ }
165
+
166
+ inline bool boolean() const {
167
+ return boolean_;
168
+ }
169
+
170
+ inline bool is_slice() const {
171
+ return type_ == TensorIndexType::Slice;
172
+ }
173
+
174
+ inline const Slice& slice() const {
175
+ return slice_;
176
+ }
177
+
178
+ inline bool is_tensor() const {
179
+ return type_ == TensorIndexType::Tensor;
180
+ }
181
+
182
+ inline const Tensor& tensor() const {
183
+ return tensor_;
184
+ }
185
+
186
+ private:
187
+ SymInt integer_ = 0;
188
+ bool boolean_ = false;
189
+ Slice slice_;
190
+ Tensor tensor_;
191
+ TensorIndexType type_;
192
+ };
193
+
194
+ TORCH_API std::ostream& operator<<(
195
+ std::ostream& stream,
196
+ const TensorIndex& tensor_index);
197
+ TORCH_API std::ostream& operator<<(
198
+ std::ostream& stream,
199
+ const std::vector<TensorIndex>& tensor_indices);
200
+
201
+ namespace impl {
202
+ inline Tensor applySlice(
203
+ const Tensor& self,
204
+ int64_t dim,
205
+ c10::SymInt start,
206
+ c10::SymInt stop,
207
+ c10::SymInt step,
208
+ bool disable_slice_optimization,
209
+ const at::Device& self_device,
210
+ const std::optional<SymIntArrayRef>& self_sizes) {
211
+ // TODO: implement negative step
212
+ TORCH_CHECK_VALUE(
213
+ step.sym_gt(0).expect_true(__FILE__, __LINE__),
214
+ "step must be greater than zero");
215
+
216
+ // See NOTE [nested tensor size for indexing]
217
+ if (self_sizes.has_value()) {
218
+ // Skip this optimization if we are tracing, as the trace may be polymorphic
219
+ // over the shape of the `self` tensor, and we still want to record
220
+ // the slice.
221
+ SymInt length = (self_device == at::kCPU || self_device == at::kCUDA)
222
+ ? (*self_sizes)[dim]
223
+ : self.sym_size(dim);
224
+ if (!disable_slice_optimization &&
225
+ TORCH_GUARD_SIZE_OBLIVIOUS(start.sym_eq(0)) &&
226
+ TORCH_GUARD_SIZE_OBLIVIOUS(length.sym_eq(stop)) && step == 1) {
227
+ return self;
228
+ }
229
+ }
230
+ return self.slice_symint(
231
+ dim, std::move(start), std::move(stop), std::move(step));
232
+ }
233
+
234
+ inline Tensor applySelect(
235
+ const Tensor& self,
236
+ int64_t dim,
237
+ SymInt index,
238
+ int64_t real_dim,
239
+ const at::Device& /*self_device*/,
240
+ const std::optional<SymIntArrayRef>& self_sizes) {
241
+ // See NOTE [nested tensor size for indexing]
242
+ if (self_sizes.has_value()) {
243
+ auto maybe_index = index.maybe_as_int();
244
+ if (maybe_index.has_value()) {
245
+ TORCH_CHECK_INDEX(
246
+ !(maybe_index.value() == 0 && dim == 0 && self_sizes->empty()),
247
+ "invalid index of a 0-dim tensor. ",
248
+ "Use `tensor.item()` in Python or `tensor.item<T>()` in C++ to convert a 0-dim tensor to a number");
249
+ }
250
+
251
+ auto size = (*self_sizes)[dim];
252
+ // Note: `size >= -index` is not equivalent to `size > -1 - index` if index
253
+ // is INT64_MIN For std::numeric_limits<int64_t>::min() result of unary
254
+ // minus is undefined by the standard but in practice is equal to self. On
255
+ // the other hand, indexing wraping is valid for all negative int64_t
256
+ // values, as x[INT64_MIN] is the same as x[INT64_MAX]
257
+ TORCH_CHECK_INDEX(
258
+ size > -1 - index && size > index,
259
+ "index ",
260
+ index,
261
+ " is out of bounds for dimension ",
262
+ real_dim,
263
+ " with size ",
264
+ size);
265
+ }
266
+
267
+ // if the index is negative, do not normalize it because that would fix the
268
+ // index on the current tensor size in the tracer. aten::select also works on
269
+ // negative indices
270
+ return self.select_symint(dim, std::move(index));
271
+ }
272
+
273
+ inline Tensor boolToIndexingTensorCPUOrCUDA(const Tensor& self, bool value) {
274
+ // booleans add a dimension of size 1. true indexes this dimension as if 0:,
275
+ // false as empty.
276
+ if (value) {
277
+ return at::empty({1}, self.options().dtype(kLong)).fill_(0.);
278
+ } else {
279
+ return at::empty({0}, self.options().dtype(kLong));
280
+ }
281
+ }
282
+
283
+ inline Tensor boolToIndexingTensorNonNativeDeviceType(
284
+ const Tensor& self,
285
+ bool value) {
286
+ // booleans add a dimension of size 1. true indexes this dimension as if 0:,
287
+ // false as empty.
288
+ if (value) {
289
+ return at::zeros({1}, self.options().dtype(kLong));
290
+ } else {
291
+ return at::empty({0}, self.options().dtype(kLong));
292
+ }
293
+ }
294
+
295
+ inline Tensor boolToIndexingTensor(
296
+ const Tensor& self,
297
+ bool value,
298
+ const at::Device& self_device) {
299
+ if (self_device == at::kCPU || self_device == at::kCUDA) {
300
+ return boolToIndexingTensorCPUOrCUDA(self, value);
301
+ } else {
302
+ return boolToIndexingTensorNonNativeDeviceType(self, value);
303
+ }
304
+ }
305
+
306
+ inline Tensor scalarToTensorNonNativeDeviceType(
307
+ const Scalar& v,
308
+ const TensorOptions& options) {
309
+ return at::scalar_tensor(v, options);
310
+ }
311
+
312
+ inline void recordTensorIndex(
313
+ const Tensor& tensor,
314
+ std::vector<Tensor>& outIndices,
315
+ int64_t* dim_ptr) {
316
+ // TODO: check scalarType
317
+ outIndices.resize(*dim_ptr + 1);
318
+ outIndices[*dim_ptr] = tensor;
319
+ (*dim_ptr)++;
320
+ };
321
+
322
+ inline c10::List<::std::optional<Tensor>> typeConvertIndices(
323
+ const Tensor& /*self*/,
324
+ std::vector<Tensor>&& indices) {
325
+ c10::List<::std::optional<Tensor>> converted_inds;
326
+ converted_inds.reserve(indices.size());
327
+ for (auto&& i : std::move(indices)) {
328
+ converted_inds.push_back(std::move(i));
329
+ }
330
+ return converted_inds;
331
+ }
332
+
333
+ // NOTE: Why do we mirror instead of replace the `count_specified_dimensions`
334
+ // function in torch/csrc/autograd/python_variable_indexing.cpp? It's because
335
+ // `count_specified_dimensions` is on the hot path of Python tensor multi-dim
336
+ // indexing (i.e. it's called by `applySlicing` which is called by
337
+ // `THPVariable_getitem` / `THPVariable_setitem` when handling indexing of more
338
+ // than one dimension). If we were to merge the Python/C++
339
+ // `count_specified_dimensions` function, on the Python side we would have to
340
+ // construct a `std::vector` container to be consumed by the C++
341
+ // `count_specified_dimensions` function, which adds 100s of nanoseconds
342
+ // overhead and is undesirable.
343
+ inline int64_t count_specified_dimensions(
344
+ const ArrayRef<TensorIndex>& indices) {
345
+ // Count the number of indexed dimensions (everything but ellipsis and None)
346
+ int64_t count = 0;
347
+ for (auto& obj : indices) {
348
+ if (obj.is_tensor()) {
349
+ auto& tensor = obj.tensor();
350
+ if (tensor.scalar_type() == kByte || tensor.scalar_type() == kBool) {
351
+ count += tensor.dim();
352
+ } else {
353
+ count++;
354
+ }
355
+ } else if (!obj.is_none() && !obj.is_ellipsis() && !obj.is_boolean()) {
356
+ count++;
357
+ }
358
+ }
359
+ return count;
360
+ }
361
+ } // namespace impl
362
+
363
+ // NOTE: Many functions below are only for consumption from Python indexing
364
+ // implementation, they include:
365
+ //
366
+ // - `Tensor scalarToTensor(...)`
367
+ // - `IntArrayRef slicePrefix1sSize(...)`
368
+ // - `void copy_to(...)`
369
+ // - `Tensor handleDimInMultiDimIndexing(...)`
370
+ // - `Tensor dispatch_index(...)`
371
+ // - `Tensor dispatch_index_put_(...)`
372
+ // - `Tensor get_item(...)`
373
+ // - `void set_item(...)`
374
+ //
375
+ // The rest of the functions are in `at::indexing::impl` namespace, signifying
376
+ // that they shouldn't be used from Python indexing implementation.
377
+ inline Tensor scalarToTensor(
378
+ const Scalar& v,
379
+ const TensorOptions& options,
380
+ const at::Device& self_device) {
381
+ if (self_device == at::kCPU && !v.isSymbolic()) {
382
+ return at::detail::scalar_tensor_static(
383
+ v, options.dtype_opt()->toScalarType(), self_device);
384
+ } else {
385
+ return impl::scalarToTensorNonNativeDeviceType(v, options);
386
+ }
387
+ }
388
+
389
+ // To match numpy semantics:
390
+ // As a special case for backwards compatibility,
391
+ // strip away unit dimensions from the left of 'src'
392
+ inline SymIntArrayRef slicePrefix1sSize(const SymIntArrayRef& sizes) {
393
+ size_t first_non1_src = sizes.size();
394
+ for (const auto i : c10::irange(sizes.size())) {
395
+ // Unbacked SymInt has different behavior, but this is sound because
396
+ // failing to slice will only ever cause an error, not divergent
397
+ // behavior
398
+ if (!sizes[i].has_hint() || sizes[i] != 1) {
399
+ first_non1_src = i;
400
+ break;
401
+ }
402
+ }
403
+
404
+ return sizes.slice(first_non1_src);
405
+ }
406
+
407
+ inline void copy_to(const Tensor& dst, const Tensor& src) {
408
+ if (dst.sym_sizes().equals(src.sym_sizes())) {
409
+ // A shortcut to avoid generating hard-coded constant sizes during tracing.
410
+ // This is not a perfect solution: when src & dst have different shapes,
411
+ // constants will still appear. Users can workaround that case by
412
+ // dst[index..] = src.reshape(..)
413
+ dst.copy_(src);
414
+ return;
415
+ } else if (src.dim() == 0 && src.device().type() == at::kCPU) {
416
+ dst.fill_(src);
417
+ return;
418
+ }
419
+ auto src_view = src.view_symint(slicePrefix1sSize(src.sym_sizes()));
420
+ c10::MaybeOwned<Tensor> b_src = expand_inplace(dst, src_view, "setitem");
421
+ dst.copy_(*b_src);
422
+ }
423
+
424
+ // See NOTE [ Setting `disable_slice_optimization` when calling C++ tensor
425
+ // indexing functions from Python ]
426
+ inline Tensor handleDimInMultiDimIndexing(
427
+ const Tensor& prev_dim_result,
428
+ const Tensor& original_tensor,
429
+ const TensorIndex& index,
430
+ int64_t* dim_ptr,
431
+ int64_t* specified_dims_ptr,
432
+ int64_t real_dim,
433
+ std::vector<Tensor>& outIndices,
434
+ bool disable_slice_optimization,
435
+ const at::Device& original_tensor_device,
436
+ const std::optional<SymIntArrayRef>& prev_dim_result_sizes) {
437
+ if (index.is_integer()) {
438
+ return impl::applySelect(
439
+ prev_dim_result,
440
+ *dim_ptr,
441
+ index.integer(),
442
+ real_dim,
443
+ original_tensor_device,
444
+ prev_dim_result_sizes);
445
+ } else if (index.is_slice()) {
446
+ Tensor result = impl::applySlice(
447
+ prev_dim_result,
448
+ *dim_ptr,
449
+ index.slice().start(),
450
+ index.slice().stop(),
451
+ index.slice().step(),
452
+ /*disable_slice_optimization=*/disable_slice_optimization,
453
+ original_tensor_device,
454
+ prev_dim_result_sizes);
455
+ (*dim_ptr)++;
456
+ return result;
457
+ } else if (index.is_ellipsis()) {
458
+ (*dim_ptr) += original_tensor.dim() - (*specified_dims_ptr);
459
+ return prev_dim_result;
460
+ } else if (index.is_none()) {
461
+ Tensor result = prev_dim_result.unsqueeze(*dim_ptr);
462
+ (*dim_ptr)++;
463
+ return result;
464
+ } else if (index.is_boolean()) {
465
+ Tensor result = prev_dim_result.unsqueeze(*dim_ptr);
466
+ impl::recordTensorIndex(
467
+ impl::boolToIndexingTensor(
468
+ result, index.boolean(), original_tensor_device),
469
+ outIndices,
470
+ dim_ptr);
471
+ return result;
472
+ } else if (index.is_tensor()) {
473
+ Tensor result = prev_dim_result;
474
+ const Tensor& tensor = index.tensor();
475
+ auto scalar_type = tensor.scalar_type();
476
+ if (tensor.dim() == 0 &&
477
+ at::isIntegralType(scalar_type, /*includeBool=*/true)) {
478
+ if (scalar_type != at::kByte && scalar_type != at::kBool) {
479
+ result = impl::applySelect(
480
+ result,
481
+ *dim_ptr,
482
+ tensor.item<int64_t>(),
483
+ real_dim,
484
+ original_tensor_device,
485
+ prev_dim_result_sizes);
486
+ } else {
487
+ result = result.unsqueeze(*dim_ptr);
488
+ if (scalar_type == at::kBool) {
489
+ impl::recordTensorIndex(
490
+ impl::boolToIndexingTensor(
491
+ result, tensor.item<bool>() != 0, original_tensor_device),
492
+ outIndices,
493
+ dim_ptr);
494
+ } else {
495
+ impl::recordTensorIndex(
496
+ impl::boolToIndexingTensor(
497
+ result, tensor.item<uint8_t>() != 0, original_tensor_device),
498
+ outIndices,
499
+ dim_ptr);
500
+ }
501
+ }
502
+ } else {
503
+ impl::recordTensorIndex(tensor, outIndices, dim_ptr);
504
+ }
505
+ return result;
506
+ } else {
507
+ TORCH_INTERNAL_ASSERT(false, "Invalid TensorIndex type");
508
+ }
509
+ }
510
+
511
+ namespace impl {
512
+ // This mirrors `applySlicing` in
513
+ // torch/csrc/autograd/python_variable_indexing.cpp
514
+ inline Tensor applySlicing(
515
+ const Tensor& self,
516
+ const ArrayRef<TensorIndex>& indices,
517
+ std::vector<Tensor>& outIndices,
518
+ bool disable_slice_optimization,
519
+ const at::Device& self_device,
520
+ const std::optional<SymIntArrayRef>& self_sizes) {
521
+ int64_t dim = 0;
522
+ int64_t specified_dims = impl::count_specified_dimensions(indices);
523
+
524
+ // See NOTE [nested tensor size for indexing]
525
+ if (self_sizes.has_value()) {
526
+ TORCH_CHECK_INDEX(
527
+ specified_dims <= (int64_t)self_sizes->size(),
528
+ "too many indices for tensor of dimension ",
529
+ (int)self_sizes->size());
530
+ }
531
+
532
+ Tensor result = self;
533
+ for (const auto i : c10::irange(indices.size())) {
534
+ auto& obj = indices[i];
535
+ // See NOTE [nested tensor size for indexing]
536
+ std::optional<SymIntArrayRef> result_sizes = result.is_nested()
537
+ ? std::optional<SymIntArrayRef>(std::nullopt)
538
+ : std::optional<SymIntArrayRef>(result.sym_sizes());
539
+ result = handleDimInMultiDimIndexing(
540
+ /*prev_dim_result=*/result,
541
+ /*original_tensor=*/self,
542
+ /*index=*/obj,
543
+ /*dim_ptr=*/&dim,
544
+ /*specified_dims_ptr=*/&specified_dims,
545
+ /*real_dim=*/static_cast<int64_t>(i),
546
+ /*outIndices=*/outIndices,
547
+ /*disable_slice_optimization=*/disable_slice_optimization,
548
+ /*original_tensor_device=*/self_device,
549
+ /*prev_dim_result_sizes=*/result_sizes);
550
+ }
551
+ return result;
552
+ }
553
+ } // namespace impl
554
+
555
+ inline Tensor dispatch_index(
556
+ const Tensor& self,
557
+ std::vector<Tensor>&& indices) {
558
+ return self.index(impl::typeConvertIndices(self, std::move(indices)));
559
+ }
560
+
561
+ inline Tensor dispatch_index_put_(
562
+ Tensor& self,
563
+ std::vector<Tensor>&& indices,
564
+ const Tensor& value) {
565
+ return self.index_put_(
566
+ impl::typeConvertIndices(self, std::move(indices)), value);
567
+ }
568
+
569
+ // NOTE [ Setting `disable_slice_optimization` when calling C++ tensor indexing
570
+ // functions from Python ]
571
+ //
572
+ // Question: When should we set `disable_slice_optimization` to `true` when
573
+ // calling C++ tensor indexing functions from Python indexing code?
574
+ //
575
+ // Answer: What "slice optimization" means: when we have a slicing expression
576
+ // like `x[0:5, 0]`, where the sliced tensor was of size 5 in dimension 0, we
577
+ // would skip dispatching the actual slice call as an optimization. However,
578
+ // here are the cases where we DON'T want this optimization:
579
+ //
580
+ // 1. When we are doing 1-D slicing (e.g. `tensor[:]`).
581
+ // Reason: we always return a shallow copy for expressions such as
582
+ // `tensor[:]` / `tensor[...]` / `tensor[:, :]`. (Note that for `tensor[:,
583
+ // :]`, we return an alias of `tensor` by doing the following:
584
+ // ```
585
+ // Tensor sliced = impl::applySlicing(self, indices, tensorIndices,
586
+ // disable_slice_optimization, self_device, self_sizes); if
587
+ // (tensorIndices.empty()) {
588
+ // if (sliced.is_same(self)) {
589
+ // // ensure we return a shallow copy for things like x[...]
590
+ // sliced = at::alias(sliced);
591
+ // }
592
+ // return sliced;
593
+ // }
594
+ // ```)
595
+ // 2. When we are doing JIT tracing.
596
+ // Reason: JIT tracing needs the `self.slice(...)` call to properly trace the
597
+ // slice operation.
598
+
599
+ // This mirrors `THPVariable_getitem` in
600
+ // torch/csrc/autograd/python_variable_indexing.cpp See NOTE [ Setting
601
+ // `disable_slice_optimization` when calling C++ tensor indexing functions from
602
+ // Python ]
603
+ inline Tensor get_item(
604
+ const Tensor& self,
605
+ const ArrayRef<TensorIndex>& indices,
606
+ bool disable_slice_optimization = false) {
607
+ at::Device self_device = self.device();
608
+ // NOTE [nested tensor size for indexing]
609
+ // nested tensor does not have a size (yet) so for now we represent its size
610
+ // as null may need to be changed after we reach a better solution for nested
611
+ // tensor size
612
+ std::optional<SymIntArrayRef> self_sizes = self.is_nested()
613
+ ? std::optional<SymIntArrayRef>(std::nullopt)
614
+ : std::optional<SymIntArrayRef>(self.sym_sizes());
615
+
616
+ // handle simple types: integers, slices, none, ellipsis, bool
617
+ if (indices.size() == 1) {
618
+ const TensorIndex& index = indices[0];
619
+ if (index.is_integer()) {
620
+ return impl::applySelect(
621
+ self, 0, index.integer(), 0, self_device, self_sizes);
622
+ } else if (index.is_slice()) {
623
+ return impl::applySlice(
624
+ self,
625
+ 0,
626
+ index.slice().start(),
627
+ index.slice().stop(),
628
+ index.slice().step(),
629
+ /*disable_slice_optimization=*/true,
630
+ self_device,
631
+ self_sizes);
632
+ } else if (index.is_none()) {
633
+ return self.unsqueeze(0);
634
+ } else if (index.is_ellipsis()) {
635
+ return at::alias(self);
636
+ } else if (index.is_boolean()) {
637
+ Tensor result = self.unsqueeze(0);
638
+ return dispatch_index(
639
+ result,
640
+ std::vector<Tensor>{impl::boolToIndexingTensor(
641
+ result, index.boolean(), self_device)});
642
+ }
643
+ }
644
+
645
+ std::vector<Tensor> tensorIndices;
646
+ Tensor sliced = impl::applySlicing(
647
+ self,
648
+ indices,
649
+ tensorIndices,
650
+ disable_slice_optimization,
651
+ self_device,
652
+ self_sizes);
653
+ if (tensorIndices.empty()) {
654
+ if (sliced.is_same(self)) {
655
+ // ensure we return a shallow copy for things like x[...]
656
+ sliced = at::alias(sliced);
657
+ }
658
+ return sliced;
659
+ }
660
+
661
+ // indexing by tensors ("advanced" indexing)
662
+ return dispatch_index(sliced, std::move(tensorIndices));
663
+ }
664
+
665
+ // This mirrors `THPVariable_setitem` in
666
+ // torch/csrc/autograd/python_variable_indexing.cpp for "the assigned value is a
667
+ // Tensor" case See NOTE [ Setting `disable_slice_optimization` when calling C++
668
+ // tensor indexing functions from Python ]
669
+ inline void set_item(
670
+ const Tensor& self,
671
+ const ArrayRef<TensorIndex>& indices,
672
+ const Tensor& value,
673
+ bool disable_slice_optimization = false) {
674
+ at::Device self_device = self.device();
675
+ SymIntArrayRef self_sizes = self.sym_sizes();
676
+
677
+ // handle simple types: integers, slices, ellipsis, bool
678
+ if (indices.size() == 1) {
679
+ const TensorIndex& index = indices[0];
680
+ if (index.is_boolean() && !index.boolean()) {
681
+ // do nothing for false (technically we should check the size, but we
682
+ // don't have real 0-sized shapes.
683
+ return;
684
+ } else if (index.is_ellipsis()) {
685
+ copy_to(self, value);
686
+ return;
687
+ } else if (index.is_none() || (index.is_boolean() && index.boolean())) {
688
+ copy_to(self.unsqueeze(0), value);
689
+ return;
690
+ } else if (index.is_integer()) {
691
+ copy_to(
692
+ impl::applySelect(
693
+ self, 0, index.integer(), 0, self_device, self_sizes),
694
+ value);
695
+ return;
696
+ } else if (index.is_slice()) {
697
+ copy_to(
698
+ impl::applySlice(
699
+ self,
700
+ 0,
701
+ index.slice().start(),
702
+ index.slice().stop(),
703
+ index.slice().step(),
704
+ /*disable_slice_optimization=*/disable_slice_optimization,
705
+ self_device,
706
+ self_sizes),
707
+ value);
708
+ return;
709
+ }
710
+ }
711
+
712
+ std::vector<Tensor> tensorIndices;
713
+ Tensor sliced = impl::applySlicing(
714
+ self,
715
+ indices,
716
+ tensorIndices,
717
+ disable_slice_optimization,
718
+ self_device,
719
+ self_sizes);
720
+ if (tensorIndices.empty()) {
721
+ copy_to(sliced, value);
722
+ return;
723
+ }
724
+
725
+ SymIntArrayRef valueSizes = value.sym_sizes();
726
+ SymIntArrayRef slicedValueSizes = slicePrefix1sSize(valueSizes);
727
+ Tensor valuesSliced;
728
+ if (!valueSizes.equals(slicedValueSizes)) {
729
+ valuesSliced = value.view_symint(slicedValueSizes);
730
+ } else {
731
+ valuesSliced = value;
732
+ }
733
+ dispatch_index_put_(sliced, std::move(tensorIndices), valuesSliced);
734
+ return;
735
+ }
736
+
737
+ } // namespace at::indexing
.venv/lib/python3.11/site-packages/torch/include/ATen/TensorIteratorInternal.h ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+ #include <ATen/native/TensorIterator.h>
3
+ #include <c10/util/SmallBuffer.h>
4
+ #include <c10/util/irange.h>
5
+
6
+ namespace at {
7
+
8
+ struct DimCounter {
9
+ DimCounter(IntArrayRef shape, Range range);
10
+
11
+ void increment(const std::array<int64_t, 2>& step);
12
+ bool is_done() const;
13
+ std::array<int64_t, 2> max_2d_step() const;
14
+
15
+ IntArrayRef shape;
16
+ Range range;
17
+ c10::SmallBuffer<int64_t, 4> values;
18
+ int64_t offset;
19
+ };
20
+
21
+ namespace internal {
22
+
23
+ inline void get_data_ptrs(
24
+ char** ptrs,
25
+ ArrayRef<char*> base,
26
+ IntArrayRef strides,
27
+ IntArrayRef counter) {
28
+ const auto ntensors = base.size();
29
+ const auto ndim = counter.size();
30
+ std::copy(base.begin(), base.end(), ptrs);
31
+ for (const auto dim : c10::irange(ndim)) {
32
+ int64_t value = counter[dim];
33
+ for (const auto arg : c10::irange(ntensors)) {
34
+ ptrs[arg] += value * strides[dim * ntensors + arg];
35
+ }
36
+ }
37
+ }
38
+
39
+ inline void serial_for_each(
40
+ IntArrayRef shape,
41
+ IntArrayRef strides,
42
+ char** base_ptrs,
43
+ size_t ntensors,
44
+ typename TensorIteratorBase::loop2d_t loop,
45
+ Range range) {
46
+ const auto ndim = shape.size();
47
+ TORCH_INTERNAL_ASSERT_DEBUG_ONLY(
48
+ strides.size() == ntensors * std::max(size_t{2}, ndim));
49
+
50
+ if (ndim <= 1) {
51
+ if (range.begin == 0) {
52
+ loop(base_ptrs, strides.data(), range.size(), 1);
53
+ } else {
54
+ c10::SmallBuffer<char*, 4> ptrs(ntensors);
55
+ get_data_ptrs(ptrs.data(), {base_ptrs, ntensors}, strides, {range.begin});
56
+ loop(ptrs.data(), strides.data(), range.size(), 1);
57
+ }
58
+ } else {
59
+ c10::SmallBuffer<char*, 4> ptrs(ntensors);
60
+ auto counter = DimCounter(shape, range);
61
+ while (!counter.is_done()) {
62
+ get_data_ptrs(
63
+ ptrs.data(), {base_ptrs, ntensors}, strides, counter.values);
64
+ auto step = counter.max_2d_step();
65
+ loop(ptrs.data(), strides.data(), step[0], step[1]);
66
+ counter.increment(step);
67
+ }
68
+ }
69
+ }
70
+
71
+ } // namespace internal
72
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/TensorOptions.h ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ #pragma once
2
+ #include <c10/core/TensorOptions.h>
.venv/lib/python3.11/site-packages/torch/include/ATen/TensorSubclassLikeUtils.h ADDED
@@ -0,0 +1,88 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+ #include <ATen/core/List.h>
3
+ #include <ATen/core/Tensor.h>
4
+ #include <c10/core/impl/TorchDispatchModeTLS.h>
5
+
6
+ #ifndef AT_PER_OPERATOR_HEADERS
7
+ #include <ATen/Functions.h>
8
+ #else
9
+ #include <ATen/ops/equal.h>
10
+ #endif
11
+
12
+ namespace at {
13
+
14
+ // Note [Tensor-subclass-like Tensors]
15
+ // Tensor-subclass-like is defined as:
16
+ // - a Tensor subclass (via __torch_dispatch__ in Python or extending
17
+ // TensorImpl in C++)
18
+ // - anything else that shares the same perils as Tensor subclasses.
19
+ // For example, many Tensor subclasses do not have storage and meta Tensors
20
+ // do not have storage either, so meta Tensors belong here.
21
+ //
22
+ // We should ensure that PyTorch internals supports Tensor-subclass-like
23
+ // objects. In particular, Tensor-subclass-like objects struggle with two
24
+ // classes of operations that are problematic for Tensor subclasses:
25
+ // 1. Because some Tensor subclasses do not have storage, .item() or
26
+ // .data_ptr() calls are not good.
27
+ // 2. Certain in-place operations can eliminate the typing of the Tensor
28
+ // subclass. For example:
29
+ // >>> torch.zeros(input.sizes(), grad.options()).diag().copy_(input)
30
+ // If input is a Tensor subclass, then the above ends up either erroring out
31
+ // or returning a regular non-Tensor-subclass Tensor!
32
+
33
+ constexpr auto kFunctorchWrappedTensors = DispatchKeySet(
34
+ {DispatchKey::FuncTorchGradWrapper,
35
+ DispatchKey::FuncTorchBatched,
36
+ DispatchKey::Functionalize});
37
+
38
+ constexpr auto kTensorSubclassLike =
39
+ kFunctorchWrappedTensors |
40
+ DispatchKeySet(
41
+ {// WARNING: DO NOT put combined backend component + functionality keys
42
+ // here, you will incorrectly always match on the functionality key
43
+ // no matter the backend component
44
+ DispatchKey::Batched,
45
+ DispatchKey::Sparse,
46
+ DispatchKey::SparseCsr,
47
+ DispatchKey::Python}) |
48
+ DispatchKeySet(BackendComponent::MetaBit);
49
+
50
+ inline bool isTensorSubclassLike(const Tensor& tensor) {
51
+ if (c10::impl::dispatch_mode_enabled())
52
+ return true;
53
+ auto key_set = tensor.unsafeGetTensorImpl()->key_set();
54
+ return !(key_set & kTensorSubclassLike).empty();
55
+ }
56
+
57
+ inline bool areAnyTensorSubclassLike(TensorList tensors) {
58
+ if (c10::impl::dispatch_mode_enabled())
59
+ return true;
60
+ return std::any_of(tensors.begin(), tensors.end(), isTensorSubclassLike);
61
+ }
62
+
63
+ inline bool areAnyOptionalTensorSubclassLike(
64
+ const c10::List<std::optional<Tensor>>& tensors) {
65
+ if (c10::impl::dispatch_mode_enabled())
66
+ return true;
67
+ return std::any_of(
68
+ tensors.begin(),
69
+ tensors.end(),
70
+ [](const std::optional<Tensor>& opt_tensor) {
71
+ return (
72
+ opt_tensor.has_value() && isTensorSubclassLike(opt_tensor.value()));
73
+ });
74
+ }
75
+
76
+ // Helper function to deal testing truthfulness of a scalar tensor
77
+ // in a Composite Compliant manner.
78
+ // NOTE: This function expects a scalar tensor of boolean dtype.
79
+ // Eg.
80
+ // Non-Composite Compliant Pattern : (t == 0).all().item<bool>()
81
+ // Composite Compliant Patter : is_salar_tensor_true((t == 0).all())
82
+ inline bool is_scalar_tensor_true(const Tensor& t) {
83
+ TORCH_INTERNAL_ASSERT(t.dim() == 0)
84
+ TORCH_INTERNAL_ASSERT(t.scalar_type() == kBool)
85
+ return at::equal(t, t.new_ones({}, t.options()));
86
+ }
87
+
88
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/TensorUtils.h ADDED
@@ -0,0 +1,190 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <ATen/DimVector.h>
4
+ #include <ATen/EmptyTensor.h>
5
+ #include <ATen/Tensor.h>
6
+ #include <ATen/TensorGeometry.h>
7
+ #include <ATen/Utils.h>
8
+
9
+ #include <utility>
10
+
11
+ // These functions are NOT in Utils.h, because this file has a dep on Tensor.h
12
+
13
+ #define TORCH_CHECK_TENSOR_ALL(cond, ...) \
14
+ TORCH_CHECK((cond)._is_all_true().item<bool>(), __VA_ARGS__);
15
+
16
+ namespace at {
17
+
18
+ // The following are utility functions for checking that arguments
19
+ // make sense. These are particularly useful for native functions,
20
+ // which do NO argument checking by default.
21
+
22
+ struct TORCH_API TensorArg {
23
+ // NOLINTNEXTLINE(cppcoreguidelines-avoid-const-or-ref-data-members)
24
+ const Tensor& tensor;
25
+ const char* name;
26
+ int pos; // 1-indexed
27
+ TensorArg(const Tensor& tensor, const char* name, int pos)
28
+ : tensor(tensor), name(name), pos(pos) {}
29
+ // Try to mitigate any possibility of dangling reference to temporaries.
30
+ // NOLINTNEXTLINE(cppcoreguidelines-rvalue-reference-param-not-moved)
31
+ TensorArg(Tensor&& tensor, const char* name, int pos) = delete;
32
+ const Tensor* operator->() const {
33
+ return &tensor;
34
+ }
35
+ const Tensor& operator*() const {
36
+ return tensor;
37
+ }
38
+ };
39
+
40
+ struct TORCH_API TensorGeometryArg {
41
+ TensorGeometry tensor;
42
+ const char* name;
43
+ int pos; // 1-indexed
44
+ /* implicit */ TensorGeometryArg(TensorArg arg)
45
+ : tensor(TensorGeometry{arg.tensor}), name(arg.name), pos(arg.pos) {}
46
+ TensorGeometryArg(TensorGeometry tensor, const char* name, int pos)
47
+ : tensor(std::move(tensor)), name(name), pos(pos) {}
48
+ const TensorGeometry* operator->() const {
49
+ return &tensor;
50
+ }
51
+ const TensorGeometry& operator*() const {
52
+ return tensor;
53
+ }
54
+ };
55
+
56
+ // A string describing which function did checks on its input
57
+ // arguments.
58
+ // TODO: Consider generalizing this into a call stack.
59
+ using CheckedFrom = const char*;
60
+
61
+ // The undefined convention: singular operators assume their arguments
62
+ // are defined, but functions which take multiple tensors will
63
+ // implicitly filter out undefined tensors (to make it easier to perform
64
+ // tests which should apply if the tensor is defined, and should not
65
+ // otherwise.)
66
+ //
67
+ // NB: This means that the n-ary operators take lists of TensorArg,
68
+ // not TensorGeometryArg, because the Tensor to TensorGeometry
69
+ // conversion will blow up if you have undefined tensors.
70
+
71
+ TORCH_API std::ostream& operator<<(
72
+ std::ostream& out,
73
+ const TensorGeometryArg& t);
74
+ TORCH_API void checkDim(
75
+ CheckedFrom c,
76
+ const Tensor& tensor,
77
+ const char* name,
78
+ int pos, // 1-indexed
79
+ int64_t dim);
80
+ TORCH_API void checkDim(CheckedFrom c, const TensorGeometryArg& t, int64_t dim);
81
+ // NB: this is an inclusive-exclusive range
82
+ TORCH_API void checkDimRange(
83
+ CheckedFrom c,
84
+ const TensorGeometryArg& t,
85
+ int64_t dim_start,
86
+ int64_t dim_end);
87
+ TORCH_API void checkSameDim(
88
+ CheckedFrom c,
89
+ const TensorGeometryArg& t1,
90
+ const TensorGeometryArg& t2);
91
+ TORCH_API void checkContiguous(CheckedFrom c, const TensorGeometryArg& t);
92
+ TORCH_API void checkAllContiguous(CheckedFrom c, at::ArrayRef<TensorArg> ts);
93
+ TORCH_API void checkSize(
94
+ CheckedFrom c,
95
+ const TensorGeometryArg& t,
96
+ IntArrayRef sizes);
97
+ TORCH_API void checkSize_symint(
98
+ CheckedFrom c,
99
+ const TensorGeometryArg& t,
100
+ c10::SymIntArrayRef sizes);
101
+ TORCH_API void checkSize(
102
+ CheckedFrom c,
103
+ const TensorGeometryArg& t,
104
+ int64_t dim,
105
+ int64_t size);
106
+ TORCH_API void checkSize_symint(
107
+ CheckedFrom c,
108
+ const TensorGeometryArg& t,
109
+ int64_t dim,
110
+ const c10::SymInt& size);
111
+ TORCH_API void checkNumel(
112
+ CheckedFrom c,
113
+ const TensorGeometryArg& t,
114
+ int64_t numel);
115
+ TORCH_API void checkSameNumel(
116
+ CheckedFrom c,
117
+ const TensorArg& t1,
118
+ const TensorArg& t2);
119
+ TORCH_API void checkAllSameNumel(CheckedFrom c, ArrayRef<TensorArg> tensors);
120
+ TORCH_API void checkScalarType(CheckedFrom c, const TensorArg& t, ScalarType s);
121
+ TORCH_API void checkScalarTypes(
122
+ CheckedFrom c,
123
+ const TensorArg& t,
124
+ at::ArrayRef<ScalarType> l);
125
+ TORCH_API void checkSameGPU(
126
+ CheckedFrom c,
127
+ const TensorArg& t1,
128
+ const TensorArg& t2);
129
+ TORCH_API void checkAllSameGPU(CheckedFrom c, ArrayRef<TensorArg> tensors);
130
+ TORCH_API void checkSameType(
131
+ CheckedFrom c,
132
+ const TensorArg& t1,
133
+ const TensorArg& t2);
134
+ TORCH_API void checkAllSameType(CheckedFrom c, ArrayRef<TensorArg> tensors);
135
+ TORCH_API void checkSameSize(
136
+ CheckedFrom c,
137
+ const TensorArg& t1,
138
+ const TensorArg& t2);
139
+ TORCH_API void checkAllSameSize(CheckedFrom c, ArrayRef<TensorArg> tensors);
140
+ TORCH_API void checkDefined(CheckedFrom c, const TensorArg& t);
141
+ TORCH_API void checkAllDefined(CheckedFrom c, at::ArrayRef<TensorArg> t);
142
+
143
+ // FixMe: does TensorArg slow things down?
144
+ TORCH_API void checkBackend(
145
+ CheckedFrom c,
146
+ at::ArrayRef<Tensor> t,
147
+ at::Backend backend);
148
+
149
+ TORCH_API void checkDeviceType(
150
+ CheckedFrom c,
151
+ at::ArrayRef<Tensor> tensors,
152
+ at::DeviceType device_type);
153
+
154
+ TORCH_API void checkLayout(CheckedFrom c, const Tensor& t, Layout layout);
155
+
156
+ TORCH_API void checkLayout(
157
+ CheckedFrom c,
158
+ at::ArrayRef<Tensor> tensors,
159
+ at::Layout layout);
160
+
161
+ // Methods for getting data_ptr if tensor is defined
162
+ TORCH_API void* maybe_data_ptr(const Tensor& tensor);
163
+ TORCH_API void* maybe_data_ptr(const TensorArg& tensor);
164
+
165
+ TORCH_API void check_dim_size(
166
+ const Tensor& tensor,
167
+ int64_t dim,
168
+ int64_t dim_size,
169
+ int64_t size);
170
+
171
+ namespace detail {
172
+ TORCH_API std::vector<int64_t> defaultStrides(IntArrayRef sizes);
173
+
174
+ TORCH_API std::optional<std::vector<int64_t>> computeStride(
175
+ IntArrayRef oldshape,
176
+ IntArrayRef oldstride,
177
+ IntArrayRef newshape);
178
+
179
+ TORCH_API std::optional<SymDimVector> computeStride(
180
+ c10::SymIntArrayRef oldshape,
181
+ c10::SymIntArrayRef oldstride,
182
+ c10::SymIntArrayRef newshape);
183
+
184
+ TORCH_API std::optional<DimVector> computeStride(
185
+ IntArrayRef oldshape,
186
+ IntArrayRef oldstride,
187
+ const DimVector& newshape);
188
+
189
+ } // namespace detail
190
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/TypeDefault.h ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <ATen/Dimname.h>
4
+ #include <c10/core/MemoryFormat.h>
5
+ #include <c10/core/QScheme.h>
6
+ #include <c10/core/Scalar.h>
7
+ #include <c10/core/TensorOptions.h>
8
+ #include <c10/macros/Export.h>
9
+ #include <c10/util/ArrayRef.h>
10
+ #include <c10/util/intrusive_ptr.h>
11
+
12
+ namespace c10 {
13
+ struct Storage;
14
+ }
15
+
16
+ namespace at {
17
+
18
+ class Tensor;
19
+ using TensorList = ArrayRef<Tensor>;
20
+
21
+ class Context;
22
+ struct Generator;
23
+
24
+ struct Quantizer;
25
+ // This is temporary typedef to enable Quantizer in aten native function API
26
+ // we'll remove them when we are actually exposing Quantizer class
27
+ // to frontend
28
+ using ConstQuantizerPtr = const c10::intrusive_ptr<Quantizer>&;
29
+
30
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/Utils.h ADDED
@@ -0,0 +1,134 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <ATen/EmptyTensor.h>
4
+ #include <ATen/Formatting.h>
5
+ #include <ATen/core/ATenGeneral.h>
6
+ #include <ATen/core/Generator.h>
7
+ #include <c10/core/ScalarType.h>
8
+ #include <c10/core/StorageImpl.h>
9
+ #include <c10/core/UndefinedTensorImpl.h>
10
+ #include <c10/util/ArrayRef.h>
11
+ #include <c10/util/Exception.h>
12
+ #include <c10/util/accumulate.h>
13
+ #include <c10/util/irange.h>
14
+
15
+ #include <algorithm>
16
+
17
+ #define AT_DISALLOW_COPY_AND_ASSIGN(TypeName) \
18
+ TypeName(const TypeName&) = delete; \
19
+ void operator=(const TypeName&) = delete
20
+
21
+ namespace at {
22
+
23
+ TORCH_API int _crash_if_asan(int);
24
+
25
+ // Converts a TensorList (i.e. ArrayRef<Tensor> to vector of TensorImpl*)
26
+ // NB: This is ONLY used by legacy TH bindings, and ONLY used by cat.
27
+ // Once cat is ported entirely to ATen this can be deleted!
28
+ inline std::vector<TensorImpl*> checked_dense_tensor_list_unwrap(
29
+ ArrayRef<Tensor> tensors,
30
+ const char* name,
31
+ int pos,
32
+ c10::DeviceType device_type,
33
+ ScalarType scalar_type) {
34
+ std::vector<TensorImpl*> unwrapped;
35
+ unwrapped.reserve(tensors.size());
36
+ for (const auto i : c10::irange(tensors.size())) {
37
+ const auto& expr = tensors[i];
38
+ if (expr.layout() != Layout::Strided) {
39
+ AT_ERROR(
40
+ "Expected dense tensor but got ",
41
+ expr.layout(),
42
+ " for sequence element ",
43
+ i,
44
+ " in sequence argument at position #",
45
+ pos,
46
+ " '",
47
+ name,
48
+ "'");
49
+ }
50
+ if (expr.device().type() != device_type) {
51
+ AT_ERROR(
52
+ "Expected object of device type ",
53
+ device_type,
54
+ " but got device type ",
55
+ expr.device().type(),
56
+ " for sequence element ",
57
+ i,
58
+ " in sequence argument at position #",
59
+ pos,
60
+ " '",
61
+ name,
62
+ "'");
63
+ }
64
+ if (expr.scalar_type() != scalar_type) {
65
+ AT_ERROR(
66
+ "Expected object of scalar type ",
67
+ scalar_type,
68
+ " but got scalar type ",
69
+ expr.scalar_type(),
70
+ " for sequence element ",
71
+ i,
72
+ " in sequence argument at position #",
73
+ pos,
74
+ " '",
75
+ name,
76
+ "'");
77
+ }
78
+ unwrapped.emplace_back(expr.unsafeGetTensorImpl());
79
+ }
80
+ return unwrapped;
81
+ }
82
+
83
+ template <size_t N>
84
+ std::array<int64_t, N> check_intlist(
85
+ ArrayRef<int64_t> list,
86
+ const char* name,
87
+ int pos) {
88
+ if (list.empty()) {
89
+ // TODO: is this necessary? We used to treat nullptr-vs-not in IntList
90
+ // differently with strides as a way of faking optional.
91
+ list = {};
92
+ }
93
+ auto res = std::array<int64_t, N>();
94
+ if (list.size() == 1 && N > 1) {
95
+ res.fill(list[0]);
96
+ return res;
97
+ }
98
+ if (list.size() != N) {
99
+ AT_ERROR(
100
+ "Expected a list of ",
101
+ N,
102
+ " ints but got ",
103
+ list.size(),
104
+ " for argument #",
105
+ pos,
106
+ " '",
107
+ name,
108
+ "'");
109
+ }
110
+ std::copy_n(list.begin(), N, res.begin());
111
+ return res;
112
+ }
113
+
114
+ using at::detail::check_size_nonnegative;
115
+
116
+ namespace detail {
117
+
118
+ template <typename T>
119
+ TORCH_API Tensor tensor_cpu(ArrayRef<T> values, const TensorOptions& options);
120
+
121
+ template <typename T>
122
+ TORCH_API Tensor
123
+ tensor_backend(ArrayRef<T> values, const TensorOptions& options);
124
+
125
+ template <typename T>
126
+ TORCH_API Tensor
127
+ tensor_complex_cpu(ArrayRef<T> values, const TensorOptions& options);
128
+
129
+ template <typename T>
130
+ TORCH_API Tensor
131
+ tensor_complex_backend(ArrayRef<T> values, const TensorOptions& options);
132
+ } // namespace detail
133
+
134
+ } // namespace at
.venv/lib/python3.11/site-packages/torch/include/ATen/VmapGeneratedPlumbing.h ADDED
The diff for this file is too large to render. See raw diff
 
.venv/lib/python3.11/site-packages/torch/include/ATen/cpp_custom_type_hack.h ADDED
@@ -0,0 +1,110 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
2
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
3
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
4
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
5
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
6
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
7
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
8
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
9
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
10
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
11
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
12
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
13
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
14
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
15
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
16
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
17
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
18
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
19
+
20
+ // YOU ARE IN THE WRONG PLACE! TURN BACK NOW!
21
+
22
+ // This code was a temporary hack to enable embedding arbitrary C++ structures
23
+ // into Tensors. THIS IS UNSAFE AND IS NOT SUPPORTED. IF YOU USE THIS CODE,
24
+ // IT __WILL__ BREAK.
25
+
26
+ // This code has been superseded by custom classes:
27
+ // https://pytorch.org/tutorials/advanced/torch_script_custom_classes.html
28
+
29
+ // Please use custom classes and **DO NOT ADD MORE CALLSITES TO THINGS DEFINED
30
+ // IN THIS FILE**.
31
+
32
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
33
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
34
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
35
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
36
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
37
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
38
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
39
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
40
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
41
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
42
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
43
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
44
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
45
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
46
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
47
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
48
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
49
+ // STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP STOP
50
+
51
+ #include <ATen/TracerMode.h>
52
+ #include <ATen/core/Tensor.h>
53
+
54
+ #ifndef AT_PER_OPERATOR_HEADERS
55
+ #include <ATen/Functions.h>
56
+ #else
57
+ #include <ATen/ops/empty.h>
58
+ #endif
59
+
60
+ namespace at::cpp_custom_type_hack {
61
+
62
+ template <typename T>
63
+ [[deprecated(
64
+ "Use custom classes instead: "
65
+ "https://pytorch.org/tutorials/advanced/torch_script_custom_classes.html")]] bool
66
+ isa(const Tensor& packed) {
67
+ return (packed.scalar_type() == kByte) &&
68
+ (packed.storage().data_ptr().get_deleter() ==
69
+ caffe2::TypeMeta::Make<T>().deleteFn());
70
+ }
71
+
72
+ template <typename T>
73
+ [[deprecated(
74
+ "Use custom classes instead: "
75
+ "https://pytorch.org/tutorials/advanced/torch_script_custom_classes.html")]] T&
76
+ cast(const Tensor& packed) {
77
+ TORCH_CHECK(
78
+ packed.scalar_type() == kByte, "Expected temporary cpp type wrapper");
79
+ TORCH_CHECK(
80
+ packed.storage().data_ptr().get_deleter() ==
81
+ caffe2::TypeMeta::Make<T>().deleteFn(),
82
+ "Expected temporary cpp type wrapper of type ",
83
+ caffe2::TypeMeta::TypeName<T>());
84
+ return *reinterpret_cast<T*>(packed.storage().data_ptr().get());
85
+ }
86
+
87
+ template <typename T>
88
+ [[deprecated(
89
+ "Use custom classes instead: "
90
+ "https://pytorch.org/tutorials/advanced/torch_script_custom_classes.html")]] Tensor
91
+ create(std::unique_ptr<T> ptr, TensorOptions options) {
92
+ // None of this should trace, so turn off Tracer dispatching
93
+ at::AutoDispatchBelowADInplaceOrView guard; // TODO: remove
94
+ at::tracer::impl::NoTracerDispatchMode tracer_guard;
95
+
96
+ // We store this instance away in a Tensor and register a deleter function
97
+ // so that we do not leak memory. On the other side, we pull out the storage's
98
+ // data_ptr and get the right typed pointer.
99
+ void* raw_ptr = ptr.release();
100
+ at::DataPtr at_ptr(
101
+ raw_ptr, raw_ptr, caffe2::TypeMeta::Make<T>().deleteFn(), at::kCPU);
102
+
103
+ // size doesn't really matter, but we can align it to the actual size
104
+ // returning variables because one likely want to use this hack from python
105
+ auto retval = at::empty({sizeof(T)}, options.device(kCPU).dtype(at::kByte));
106
+ retval.storage().set_data_ptr_noswap(std::move(at_ptr));
107
+ return retval;
108
+ }
109
+
110
+ } // namespace at::cpp_custom_type_hack
.venv/lib/python3.11/site-packages/torch/include/ATen/cuda/ATenCUDAGeneral.h ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <cuda.h>
4
+ #include <cuda_runtime.h>
5
+ #include <cuda_fp16.h>
6
+
7
+ #include <c10/macros/Export.h>
8
+
9
+ // Use TORCH_CUDA_CPP_API or TORCH_CUDA_CU_API for exports from this folder
.venv/lib/python3.11/site-packages/torch/include/ATen/cuda/ApplyGridUtils.cuh ADDED
@@ -0,0 +1,47 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #include <ATen/cuda/CUDAContext.h>
2
+
3
+ #include <cuda_runtime.h>
4
+
5
+ namespace at::cuda {
6
+
7
+ /**
8
+ Computes ceil(a / b)
9
+ */
10
+ template <typename T>
11
+ __host__ __device__ __forceinline__ T ATenCeilDiv(T a, T b) {
12
+ return (a + b - 1) / b;
13
+ }
14
+
15
+ namespace {
16
+
17
+ // Threads per block for our apply kernel
18
+ // FIXME: use occupancy calculator instead
19
+ constexpr uint32_t AT_APPLY_THREADS_PER_BLOCK = 512;
20
+ constexpr uint32_t AT_APPLY_BLOCKS_PER_SM = 4;
21
+
22
+ template <int step = 1>
23
+ inline bool getApplyGrid(uint64_t totalElements, dim3& grid, c10::DeviceIndex curDevice, int max_threads_per_block=AT_APPLY_THREADS_PER_BLOCK) {
24
+ if (curDevice == -1) return false;
25
+ uint64_t numel_per_thread = static_cast<uint64_t>(max_threads_per_block) * static_cast<uint64_t>(step);
26
+ uint64_t numBlocks = ATenCeilDiv(totalElements, numel_per_thread);
27
+ uint64_t maxGridX = at::cuda::getDeviceProperties(curDevice)->maxGridSize[0];
28
+ if (numBlocks > maxGridX)
29
+ numBlocks = maxGridX;
30
+ grid = dim3(numBlocks);
31
+ return true;
32
+ }
33
+
34
+ constexpr int getApplyBlocksPerSM() {
35
+ return AT_APPLY_BLOCKS_PER_SM;
36
+ }
37
+
38
+ constexpr int getApplyBlockSize() {
39
+ return AT_APPLY_THREADS_PER_BLOCK;
40
+ }
41
+
42
+ inline dim3 getApplyBlock(int max_threads_per_block=AT_APPLY_THREADS_PER_BLOCK) {
43
+ return dim3(max_threads_per_block);
44
+ }
45
+
46
+ } // anonymous namespace
47
+ } // namespace at::cuda
.venv/lib/python3.11/site-packages/torch/include/ATen/cuda/Atomic.cuh ADDED
@@ -0,0 +1,514 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <cuda.h>
4
+ #include <c10/util/Half.h>
5
+ #include <c10/util/BFloat16.h>
6
+
7
+ #include <ATen/NumericUtils.h>
8
+
9
+ #if !(defined(USE_ROCM) || ((defined(__CUDA_ARCH__) && (__CUDA_ARCH__ < 800))))
10
+ #include <cuda_bf16.h>
11
+ #endif
12
+
13
+ template <typename T>
14
+ struct AtomicFPOp;
15
+
16
+ template <>
17
+ struct AtomicFPOp<at::Half> {
18
+ template <typename func_t>
19
+ inline __device__ at::Half operator() (at::Half *address, at::Half val, const func_t& func) {
20
+ unsigned int * address_as_ui =
21
+ (unsigned int *) ((char *)address - ((size_t)address & 2));
22
+ unsigned int old = *address_as_ui;
23
+ unsigned int assumed;
24
+
25
+ at::Half hsum;
26
+ do {
27
+ assumed = old;
28
+ hsum.x = (size_t)address & 2 ? (old >> 16) : (old & 0xffff);
29
+ hsum = func(hsum, val);
30
+ old = (size_t)address & 2 ? (old & 0xffff) | (hsum.x << 16) : (old & 0xffff0000) | hsum.x;
31
+ old = atomicCAS(address_as_ui, assumed, old);
32
+ } while (assumed != old);
33
+ hsum.x = (size_t)address & 2 ? (old >> 16) : (old & 0xffff);
34
+ return hsum;
35
+ }
36
+ };
37
+
38
+ template <>
39
+ struct AtomicFPOp<at::BFloat16> {
40
+ template <typename func_t>
41
+ inline __device__ at::BFloat16 operator() (at::BFloat16 *address, at::BFloat16 val, const func_t& func) {
42
+ unsigned int * address_as_ui =
43
+ (unsigned int *) ((char *)address - ((size_t)address & 2));
44
+ unsigned int old = *address_as_ui;
45
+ unsigned int assumed;
46
+
47
+ at::BFloat16 bsum;
48
+ do {
49
+ assumed = old;
50
+ bsum.x = (size_t)address & 2 ? (old >> 16) : (old & 0xffff);
51
+ bsum = func(bsum, val);
52
+ old = (size_t)address & 2 ? (old & 0xffff) | (bsum.x << 16) : (old & 0xffff0000) | bsum.x;
53
+ old = atomicCAS(address_as_ui, assumed, old);
54
+ } while (assumed != old);
55
+ bsum.x = (size_t)address & 2 ? (old >> 16) : (old & 0xffff);
56
+ return bsum.x;
57
+ }
58
+ };
59
+
60
+ template <>
61
+ struct AtomicFPOp<double> {
62
+ template <typename func_t>
63
+ inline __device__ double operator() (double * address, double val, const func_t& func) {
64
+ unsigned long long int* address_as_ull = (unsigned long long int*)address;
65
+ unsigned long long int old = *address_as_ull;
66
+ unsigned long long int assumed;
67
+
68
+ do {
69
+ assumed = old;
70
+ old = atomicCAS(address_as_ull, assumed, func(val, assumed));
71
+ // Note: uses integer comparison to avoid hang in case of NaN (since NaN != NaN)
72
+ } while (assumed != old);
73
+
74
+ return __longlong_as_double(old);
75
+ }
76
+ };
77
+
78
+ #define ATOMIC_INTEGER_IMPL(NAME) \
79
+ template <typename T, size_t n> \
80
+ struct Atomic##NAME##IntegerImpl; \
81
+ \
82
+ template<typename T> \
83
+ struct Atomic##NAME##IntegerImpl<T, 1> { \
84
+ template <typename func_t> \
85
+ inline __device__ void operator()(T *address, T val, const func_t& func) { \
86
+ size_t offset = (size_t)address & 3; \
87
+ uint32_t * address_as_ui = (uint32_t *)((char *)address - offset); \
88
+ uint32_t old = *address_as_ui; \
89
+ uint32_t shift = offset * 8; \
90
+ uint32_t old_byte; \
91
+ uint32_t newval; \
92
+ uint32_t assumed; \
93
+ \
94
+ do { \
95
+ assumed = old; \
96
+ old_byte = (old >> shift) & 0xff; \
97
+ newval = static_cast<uint8_t>(func(val, static_cast<T>(old_byte))); \
98
+ newval = (old & ~(0x000000ff << shift)) | (newval << shift); \
99
+ old = atomicCAS(address_as_ui, assumed, newval); \
100
+ } while (assumed != old); \
101
+ } \
102
+ }; \
103
+ \
104
+ template<typename T> \
105
+ struct Atomic##NAME##IntegerImpl<T, 2> { \
106
+ template <typename func_t> \
107
+ inline __device__ void operator()(T *address, T val, const func_t& func) { \
108
+ size_t offset = (size_t)address & 2; \
109
+ uint32_t * address_as_ui = (uint32_t *)((char *)address - offset); \
110
+ bool is_32_align = offset; \
111
+ uint32_t old = *address_as_ui; \
112
+ uint32_t old_bytes; \
113
+ uint32_t newval; \
114
+ uint32_t assumed; \
115
+ \
116
+ do { \
117
+ assumed = old; \
118
+ old_bytes = is_32_align ? old >> 16 : old & 0xffff; \
119
+ newval = static_cast<uint16_t>(func(val, static_cast<T>(old_bytes))); \
120
+ newval = is_32_align ? (old & 0xffff) | (newval << 16) : (old & 0xffff0000) | newval; \
121
+ old = atomicCAS(address_as_ui, assumed, newval); \
122
+ } while (assumed != old); \
123
+ } \
124
+ }; \
125
+ \
126
+ template<typename T> \
127
+ struct Atomic##NAME##IntegerImpl<T, 4> { \
128
+ template <typename func_t> \
129
+ inline __device__ void operator()(T *address, T val, const func_t& func) { \
130
+ uint32_t * address_as_ui = (uint32_t *) (address); \
131
+ uint32_t old = *address_as_ui; \
132
+ uint32_t newval; \
133
+ uint32_t assumed; \
134
+ \
135
+ do { \
136
+ assumed = old; \
137
+ newval = static_cast<uint32_t>(func(val, static_cast<T>(old))); \
138
+ old = atomicCAS(address_as_ui, assumed, newval); \
139
+ } while (assumed != old); \
140
+ } \
141
+ }; \
142
+ \
143
+ template<typename T> \
144
+ struct Atomic##NAME##IntegerImpl<T, 8> { \
145
+ template <typename func_t> \
146
+ inline __device__ void operator()(T *address, T val, const func_t& func) { \
147
+ unsigned long long * address_as_ui = (unsigned long long *) (address); \
148
+ unsigned long long old = *address_as_ui; \
149
+ unsigned long long newval; \
150
+ unsigned long long assumed; \
151
+ \
152
+ do { \
153
+ assumed = old; \
154
+ newval = static_cast<uint64_t>(func(val, static_cast<T>(old))); \
155
+ old = atomicCAS(address_as_ui, assumed, newval); \
156
+ } while (assumed != old); \
157
+ } \
158
+ };
159
+
160
+
161
+ # define GPU_ATOMIC_INTEGER(NAME, OP, DTYPE) \
162
+ inline __device__ void gpuAtomic##NAME(DTYPE *address, DTYPE val) { \
163
+ Atomic##NAME##IntegerImpl<DTYPE, sizeof(DTYPE)>()(address, \
164
+ val, \
165
+ [](DTYPE a, DTYPE b) { \
166
+ return OP; \
167
+ }); \
168
+ } \
169
+
170
+ ATOMIC_INTEGER_IMPL(Add)
171
+ GPU_ATOMIC_INTEGER(Add, a || b, bool)
172
+
173
+ // Don't instantiate gpuAtomicAdd with the macro as it seems non-standard (see int32, int64)
174
+ inline __device__ void gpuAtomicAdd(uint8_t *address, uint8_t val) {
175
+ AtomicAddIntegerImpl<uint8_t, sizeof(uint8_t)>()(address,
176
+ val,
177
+ [](uint8_t a, uint8_t b) {
178
+ return a + b;
179
+ });
180
+ }
181
+
182
+ inline __device__ void gpuAtomicAdd(int8_t *address, int8_t val) {
183
+ AtomicAddIntegerImpl<int8_t, sizeof(int8_t)>()(address,
184
+ val,
185
+ [](int8_t a, int8_t b) {
186
+ return a + b;
187
+ });
188
+ }
189
+
190
+ inline __device__ void gpuAtomicAdd(int16_t *address, int16_t val) {
191
+ AtomicAddIntegerImpl<int16_t, sizeof(int16_t)>()(address,
192
+ val,
193
+ [](int16_t a, int16_t b) {
194
+ return a + b;
195
+ });
196
+ }
197
+
198
+ inline __device__ int32_t gpuAtomicAdd(int32_t *address, int32_t val) {
199
+ return atomicAdd(address, val);
200
+ }
201
+
202
+ inline __device__ void gpuAtomicAdd(int64_t *address, int64_t val) {
203
+ #if defined(USE_ROCM)
204
+ __atomic_fetch_add(address, val, __ATOMIC_RELAXED);
205
+ #else
206
+ static_assert(sizeof(unsigned long long int) == sizeof(int64_t), "bitwidth change is not allowed");
207
+ atomicAdd(reinterpret_cast<unsigned long long int *>(address), static_cast<unsigned long long int>(val));
208
+ #endif
209
+ }
210
+
211
+ inline __device__ at::Half gpuAtomicAdd(at::Half *address, at::Half val) {
212
+ #if defined(USE_ROCM) || ((defined(__CUDA_ARCH__) && (__CUDA_ARCH__ < 700)))
213
+ return AtomicFPOp<at::Half>()(address, val,
214
+ [](at::Half hsum, at::Half val) {
215
+ return hsum + val;
216
+ });
217
+ #else
218
+ return atomicAdd(reinterpret_cast<__half*>(address), val);
219
+ #endif
220
+ }
221
+
222
+ inline __device__ at::BFloat16 gpuAtomicAdd(at::BFloat16 *address, at::BFloat16 val) {
223
+ #if defined(USE_ROCM) || ((defined(__CUDA_ARCH__) && (__CUDA_ARCH__ < 800)))
224
+ return AtomicFPOp<at::BFloat16>()(address, val,
225
+ [](at::BFloat16 bsum, at::BFloat16 val) {
226
+ return bsum + val;
227
+ });
228
+ #else
229
+ __nv_bfloat16 r = atomicAdd(reinterpret_cast<__nv_bfloat16*>(address), *reinterpret_cast<__nv_bfloat16*>(&val));
230
+ return *reinterpret_cast<c10::BFloat16*>(&r);
231
+ #endif
232
+ }
233
+
234
+ #if defined(__CUDA_ARCH__) && (__CUDA_ARCH__ < 600)
235
+ // from CUDA C Programmic Guide
236
+ inline __device__ double atomicAdd(double* address, double val)
237
+ #if defined(__clang__) && defined(__CUDA__)
238
+ #pragma GCC diagnostic push
239
+ #pragma GCC diagnostic ignored "-Wgcc-compat"
240
+ __attribute__((enable_if(true, "")))
241
+ #pragma GCC diagnostic pop
242
+ #endif
243
+ {
244
+
245
+ return AtomicFPOp<double>()(address, val,
246
+ [](double val, unsigned long long int assumed) {
247
+ return __double_as_longlong(val + __longlong_as_double(assumed));
248
+ });
249
+ }
250
+ #elif defined(USE_ROCM) || !(defined(__CUDA_ARCH__))
251
+
252
+ /* Note [hip-clang differences to hcc]
253
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
254
+ * The upcoming hip-clang compiler for ROCm differs from hcc in a few details.
255
+ * It exports the __HIP__ macro, we can hence differentiate between hcc and
256
+ * hip-clang. In the below, hcc only received support for atomicAdd with double
257
+ * typing after work week 18312. hip-clang had support from the first version.
258
+ * In general, the code-visible differences between hip-clang and hcc will be
259
+ * minimal.
260
+ */
261
+
262
+ #if defined(USE_ROCM) && __hcc_workweek__ < 18312 && !__HIP__
263
+ // This needs to be defined for the host side pass
264
+ inline __device__ double atomicAdd(double *address, double val) { }
265
+ #endif
266
+ #endif
267
+
268
+ inline __device__ double gpuAtomicAdd(double *address, double val) {
269
+ return atomicAdd(address, val);
270
+ }
271
+
272
+ inline __device__ float gpuAtomicAdd(float *address, float val) {
273
+ return atomicAdd(address, val);
274
+ }
275
+
276
+ template<typename T>
277
+ inline __device__ void gpuAtomicAdd(c10::complex<T> *address, c10::complex<T> val) {
278
+ gpuAtomicAdd(&address->real_, val.real_);
279
+ gpuAtomicAdd(&address->imag_, val.imag_);
280
+ }
281
+
282
+ /* Note [gpuAtomicAdd vs atomicAdd]
283
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
284
+ * Some extensions such as torchvision call atomicAdd()
285
+ * directly and require non-library provided data type support. Only for these, we
286
+ * continue to provide atomicAdd overloads.
287
+ */
288
+ inline __device__ at::Half atomicAdd(at::Half *address, at::Half val) {
289
+ return gpuAtomicAdd(address, val);
290
+ }
291
+
292
+ inline __device__ at::BFloat16 atomicAdd(at::BFloat16 *address, at::BFloat16 val) {
293
+ return gpuAtomicAdd(address, val);
294
+ }
295
+
296
+ inline __device__ void atomicAdd(uint8_t *address, uint8_t val) {
297
+ gpuAtomicAdd(address, val);
298
+ }
299
+
300
+ inline __device__ void atomicAdd(int8_t *address, int8_t val) {
301
+ gpuAtomicAdd(address, val);
302
+ }
303
+
304
+ inline __device__ void atomicAdd(int16_t *address, int16_t val) {
305
+ gpuAtomicAdd(address, val);
306
+ }
307
+
308
+ inline __device__ void atomicAdd(int64_t *address, int64_t val) {
309
+ gpuAtomicAdd(address, val);
310
+ }
311
+
312
+ inline __device__ void atomicAdd(bool *address, bool val) {
313
+ gpuAtomicAdd(address, val);
314
+ }
315
+
316
+ /* Note [explicitly non-returning atomics]
317
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
318
+ * AMD's MI100 (gfx908) provides an optimized fp32 atomicAdd, exposed via atomicAddNoRet().
319
+ * Due to compiler limitations, callers must opt-in to guarantee the optimized instruction.
320
+ * This non-returning atomicAddNoRet cannot be used to implement the returning atomicAdd,
321
+ * therefore we need a new API 'gpuAtomicAddNoReturn'.
322
+ */
323
+ template<typename T>
324
+ inline __device__ void gpuAtomicAddNoReturn(c10::complex<T> *address, c10::complex<T> val) { gpuAtomicAdd(address, val); }
325
+ inline __device__ void gpuAtomicAddNoReturn(uint8_t *address, uint8_t val) { gpuAtomicAdd(address, val); }
326
+ inline __device__ void gpuAtomicAddNoReturn(int8_t *address, int8_t val) { gpuAtomicAdd(address, val); }
327
+ inline __device__ void gpuAtomicAddNoReturn(int16_t *address, int16_t val) { gpuAtomicAdd(address, val); }
328
+ inline __device__ void gpuAtomicAddNoReturn(int32_t *address, int32_t val) { gpuAtomicAdd(address, val); }
329
+ inline __device__ void gpuAtomicAddNoReturn(int64_t *address, int64_t val) { gpuAtomicAdd(address, val); }
330
+ inline __device__ void gpuAtomicAddNoReturn(bool *address, bool val) { gpuAtomicAdd(address, val); }
331
+ inline __device__ void gpuAtomicAddNoReturn(at::Half *address, at::Half val) { gpuAtomicAdd(address, val); }
332
+ inline __device__ void gpuAtomicAddNoReturn(at::BFloat16 *address, at::BFloat16 val) { gpuAtomicAdd(address, val); }
333
+ inline __device__ void gpuAtomicAddNoReturn(double *address, double val) { gpuAtomicAdd(address, val); }
334
+
335
+ /* Special case fp32 atomic. */
336
+ #if defined(USE_ROCM)
337
+ inline __device__ void gpuAtomicAddNoReturn(float *address, float val) {
338
+ #if defined(__gfx908__)
339
+ atomicAddNoRet(address, val);
340
+ #else
341
+ (void)unsafeAtomicAdd(address, val);
342
+ #endif
343
+ }
344
+ #else
345
+ inline __device__ void gpuAtomicAddNoReturn(float *address, float val) { gpuAtomicAdd(address, val); }
346
+ #endif
347
+
348
+ // Atomic multiplication implementation.
349
+
350
+ ATOMIC_INTEGER_IMPL(Mul)
351
+ GPU_ATOMIC_INTEGER(Mul, a * b, uint8_t)
352
+ GPU_ATOMIC_INTEGER(Mul, a * b, int8_t)
353
+ GPU_ATOMIC_INTEGER(Mul, a * b, int16_t)
354
+ GPU_ATOMIC_INTEGER(Mul, a * b, int32_t)
355
+ GPU_ATOMIC_INTEGER(Mul, a * b, int64_t)
356
+
357
+ inline __device__ at::Half gpuAtomicMul(at::Half * address, at::Half val) {
358
+ return AtomicFPOp<at::Half>()(address, val,
359
+ [](at::Half bsum, at::Half val) {
360
+ return bsum * val;
361
+ });
362
+ }
363
+
364
+ inline __device__ at::BFloat16 gpuAtomicMul(at::BFloat16 * address, at::BFloat16 val) {
365
+ return AtomicFPOp<at::BFloat16>()(address, val,
366
+ [](at::BFloat16 bsum, at::BFloat16 val) {
367
+ return bsum * val;
368
+ });
369
+ }
370
+
371
+ inline __device__ double gpuAtomicMul(double * address, double val) {
372
+ return AtomicFPOp<double>()(address, val,
373
+ [](double val, unsigned long long int assumed) {
374
+ return __double_as_longlong(val * __longlong_as_double(assumed));
375
+ });
376
+ }
377
+
378
+ // Dont use a templated function for this since the addition function defaults to the CUDA built-in.
379
+ inline __device__ float gpuAtomicMul (float * address, float val) {
380
+ unsigned int* address_as_ull = (unsigned int*)address;
381
+ unsigned int old = *address_as_ull;
382
+ unsigned int assumed;
383
+
384
+ do {
385
+ assumed = old;
386
+ old = atomicCAS(address_as_ull, assumed,
387
+ __float_as_int(val *
388
+ __int_as_float(assumed)));
389
+
390
+ // Note: uses integer comparison to avoid hang in case of NaN (since NaN != NaN)
391
+ } while (assumed != old);
392
+
393
+ return __int_as_float(old);
394
+ }
395
+
396
+ // Atomic maximum implementation.
397
+
398
+ template <typename T>
399
+ __host__ __device__ T safe_max(T a, T b) {
400
+ #if defined(__HIPCC__)
401
+ // TODO: remove this special case for HIP when issue is fixed:
402
+ // https://github.com/ROCm-Developer-Tools/HIP/issues/2209
403
+ T max = at::_isnan(a) ? a : (at::_isnan(b) ? b : std::max<T>(a, b));
404
+ #else
405
+ T max = at::_isnan(b) ? b : std::max<T>(a, b);
406
+ #endif
407
+
408
+ return max;
409
+ }
410
+
411
+ ATOMIC_INTEGER_IMPL(Max)
412
+ GPU_ATOMIC_INTEGER(Max, safe_max(a, b), uint8_t)
413
+ GPU_ATOMIC_INTEGER(Max, safe_max(a, b), int8_t)
414
+ GPU_ATOMIC_INTEGER(Max, safe_max(a, b), int16_t)
415
+ GPU_ATOMIC_INTEGER(Max, safe_max(a, b), int32_t)
416
+ GPU_ATOMIC_INTEGER(Max, safe_max(a, b), int64_t)
417
+
418
+ inline __device__ at::Half gpuAtomicMax(at::Half * address, at::Half val) {
419
+ return AtomicFPOp<at::Half>()(address, val,
420
+ [](at::Half bsum, at::Half val) {
421
+ return safe_max(bsum, val);
422
+ });
423
+ }
424
+
425
+ inline __device__ at::BFloat16 gpuAtomicMax(at::BFloat16 * address, at::BFloat16 val) {
426
+ return AtomicFPOp<at::BFloat16>()(address, val,
427
+ [](at::BFloat16 bsum, at::BFloat16 val) {
428
+ return safe_max(bsum, val);
429
+ });
430
+ }
431
+
432
+ inline __device__ double gpuAtomicMax(double * address, double val) {
433
+ return AtomicFPOp<double>()(address, val,
434
+ [](double val, unsigned long long int assumed) {
435
+ return __double_as_longlong(safe_max(val, __longlong_as_double(assumed)));
436
+ });
437
+ }
438
+
439
+ // Dont use a templated function for this since the addition function defaults to the CUDA built-in.
440
+ inline __device__ float gpuAtomicMax(float * address, float val) {
441
+ unsigned int* address_as_ull = (unsigned int*)address;
442
+ unsigned int old = *address_as_ull;
443
+ unsigned int assumed;
444
+
445
+ do {
446
+ assumed = old;
447
+ old = atomicCAS(address_as_ull, assumed,
448
+ __float_as_int(safe_max(val, __int_as_float(assumed))));
449
+
450
+ // Note: uses integer comparison to avoid hang in case of NaN (since NaN != NaN)
451
+ } while (assumed != old);
452
+
453
+ return __int_as_float(old);
454
+ }
455
+
456
+ // Atomic minimum implementation.
457
+
458
+ template <typename T>
459
+ __host__ __device__ T safe_min(T a, T b) {
460
+ #if defined(__HIPCC__)
461
+ // TODO: remove this special case for HIP when issue is fixed:
462
+ // https://github.com/ROCm-Developer-Tools/HIP/issues/2209
463
+ T min = at::_isnan(a) ? a : (at::_isnan(b) ? b : std::min<T>(a, b));
464
+ #else
465
+ T min = at::_isnan(b) ? b : std::min<T>(a, b);
466
+ #endif
467
+
468
+ return min;
469
+ }
470
+
471
+ ATOMIC_INTEGER_IMPL(Min)
472
+ GPU_ATOMIC_INTEGER(Min, safe_min(a, b), uint8_t)
473
+ GPU_ATOMIC_INTEGER(Min, safe_min(a, b), int8_t)
474
+ GPU_ATOMIC_INTEGER(Min, safe_min(a, b), int16_t)
475
+ GPU_ATOMIC_INTEGER(Min, safe_min(a, b), int32_t)
476
+ GPU_ATOMIC_INTEGER(Min, safe_min(a, b), int64_t)
477
+
478
+ inline __device__ at::Half gpuAtomicMin(at::Half * address, at::Half val) {
479
+ return AtomicFPOp<at::Half>()(address, val,
480
+ [](at::Half bsum, at::Half val) {
481
+ return safe_min(bsum, val);
482
+ });
483
+ }
484
+
485
+ inline __device__ at::BFloat16 gpuAtomicMin(at::BFloat16 * address, at::BFloat16 val) {
486
+ return AtomicFPOp<at::BFloat16>()(address, val,
487
+ [](at::BFloat16 bsum, at::BFloat16 val) {
488
+ return safe_min(bsum, val);
489
+ });
490
+ }
491
+
492
+ inline __device__ double gpuAtomicMin(double * address, double val) {
493
+ return AtomicFPOp<double>()(address, val,
494
+ [](double val, unsigned long long int assumed) {
495
+ return __double_as_longlong(safe_min(val, __longlong_as_double(assumed)));
496
+ });
497
+ }
498
+
499
+ // Dont use a templated function for this since the addition function defaults to the CUDA built-in.
500
+ inline __device__ float gpuAtomicMin(float * address, float val) {
501
+ unsigned int* address_as_ull = (unsigned int*)address;
502
+ unsigned int old = *address_as_ull;
503
+ unsigned int assumed;
504
+
505
+ do {
506
+ assumed = old;
507
+ old = atomicCAS(address_as_ull, assumed,
508
+ __float_as_int(safe_min(val, __int_as_float(assumed))));
509
+
510
+ // Note: uses integer comparison to avoid hang in case of NaN (since NaN != NaN)
511
+ } while (assumed != old);
512
+
513
+ return __int_as_float(old);
514
+ }
.venv/lib/python3.11/site-packages/torch/include/ATen/cuda/CUDAApplyUtils.cuh ADDED
@@ -0,0 +1,537 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <ATen/cuda/ApplyGridUtils.cuh>
4
+ #include <ATen/cuda/detail/IndexUtils.cuh>
5
+ #include <ATen/core/TensorBase.h>
6
+ #include <ATen/ceil_div.h>
7
+ #include <ATen/cuda/Atomic.cuh>
8
+ #include <ATen/cuda/CUDAContext.h>
9
+ #include <c10/macros/Macros.h>
10
+ #include <ATen/native/Copy.h>
11
+
12
+ #include <math.h>
13
+
14
+ //
15
+ // This file contains pointwise operation functions and kernels that
16
+ // work on both contiguous and non-contiguous tensor arguments of
17
+ // arbitrary (up to MAX_CUTORCH_DIMS) dimensioned arguments without
18
+ // copying or temporary storage.
19
+ //
20
+
21
+ /*
22
+ NOTE [ CUDA_tensor_applyN helpers ]
23
+
24
+ The following CUDA_tensor_applyN (where N currently can be 1, 2, 3, or 4)
25
+ functions apply a pointwise operator to N tensor(s).
26
+
27
+ The calling convention is
28
+
29
+ 1. The template arguments should be, sequentially,
30
+ - First N typename args specify the scalar types of each of the N tensors.
31
+ - (Optional) `int step` arg specifies the number of elements processed
32
+ together at the same time.
33
+ Default is 1.
34
+ - A usually omitted (i.e., inferred) typename arg specifies the type of the
35
+ function/functor applied on `N * step` values in each iteration of each
36
+ CUDA thread.
37
+ 2. The arguments should be, sequentially,
38
+ - N tensors
39
+ - op: a function/functor that processes `N * step` values at the same time.
40
+ - If `step == 1`, it must have signature
41
+ `void(*)(scalar1_t&, scalar2_t&, ..., scalarN_t&)`, where
42
+ `scalar*_t`s are the first N typename template args, and the inputs
43
+ are the `N` values from the `N` tensors retrieved at a common index.
44
+ - Otherwise, it must must have signature
45
+ void(*)(int n, scalar1_t&, scalar1_t&, ..., scalar1_t&, // repeat `step` times
46
+ scalar2_t&, scalar2_t&, ..., scalar2_t&, // repeat `step` times
47
+ ...,
48
+ scalarN_t&, scalarN_t&, ..., scalarN_t&) // repeat `step` times
49
+ Different from `step == 1` case, it processes `N * step` values taken
50
+ from `step` common indices. Moreover, the first input `n` represents the
51
+ number of valid indices (it will always have `0 < n <= step`). It will
52
+ almost always be `step`, but at the boundary we may not have full `step`
53
+ elements and `n` can be a lesser value.
54
+
55
+ E.g., if `step == 4` and `N == 2`, `op` could be
56
+
57
+ [](int n, scalar1_t &u1, scalar1_t &u2, scalar1_t &u3, scalar1_t &u4,
58
+ scalar2_t &v1, scalar2_t &v2, scalar2_t &v3, scalar2_t &v4) {
59
+ // Only process u1, ..., un and v1, ..., vn.
60
+ // So if `n == 3`, `u4` and `v4` need not to be considered.
61
+ }
62
+
63
+ In both cases, the references can actually be const, but at least one of
64
+ them should be non-const in order to write the output.
65
+ - (Optional, but recommended) N TensorArgType args that specify for each
66
+ tensor whether `op` reads AND writes ] (i.e., TensorArgType::ReadWrite),
67
+ or only reads (i.e., TensorArgType::ReadOnly).
68
+ Default is TensorArgType::ReadWrite for first Tensor, and
69
+ TensorArgType::ReadOnly for the rest.
70
+
71
+ E.g.,
72
+
73
+ to compute a = b^2 for a and b of same dtype, we can call
74
+
75
+ CUDA_tensor_apply2<scalar, scalar>(
76
+ a, b,
77
+ [] __device__ (scalar &a_val, const scalar &b_val) { a_val = b_val * b_val; }
78
+ );
79
+
80
+ to work on 2 values at the same time, we can call
81
+
82
+ CUDA_tensor_apply2<scalar1, scalar2, 2>(
83
+ a, b,
84
+ [] __device__ (int n, scalar1 &a_val1, scalar1 &a_val2,
85
+ const scalar2 &b_val1, const scalar2 &b_val2) {
86
+ // call special vectorized op here, or just do elementwise and enjoy unrolling...
87
+ // if n == 1, only process a_val1 and b_val1
88
+ }
89
+ );
90
+ */
91
+
92
+ namespace at::cuda {
93
+
94
+ // TODO: combine with TensorArg? So far that's been for debugging, and this is functional...
95
+ enum class TensorArgType { ReadWrite, ReadOnly };
96
+
97
+ namespace {
98
+
99
+ // Rearrange dimensions for pointwise operations so that strides are in
100
+ // decreasing order as much as possible, so that kernels have better memory
101
+ // access patterns.
102
+ //
103
+ // For example, consider a binary operation on two "transposed" 2-dim tensors:
104
+ // sizes: 256 512
105
+ // aInfo->strides: 1 256
106
+ // bInfo->strides: 1 256
107
+ //
108
+ // Given this, each concurrent memory access inside kernelPointwiseApply2() is
109
+ // exactly 256 elements apart, resulting in poor performance.
110
+ //
111
+ // This function exchanges dimensions so that memory access is contiguous:
112
+ // sizes: 512 256
113
+ // aInfo->strides: 256 1
114
+ // bInfo->strides: 256 1
115
+ //
116
+ // (Actually, it becomes even better because now collapseDims() can turn each
117
+ // input into one contiguous array.)
118
+ //
119
+ // In general, given M (<=4) TensorInfo's with N dimensions, we can view each
120
+ // strides[i] (0 <= i < N) as an M-tuple. Given each pair i < j, we exchange
121
+ // strides[i] and [j] if
122
+ // (1) strides[i][k] < strides[j][k] for some k (0 <= k < M)
123
+ // (exchanging them will benefit input #k), and
124
+ // (2) strides[i][k] <= strieds[j][k] for all k
125
+ // (exchanging them will not make any input worse).
126
+ template <typename T1, typename IndexType,
127
+ typename T2 = void, typename T3 = void, typename T4 = void>
128
+ inline void rearrangeDims(detail::TensorInfo<T1, IndexType>* aInfo,
129
+ detail::TensorInfo<T2, IndexType>* bInfo = nullptr,
130
+ detail::TensorInfo<T3, IndexType>* cInfo = nullptr,
131
+ detail::TensorInfo<T4, IndexType>* dInfo = nullptr) {
132
+ int numInfos = 1;
133
+ int dims = aInfo->dims;
134
+ IndexType *sizes[4] = { aInfo->sizes, };
135
+ IndexType *strides[4] = { aInfo->strides, };
136
+
137
+ if (bInfo != nullptr) {
138
+ ++numInfos;
139
+ if (bInfo->dims != dims) return;
140
+ sizes[1] = bInfo->sizes;
141
+ strides[1] = bInfo->strides;
142
+ }
143
+
144
+ if (cInfo != nullptr) {
145
+ ++numInfos;
146
+ if (cInfo->dims != dims) return;
147
+ sizes[2] = cInfo->sizes;
148
+ strides[2] = cInfo->strides;
149
+ }
150
+
151
+ if (dInfo != nullptr) {
152
+ ++numInfos;
153
+ if (dInfo->dims != dims) return;
154
+ sizes[3] = dInfo->sizes;
155
+ strides[3] = dInfo->strides;
156
+ }
157
+
158
+ // Bail out if sizes do not match: we are using "deprecated pointwise
159
+ // behavior" among tensors of different shapes but same number of elements.
160
+ for (int i = 1; i < numInfos; ++i) {
161
+ for (int j = 0; j < dims; ++j) {
162
+ if (sizes[i][j] != sizes[0][j]) return;
163
+ }
164
+ }
165
+
166
+ for (int i = 0; i < dims - 1; ++i) {
167
+ // No need to consider dimensions of size 1.
168
+ if (sizes[0][i] == 1) continue;
169
+
170
+ for (int j = i + 1; j < dims; ++j) {
171
+ if (sizes[0][j] == 1) continue;
172
+
173
+ // Compare the relative sizes of strides between dim #i and dim #j.
174
+ bool hasIncreasingStrides = false;
175
+ bool hasDecreasingStrides = false;
176
+
177
+ for (int k = 0; k < numInfos; k++) {
178
+ IndexType stride_i = strides[k][i];
179
+ IndexType stride_j = strides[k][j];
180
+ if (stride_i < stride_j) {
181
+ hasIncreasingStrides = true;
182
+ } else if (stride_i > stride_j) {
183
+ hasDecreasingStrides = true;
184
+ }
185
+ }
186
+
187
+ if (hasIncreasingStrides && !hasDecreasingStrides) {
188
+ for (int k = 0; k < numInfos; k++) {
189
+ IndexType size = sizes[k][i];
190
+ sizes[k][i] = sizes[k][j];
191
+ sizes[k][j] = size;
192
+
193
+ IndexType stride = strides[k][i];
194
+ strides[k][i] = strides[k][j];
195
+ strides[k][j] = stride;
196
+ }
197
+ }
198
+ }
199
+ }
200
+ }
201
+
202
+ // The `remaining_steps` argument is used to support Op that operates on
203
+ // multiple elements at the same time. Generally, the strategy of ApplyOpN is to
204
+ // 1. Initialize `remaining_steps = step`, where `step` is the template arg of
205
+ // CUDA_tensor_applyN helpers. The input arg `n` to `apply()` represents the
206
+ // number of elements in bound for this call. It will almost always equal to
207
+ // `step` except at boundaries.
208
+ // 2. If `remaining_steps > 0` convert the current linearIndex to offset (if in
209
+ // bound), and recursively call `ApplyOpN` with `remaining_steps - 1`.
210
+ // 3. At `remaining_steps = 0`,
211
+ // if `step = 1`, call `op(tensor1_val, tensor2_val, ...)`;
212
+ // if `step > 1`, call `op(n, tensor1_val1, tensor1_val2, ..., tesor1_valstep,
213
+ // tensor2_val1, tensor2_val2, ..., tesor2_valstep,
214
+ // ...
215
+ // tensorN_val1, tensorN_val2, ..., tesorN_valstep);`
216
+ //
217
+ // See NOTE [ CUDA_tensor_applyN helpers ] above for how Op may look like.
218
+
219
+ template <typename Op,
220
+ typename scalar,
221
+ typename IndexType,
222
+ int ADims,
223
+ int remaining_steps,
224
+ typename... Offsets>
225
+ struct ApplyOp1 {
226
+ __device__ __forceinline__
227
+ static void apply(detail::TensorInfo<scalar, IndexType> &a, const Op &op, int n,
228
+ IndexType linearIndex, Offsets... aOffsets) {
229
+ // Convert `linearIndex` into an offset of `a`
230
+ const IndexType aOffset = sizeof...(Offsets) < n ?
231
+ detail::IndexToOffset<scalar, IndexType, ADims>::get(linearIndex, a) : 0;
232
+
233
+ ApplyOp1<Op, scalar, IndexType, ADims, remaining_steps - 1, const IndexType, Offsets...>::apply(
234
+ a, op, n, linearIndex + 1, aOffsets..., aOffset
235
+ );
236
+ }
237
+ };
238
+
239
+ // Specialize `step=1` case (i.e., `remaining_steps=0` and `len(Offsets)=1`).
240
+ // We don't need to pass in how many elements need to processed in this case.
241
+ template <typename Op,
242
+ typename scalar,
243
+ typename IndexType,
244
+ int ADims,
245
+ typename Offset>
246
+ struct ApplyOp1<Op, scalar, IndexType, ADims, 0, Offset> {
247
+ __device__ __forceinline__
248
+ static void apply(detail::TensorInfo<scalar, IndexType> &a, const Op &op,
249
+ int n, IndexType linearIndex, Offset offset) {
250
+ op(a.data[offset]);
251
+ }
252
+ };
253
+
254
+ template <typename Op,
255
+ typename scalar,
256
+ typename IndexType,
257
+ int ADims,
258
+ typename... Offsets>
259
+ struct ApplyOp1<Op, scalar, IndexType, ADims, 0, Offsets...> {
260
+ __device__ __forceinline__
261
+ static void apply(detail::TensorInfo<scalar, IndexType> &a, const Op &op, int n,
262
+ IndexType linearIndex, Offsets... offsets) {
263
+ op(n, a.data[offsets]...);
264
+ }
265
+ };
266
+
267
+ template <typename Op,
268
+ typename scalar,
269
+ typename IndexType,
270
+ int ADims,
271
+ int step>
272
+ #if __CUDA_ARCH__ >= 350 || defined(USE_ROCM)
273
+ C10_LAUNCH_BOUNDS_2(AT_APPLY_THREADS_PER_BLOCK, AT_APPLY_BLOCKS_PER_SM)
274
+ #endif
275
+ __global__ void kernelPointwiseApply1(detail::TensorInfo<scalar, IndexType> a,
276
+ IndexType totalElements, const Op op) {
277
+ for (IndexType linearIndex = (blockIdx.x * blockDim.x + threadIdx.x) * step;
278
+ linearIndex < totalElements;
279
+ linearIndex += gridDim.x * blockDim.x * step) {
280
+ ApplyOp1<Op, scalar, IndexType, ADims, step>::apply(
281
+ a, op, ::min(step, static_cast<int>(totalElements - linearIndex)), linearIndex);
282
+ }
283
+ }
284
+
285
+
286
+ template <typename Op,
287
+ typename scalar1,
288
+ typename scalar2,
289
+ typename IndexType,
290
+ int ADims,
291
+ int BDims,
292
+ int remaining_steps,
293
+ typename... Offsets>
294
+ struct ApplyOp2 {
295
+ __device__ __forceinline__
296
+ static void apply(detail::TensorInfo<scalar1, IndexType> &a,
297
+ detail::TensorInfo<scalar2, IndexType> &b,
298
+ const Op &op, int64_t n, IndexType linearIndex,
299
+ Offsets... aOffsets, Offsets... bOffsets) {
300
+ // Convert `linearIndex` into an offset of `a`
301
+ const IndexType aOffset = static_cast<int64_t>(sizeof...(Offsets)) < n ?
302
+ detail::IndexToOffset<scalar1, IndexType, ADims>::get(linearIndex, a) : 0;
303
+
304
+ // Convert `linearIndex` into an offset of `b`
305
+ const IndexType bOffset = static_cast<int64_t>(sizeof...(Offsets)) < n ?
306
+ detail::IndexToOffset<scalar2, IndexType, BDims>::get(linearIndex, b) : 0;
307
+
308
+ ApplyOp2<Op, scalar1, scalar2, IndexType, ADims, BDims, remaining_steps - 1, const IndexType, Offsets...>::apply(
309
+ a, b, op, n, linearIndex + 1, aOffsets..., aOffset, bOffsets..., bOffset
310
+ );
311
+ }
312
+ };
313
+
314
+ // Specialize `step=1` case (i.e., `remaining_steps=0` and `len(Offsets)=1`).
315
+ // We don't need to pass in how many elements need to processed in this case.
316
+ template <typename Op,
317
+ typename scalar1,
318
+ typename scalar2,
319
+ typename IndexType,
320
+ int ADims,
321
+ int BDims,
322
+ typename Offset>
323
+ struct ApplyOp2<Op, scalar1, scalar2, IndexType, ADims, BDims, 0, Offset> {
324
+ __device__ __forceinline__
325
+ static void apply(detail::TensorInfo<scalar1, IndexType> &a,
326
+ detail::TensorInfo<scalar2, IndexType> &b,
327
+ const Op &op, int /*n*/, IndexType /*linearIndex*/,
328
+ Offset aOffset, Offset bOffset) {
329
+ op(a.data[aOffset], b.data[bOffset]);
330
+ }
331
+ };
332
+
333
+ template <typename Op,
334
+ typename scalar1,
335
+ typename scalar2,
336
+ typename IndexType,
337
+ int ADims,
338
+ int BDims,
339
+ typename... Offsets>
340
+ struct ApplyOp2<Op, scalar1, scalar2, IndexType, ADims, BDims, 0, Offsets...> {
341
+ __device__ __forceinline__
342
+ static void apply(detail::TensorInfo<scalar1, IndexType> &a,
343
+ detail::TensorInfo<scalar2, IndexType> &b,
344
+ const Op &op, int n, IndexType linearIndex,
345
+ Offsets... aOffsets, Offsets... bOffsets) {
346
+ op(n, a.data[aOffsets]..., b.data[bOffsets]...);
347
+ }
348
+ };
349
+
350
+ template <typename Op,
351
+ typename scalar1,
352
+ typename scalar2,
353
+ typename IndexType,
354
+ int ADims, int BDims,
355
+ int step,
356
+ int max_threads_per_block=AT_APPLY_THREADS_PER_BLOCK,
357
+ int min_blocks_per_sm=AT_APPLY_BLOCKS_PER_SM>
358
+ #if __CUDA_ARCH__ >= 350 || defined(USE_ROCM)
359
+ C10_LAUNCH_BOUNDS_2(max_threads_per_block, min_blocks_per_sm)
360
+ #endif
361
+ __global__ void
362
+ kernelPointwiseApply2(detail::TensorInfo<scalar1, IndexType> a,
363
+ detail::TensorInfo<scalar2, IndexType> b,
364
+ IndexType totalElements,
365
+ const Op op) {
366
+ for (IndexType linearIndex = (blockIdx.x * blockDim.x + threadIdx.x) * step;
367
+ linearIndex < totalElements;
368
+ linearIndex += gridDim.x * blockDim.x * step) {
369
+ ApplyOp2<Op, scalar1, scalar2, IndexType, ADims, BDims, step>::apply(
370
+ a, b, op, ::min(step, static_cast<int>(totalElements - linearIndex)),
371
+ linearIndex);
372
+ }
373
+ }
374
+
375
+ } // anonymous namespace
376
+
377
+ template <typename scalar1, typename scalar2, int step, typename Op,
378
+ int max_threads_per_block=AT_APPLY_THREADS_PER_BLOCK,
379
+ int min_blocks_per_sm=AT_APPLY_BLOCKS_PER_SM>
380
+ inline bool CUDA_tensor_apply2(at::TensorBase a,
381
+ at::TensorBase b,
382
+ const Op op,
383
+ TensorArgType aType = TensorArgType::ReadWrite,
384
+ TensorArgType bType = TensorArgType::ReadOnly) {
385
+ TORCH_CHECK(a.device().is_cuda() && b.device().is_cuda(),
386
+ "CUDA_tensor_apply2: Expected tensors to have CUDA DeviceType, but got "
387
+ "tensors with type ", a.device().type(), " and ", b.device().type());
388
+ int64_t totalElements = a.numel();
389
+
390
+ if (totalElements != b.numel()) {
391
+ return false;
392
+ }
393
+
394
+ if (a.dim() > MAX_TENSORINFO_DIMS ||
395
+ b.dim() > MAX_TENSORINFO_DIMS) {
396
+ return false;
397
+ }
398
+
399
+ if (a.numel() == 0) {
400
+ // Empty tensor; do nothing
401
+ return true;
402
+ }
403
+ const dim3 block = getApplyBlock(max_threads_per_block);
404
+
405
+ dim3 grid;
406
+ auto curDevice = current_device();
407
+ if (curDevice == -1) return false;
408
+ if (!getApplyGrid<step>(totalElements, grid, curDevice, max_threads_per_block)) {
409
+ return false;
410
+ }
411
+
412
+ /*
413
+ Expands readable/writable tensors whose indices may be "overlapped."
414
+ This ensures that each element of the tensor is operated on once and only
415
+ once.
416
+ */
417
+ TensorBase oldA;
418
+ TensorBase oldB;
419
+
420
+ if (aType == TensorArgType::ReadWrite && detail::maybeOverlappingIndices(a)) {
421
+ // Must perform in contiguous space
422
+ oldA = std::exchange(a, a.contiguous());
423
+ }
424
+ if (bType == TensorArgType::ReadWrite && detail::maybeOverlappingIndices(b)) {
425
+ // Must perform in contiguous space
426
+ oldB = std::exchange(b, b.contiguous());
427
+ }
428
+
429
+ // It is possible that the tensor dimensions are able to be collapsed,
430
+ // and thus we can reduce the actual code complexity of the copy by
431
+ // exploiting this knowledge statically, since the div/mod is the
432
+ // most expensive part of the operation, more so than memory accesses.
433
+ // For instance, when copying a non-contiguous to a contiguous tensor
434
+ // (or vice versa), the contiguous tensor can be collapsed to one
435
+ // dimension, and the loop to translate the linear index to the array
436
+ // index can be similarly collapsed. That is what this unrolling is for.
437
+
438
+ #define HANDLE_CASE(TYPE, A, B) \
439
+ kernelPointwiseApply2<Op, \
440
+ scalar1, \
441
+ scalar2, \
442
+ TYPE, A, B, step, \
443
+ max_threads_per_block, \
444
+ min_blocks_per_sm> \
445
+ <<<grid, block, 0, at::cuda::getCurrentCUDAStream(curDevice)>>>( \
446
+ aInfo, bInfo, static_cast<TYPE>(totalElements), op); \
447
+ C10_CUDA_KERNEL_LAUNCH_CHECK();
448
+
449
+ #define HANDLE_B_CASE(TYPE, A, B) { \
450
+ switch (B) { \
451
+ case 1: \
452
+ HANDLE_CASE(TYPE, A, 1); \
453
+ break; \
454
+ case 2: \
455
+ HANDLE_CASE(TYPE, A, 2); \
456
+ break; \
457
+ default: \
458
+ HANDLE_CASE(TYPE, A, -1); \
459
+ break; \
460
+ } \
461
+ }
462
+
463
+ #define HANDLE_A_CASE(TYPE, A, B) { \
464
+ switch (A) { \
465
+ case 1: \
466
+ HANDLE_B_CASE(TYPE, 1, B); \
467
+ break; \
468
+ case 2: \
469
+ HANDLE_B_CASE(TYPE, 2, B); \
470
+ break; \
471
+ default: \
472
+ HANDLE_B_CASE(TYPE, -1, B); \
473
+ break; \
474
+ } \
475
+ }
476
+
477
+ if (detail::canUse32BitIndexMath(a) &&
478
+ detail::canUse32BitIndexMath(b)) {
479
+ detail::TensorInfo<scalar1, unsigned int> aInfo =
480
+ detail::getTensorInfo<scalar1, unsigned int>(a);
481
+
482
+ detail::TensorInfo<scalar2, unsigned int> bInfo =
483
+ detail::getTensorInfo<scalar2, unsigned int>(b);
484
+ rearrangeDims(&aInfo, &bInfo);
485
+ aInfo.collapseDims();
486
+ bInfo.collapseDims();
487
+
488
+ HANDLE_A_CASE(unsigned int, aInfo.dims, bInfo.dims);
489
+ } else {
490
+ detail::TensorInfo<scalar1, uint64_t> aInfo =
491
+ detail::getTensorInfo<scalar1, uint64_t>(a);
492
+
493
+ detail::TensorInfo<scalar2, uint64_t> bInfo =
494
+ detail::getTensorInfo<scalar2, uint64_t>(b);
495
+ rearrangeDims(&aInfo, &bInfo);
496
+ aInfo.collapseDims();
497
+ bInfo.collapseDims();
498
+
499
+ /*
500
+ Only instantiates the all 1D special case and the fallback all nD case for
501
+ large (64-bit indexed) tensors to reduce compilation time.
502
+ */
503
+ if (aInfo.dims == 1 && bInfo.dims == 1) {
504
+ HANDLE_CASE(uint64_t, 1, 1);
505
+ } else {
506
+ HANDLE_CASE(uint64_t, -1, -1);
507
+ }
508
+ }
509
+ #undef HANDLE_CASE
510
+ #undef HANDLE_B_CASE
511
+ #undef HANDLE_A_CASE
512
+
513
+ if (oldA.defined()) {
514
+ at::native::copy_ignoring_overlaps(oldA, a);
515
+ }
516
+
517
+ if (oldB.defined()) {
518
+ at::native::copy_ignoring_overlaps(oldB, b);
519
+ }
520
+
521
+ return true;
522
+ }
523
+
524
+ /* Provides default step = 1 to CUDA_tensor_apply2. */
525
+ template <typename scalar1, typename scalar2, typename Op,
526
+ int max_threads_per_block=AT_APPLY_THREADS_PER_BLOCK,
527
+ int min_blocks_per_sm=AT_APPLY_BLOCKS_PER_SM>
528
+ inline bool CUDA_tensor_apply2(const at::TensorBase &a,
529
+ const at::TensorBase &b,
530
+ const Op op,
531
+ TensorArgType aType = TensorArgType::ReadWrite,
532
+ TensorArgType bType = TensorArgType::ReadOnly) {
533
+ return CUDA_tensor_apply2<scalar1, scalar2, 1, Op,
534
+ max_threads_per_block, min_blocks_per_sm>(a, b, op, aType, bType);
535
+ }
536
+
537
+ } // namespace at::cuda
.venv/lib/python3.11/site-packages/torch/include/ATen/cuda/CUDABlas.h ADDED
@@ -0,0 +1,358 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+ /*
3
+ Provides a subset of CUDA BLAS functions as templates:
4
+
5
+ gemm<Dtype>(transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c,
6
+ ldc)
7
+
8
+ gemv<Dtype>(transa, m, n, alpha, a, lda, x, incx, beta, y, incy)
9
+
10
+ dot<Dtype>(n, x, incx, y, incy, result)
11
+
12
+ where Dtype is double, float, at::Half or at::BFloat16 (ROCm, NOT for dot).
13
+ The functions are available in at::cuda::blas namespace.
14
+ */
15
+
16
+ #include <ATen/cuda/CUDAContext.h>
17
+ #include <ATen/OpMathType.h>
18
+
19
+ namespace at::cuda::blas {
20
+
21
+ // RAII guard that sets the CuBLAS pointer mode and restores it to
22
+ // its previous value when the guard is destroyed
23
+ class PointerModeGuard {
24
+ public:
25
+ PointerModeGuard(cublasHandle_t handle, cublasPointerMode_t mode) :
26
+ handle(handle) {
27
+ TORCH_CUDABLAS_CHECK(cublasGetPointerMode(handle, &previous_mode));
28
+ TORCH_CUDABLAS_CHECK(cublasSetPointerMode(handle, mode));
29
+ }
30
+
31
+ ~PointerModeGuard() {
32
+ cublasSetPointerMode(handle, previous_mode);
33
+ }
34
+
35
+ private:
36
+ cublasHandle_t handle;
37
+ cublasPointerMode_t previous_mode;
38
+ };
39
+
40
+ /* LEVEL 3 BLAS FUNCTIONS */
41
+
42
+ #define CUDABLAS_GEMM_ARGTYPES(Dtype) \
43
+ char transa, char transb, int64_t m, int64_t n, int64_t k, at::opmath_type<Dtype> alpha, \
44
+ const Dtype *a, int64_t lda, const Dtype *b, int64_t ldb, at::opmath_type<Dtype> beta,\
45
+ Dtype *c, int64_t ldc
46
+
47
+ #define CUDABLAS_GEMM_ARGS(Dtype) transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc
48
+
49
+ template <typename Dtype>
50
+ inline void gemm(CUDABLAS_GEMM_ARGTYPES(Dtype)) {
51
+ static_assert(false&&sizeof(Dtype),"at::cuda::blas::gemm: not implemented");
52
+ }
53
+
54
+ template <>
55
+ void gemm<double>(CUDABLAS_GEMM_ARGTYPES(double));
56
+ template <>
57
+ void gemm<float>(CUDABLAS_GEMM_ARGTYPES(float));
58
+ template <>
59
+ void gemm<c10::complex<double>>(CUDABLAS_GEMM_ARGTYPES(c10::complex<double>));
60
+ template <>
61
+ void gemm<c10::complex<float>>(CUDABLAS_GEMM_ARGTYPES(c10::complex<float>));
62
+ template <>
63
+ void gemm<at::Half>(CUDABLAS_GEMM_ARGTYPES(at::Half));
64
+ template <>
65
+ void gemm<at::BFloat16>(CUDABLAS_GEMM_ARGTYPES(at::BFloat16));
66
+
67
+ template <typename Dtype>
68
+ inline void gemm_internal(CUDABLAS_GEMM_ARGTYPES(Dtype)) {
69
+ static_assert(false&&sizeof(Dtype),"at::cuda::blas::gemm_internal: not implemented");
70
+ }
71
+
72
+ template <>
73
+ void gemm_internal<double>(CUDABLAS_GEMM_ARGTYPES(double));
74
+ template <>
75
+ void gemm_internal<float>(CUDABLAS_GEMM_ARGTYPES(float));
76
+ template <>
77
+ void gemm_internal<c10::complex<double>>(CUDABLAS_GEMM_ARGTYPES(c10::complex<double>));
78
+ template <>
79
+ void gemm_internal<c10::complex<float>>(CUDABLAS_GEMM_ARGTYPES(c10::complex<float>));
80
+ template <>
81
+ void gemm_internal<at::Half>(CUDABLAS_GEMM_ARGTYPES(at::Half));
82
+ template <>
83
+ void gemm_internal<at::BFloat16>(CUDABLAS_GEMM_ARGTYPES(at::BFloat16));
84
+
85
+ enum GEMMAndBiasActivationEpilogue {
86
+ None,
87
+ RELU,
88
+ GELU,
89
+ };
90
+
91
+ // NOTE: GELU activation is not supported prior to CUDA 11.4 and will
92
+ // do nothing if passed in that case.
93
+ template <typename Dtype>
94
+ void gemm_and_bias(
95
+ bool transpose_mat1,
96
+ bool transpose_mat2,
97
+ int64_t m,
98
+ int64_t n,
99
+ int64_t k,
100
+ at::opmath_type<Dtype> alpha_val,
101
+ const Dtype* mat1_ptr,
102
+ int64_t mat1_ld,
103
+ const Dtype* mat2_ptr,
104
+ int64_t mat2_ld,
105
+ const Dtype* bias,
106
+ Dtype* result_ptr,
107
+ int64_t result_ld,
108
+ GEMMAndBiasActivationEpilogue activation = GEMMAndBiasActivationEpilogue::None);
109
+
110
+ void int8_gemm(
111
+ bool transpose_mat1,
112
+ bool transpose_mat2,
113
+ int64_t m,
114
+ int64_t n,
115
+ int64_t k,
116
+ const int8_t* mat1_ptr,
117
+ int64_t mat1_ld,
118
+ const int8_t* mat2_ptr,
119
+ int64_t mat2_ld,
120
+ int32_t* result_ptr,
121
+ int64_t result_ld);
122
+
123
+ void scaled_gemm(
124
+ char transa,
125
+ char transb,
126
+ int64_t m,
127
+ int64_t n,
128
+ int64_t k,
129
+ const void* mat1_ptr,
130
+ const void* mat1_scale_ptr,
131
+ int64_t mat1_ld,
132
+ ScalarType mat1_dtype,
133
+ const void* mat2_ptr,
134
+ const void* mat2_scale_ptr,
135
+ int64_t mat2_ld,
136
+ ScalarType mat2_dtype,
137
+ const void* bias_ptr,
138
+ ScalarType bias_dtype,
139
+ void* result_ptr,
140
+ const void* result_scale_ptr,
141
+ int64_t result_ld,
142
+ ScalarType result_dtype,
143
+ void* amax_ptr,
144
+ bool use_fast_accum);
145
+
146
+ #define CUDABLAS_BGEMM_ARGTYPES(Dtype) \
147
+ char transa, char transb, int64_t m, int64_t n, int64_t k, at::opmath_type<Dtype> alpha, \
148
+ const Dtype *a, int64_t lda, int64_t stridea, \
149
+ const Dtype *b, int64_t ldb, int64_t strideb, \
150
+ at::opmath_type<Dtype> beta, Dtype *c, int64_t ldc, int64_t stridec, int64_t num_batches
151
+
152
+ #define CUDABLAS_BGEMM_ARGS(Dtype) \
153
+ transa, transb, m, n, k, alpha, a, lda, stridea, b, ldb, strideb, beta, c, ldc, stridec, num_batches
154
+
155
+ template <typename Dtype>
156
+ inline void bgemm(CUDABLAS_BGEMM_ARGTYPES(Dtype)) {
157
+ static_assert(false&&sizeof(Dtype),"at::cuda::blas::bgemm: not implemented");
158
+ }
159
+
160
+ template <>
161
+ void bgemm<double>(CUDABLAS_BGEMM_ARGTYPES(double));
162
+ template <>
163
+ void bgemm<float>(CUDABLAS_BGEMM_ARGTYPES(float));
164
+ template <>
165
+ void bgemm<c10::complex<double>>(CUDABLAS_BGEMM_ARGTYPES(c10::complex<double>));
166
+ template <>
167
+ void bgemm<c10::complex<float>>(CUDABLAS_BGEMM_ARGTYPES(c10::complex<float>));
168
+ template <>
169
+ void bgemm<at::Half>(CUDABLAS_BGEMM_ARGTYPES(at::Half));
170
+ template <>
171
+ void bgemm<at::BFloat16>(CUDABLAS_BGEMM_ARGTYPES(at::BFloat16));
172
+
173
+ template <typename Dtype>
174
+ inline void bgemm_internal(CUDABLAS_BGEMM_ARGTYPES(Dtype)) {
175
+ static_assert(false&&sizeof(Dtype),"at::cuda::blas::bgemm_internal: not implemented");
176
+ }
177
+
178
+ template <>
179
+ void bgemm_internal<double>(CUDABLAS_BGEMM_ARGTYPES(double));
180
+ template <>
181
+ void bgemm_internal<float>(CUDABLAS_BGEMM_ARGTYPES(float));
182
+ template <>
183
+ void bgemm_internal<c10::complex<double>>(CUDABLAS_BGEMM_ARGTYPES(c10::complex<double>));
184
+ template <>
185
+ void bgemm_internal<c10::complex<float>>(CUDABLAS_BGEMM_ARGTYPES(c10::complex<float>));
186
+ template <>
187
+ void bgemm_internal<at::Half>(CUDABLAS_BGEMM_ARGTYPES(at::Half));
188
+ template <>
189
+ void bgemm_internal<at::BFloat16>(CUDABLAS_BGEMM_ARGTYPES(at::BFloat16));
190
+
191
+ #define CUDABLAS_TRSM_ARGTYPES(Dtype) \
192
+ cublasHandle_t handle, cublasSideMode_t side, cublasFillMode_t uplo, \
193
+ cublasOperation_t trans, cublasDiagType_t diag, int m, int n, \
194
+ const Dtype *alpha, const Dtype *A, int lda, Dtype *B, int ldb
195
+
196
+ template <typename Dtype>
197
+ inline void trsm(CUDABLAS_TRSM_ARGTYPES(Dtype)) {
198
+ static_assert(false&&sizeof(Dtype), "at::cuda::blas::trsm: not implemented");
199
+ }
200
+
201
+ template <>
202
+ TORCH_CUDA_CU_API void trsm<float>(CUDABLAS_TRSM_ARGTYPES(float));
203
+ template <>
204
+ TORCH_CUDA_CU_API void trsm<double>(CUDABLAS_TRSM_ARGTYPES(double));
205
+ template <>
206
+ TORCH_CUDA_CU_API void trsm<c10::complex<float>>(CUDABLAS_TRSM_ARGTYPES(c10::complex<float>));
207
+ template <>
208
+ TORCH_CUDA_CU_API void trsm<c10::complex<double>>(CUDABLAS_TRSM_ARGTYPES(c10::complex<double>));
209
+
210
+ #define CUDABLAS_TRSM_BATCHED_ARGTYPES(Dtype) \
211
+ cublasHandle_t handle, cublasSideMode_t side, cublasFillMode_t uplo, \
212
+ cublasOperation_t trans, cublasDiagType_t diag, int m, int n, \
213
+ const Dtype *alpha, Dtype *A[], int lda, Dtype *B[], int ldb, \
214
+ int batchCount
215
+
216
+ template <typename Dtype>
217
+ inline void trsmBatched(CUDABLAS_TRSM_BATCHED_ARGTYPES(Dtype)) {
218
+ static_assert(false&&sizeof(Dtype), "at::cuda::blas::trsmBatched: not implemented");
219
+ }
220
+
221
+ template <>
222
+ TORCH_CUDA_CU_API void trsmBatched<float>(CUDABLAS_TRSM_BATCHED_ARGTYPES(float));
223
+ template <>
224
+ TORCH_CUDA_CU_API void trsmBatched<double>(CUDABLAS_TRSM_BATCHED_ARGTYPES(double));
225
+ template <>
226
+ TORCH_CUDA_CU_API void trsmBatched<c10::complex<float>>(CUDABLAS_TRSM_BATCHED_ARGTYPES(c10::complex<float>));
227
+ template <>
228
+ TORCH_CUDA_CU_API void trsmBatched<c10::complex<double>>(CUDABLAS_TRSM_BATCHED_ARGTYPES(c10::complex<double>));
229
+
230
+ /* LEVEL 2 BLAS FUNCTIONS */
231
+
232
+ #define CUDABLAS_GEMV_ARGTYPES(Dtype) \
233
+ char trans, int64_t m, int64_t n, Dtype alpha, const Dtype *a, int64_t lda, \
234
+ const Dtype *x, int64_t incx, Dtype beta, Dtype *y, int64_t incy
235
+
236
+ template <typename Dtype>
237
+ inline void gemv(CUDABLAS_GEMV_ARGTYPES(Dtype)) {
238
+ static_assert(false&&sizeof(Dtype), "at::cuda::blas::gemv: not implemented");
239
+ }
240
+
241
+ template <>
242
+ void gemv<double>(CUDABLAS_GEMV_ARGTYPES(double));
243
+ template <>
244
+ void gemv<float>(CUDABLAS_GEMV_ARGTYPES(float));
245
+ template <>
246
+ void gemv<c10::complex<double>>(CUDABLAS_GEMV_ARGTYPES(c10::complex<double>));
247
+ template <>
248
+ void gemv<c10::complex<float>>(CUDABLAS_GEMV_ARGTYPES(c10::complex<float>));
249
+ template <>
250
+ void gemv<at::Half>(CUDABLAS_GEMV_ARGTYPES(at::Half));
251
+ template <>
252
+ void gemv<at::BFloat16>(CUDABLAS_GEMV_ARGTYPES(at::BFloat16));
253
+
254
+ /* LEVEL 1 BLAS FUNCTIONS */
255
+
256
+ #define CUDABLAS_DOT_ARGTYPES(Dtype) \
257
+ cublasHandle_t handle, int n, const Dtype *x, int incx, const Dtype *y, \
258
+ int incy, Dtype *result
259
+
260
+ template <typename Dtype>
261
+ inline void dot(CUDABLAS_DOT_ARGTYPES(Dtype)) {
262
+ static_assert(false&&sizeof(Dtype),"at::cuda::blas::dot: not implemented");
263
+ }
264
+
265
+ template <>
266
+ void dot<double>(CUDABLAS_DOT_ARGTYPES(double));
267
+ template <>
268
+ void dot<float>(CUDABLAS_DOT_ARGTYPES(float));
269
+ template <>
270
+ void dot<at::Half>(CUDABLAS_DOT_ARGTYPES(at::Half));
271
+ template <>
272
+ void dot<at::BFloat16>(CUDABLAS_DOT_ARGTYPES(at::BFloat16));
273
+ template <>
274
+ void dot<c10::complex<double>>(CUDABLAS_DOT_ARGTYPES(c10::complex<double>));
275
+ template <>
276
+ void dot<c10::complex<float>>(CUDABLAS_DOT_ARGTYPES(c10::complex<float>));
277
+
278
+ template <typename Dtype>
279
+ inline void vdot(CUDABLAS_DOT_ARGTYPES(Dtype)) {
280
+ static_assert(false&&sizeof(Dtype),"at::cuda::blas::vdot: not implemented");
281
+ }
282
+
283
+ template <>
284
+ void vdot<c10::complex<float>>(CUDABLAS_DOT_ARGTYPES(c10::complex<float>));
285
+ template <>
286
+ void vdot<c10::complex<double>>(CUDABLAS_DOT_ARGTYPES(c10::complex<double>));
287
+
288
+ #define CUDABLAS_GETRS_ARGTYPES(Dtype) \
289
+ cublasHandle_t handle, cublasOperation_t trans, \
290
+ int n, int nrhs, Dtype** dA_array, int lda, int* ipiv_array, \
291
+ Dtype** dB_array, int ldb, int* info_array, int batchsize
292
+
293
+ template<class Dtype>
294
+ void getrsBatched(CUDABLAS_GETRS_ARGTYPES(Dtype)) {
295
+ static_assert(false&&sizeof(Dtype),"at::cuda::blas::getrsBatched: not implemented");
296
+ }
297
+ template<>
298
+ TORCH_CUDA_CU_API void getrsBatched<float>(CUDABLAS_GETRS_ARGTYPES(float));
299
+ template<>
300
+ TORCH_CUDA_CU_API void getrsBatched<double>(CUDABLAS_GETRS_ARGTYPES(double));
301
+ template<>
302
+ TORCH_CUDA_CU_API void getrsBatched<c10::complex<float>>(CUDABLAS_GETRS_ARGTYPES(c10::complex<float>));
303
+ template<>
304
+ TORCH_CUDA_CU_API void getrsBatched<c10::complex<double>>(CUDABLAS_GETRS_ARGTYPES(c10::complex<double>));
305
+
306
+ #define CUDABLAS_GEQRF_BATCHED_ARGTYPES(Dtype) \
307
+ cublasHandle_t handle, int m, int n, Dtype **A_array, int lda, \
308
+ Dtype **tau_array, int *info, int batchsize
309
+
310
+ template <class Dtype>
311
+ void geqrfBatched(CUDABLAS_GEQRF_BATCHED_ARGTYPES(Dtype)) {
312
+ static_assert(false&&sizeof(Dtype), "at::cuda::blas::geqrfBatched: not implemented");
313
+ }
314
+ template <>
315
+ TORCH_CUDA_CU_API void geqrfBatched<float>(CUDABLAS_GEQRF_BATCHED_ARGTYPES(float));
316
+ template <>
317
+ TORCH_CUDA_CU_API void geqrfBatched<double>(CUDABLAS_GEQRF_BATCHED_ARGTYPES(double));
318
+ template <>
319
+ TORCH_CUDA_CU_API void geqrfBatched<c10::complex<double>>(
320
+ CUDABLAS_GEQRF_BATCHED_ARGTYPES(c10::complex<double>));
321
+ template <>
322
+ TORCH_CUDA_CU_API void geqrfBatched<c10::complex<float>>(
323
+ CUDABLAS_GEQRF_BATCHED_ARGTYPES(c10::complex<float>));
324
+
325
+ #define CUDABLAS_GETRF_ARGTYPES(Dtype) \
326
+ int n, Dtype** dA_array, int ldda, int* ipiv_array, int* info_array, int batchsize
327
+
328
+ template<class Dtype>
329
+ void getrfBatched(CUDABLAS_GETRF_ARGTYPES(Dtype)) {
330
+ TORCH_CHECK(false, "at::cuda::blas::getrfBatched: not implemented");
331
+ }
332
+ template<>
333
+ TORCH_CUDA_CU_API void getrfBatched<float>(CUDABLAS_GETRF_ARGTYPES(float));
334
+ template<>
335
+ TORCH_CUDA_CU_API void getrfBatched<double>(CUDABLAS_GETRF_ARGTYPES(double));
336
+ template<>
337
+ TORCH_CUDA_CU_API void getrfBatched<c10::complex<double>>(CUDABLAS_GETRF_ARGTYPES(c10::complex<double>));
338
+ template<>
339
+ TORCH_CUDA_CU_API void getrfBatched<c10::complex<float>>(CUDABLAS_GETRF_ARGTYPES(c10::complex<float>));
340
+
341
+ #define CUDABLAS_GELS_BATCHED_ARGTYPES(Dtype) \
342
+ cublasHandle_t handle, cublasOperation_t trans, int m, int n, int nrhs, Dtype** dA_array, int ldda, Dtype** dC_array, int lddc, int* info, int *devInfoArray, int batchSize
343
+
344
+ template <class Dtype>
345
+ void gelsBatched(CUDABLAS_GELS_BATCHED_ARGTYPES(Dtype)) {
346
+ static_assert(false&&sizeof(Dtype),"at::cuda::blas::gelsBatched: not implemented");
347
+ }
348
+
349
+ template<>
350
+ TORCH_CUDA_CU_API void gelsBatched<double>(CUDABLAS_GELS_BATCHED_ARGTYPES(double));
351
+ template<>
352
+ TORCH_CUDA_CU_API void gelsBatched<float>(CUDABLAS_GELS_BATCHED_ARGTYPES(float));
353
+ template<>
354
+ TORCH_CUDA_CU_API void gelsBatched<c10::complex<double>>(CUDABLAS_GELS_BATCHED_ARGTYPES(c10::complex<double>));
355
+ template<>
356
+ TORCH_CUDA_CU_API void gelsBatched<c10::complex<float>>(CUDABLAS_GELS_BATCHED_ARGTYPES(c10::complex<float>));
357
+
358
+ } // namespace at::cuda::blas
.venv/lib/python3.11/site-packages/torch/include/ATen/cuda/CUDAConfig.h ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ // Test these using #if AT_CUDNN_ENABLED(), not #ifdef, so that it's
4
+ // obvious if you forgot to include Config.h
5
+ // c.f. https://stackoverflow.com/questions/33759787/generating-an-error-if-checked-boolean-macro-is-not-defined
6
+ //
7
+ // NB: This header MUST NOT be included from other headers; it should
8
+ // only be included from C++ files.
9
+ #define AT_CUDNN_ENABLED() 1
10
+ #define AT_CUSPARSELT_ENABLED() 1
11
+ #define AT_ROCM_ENABLED() 0
12
+ #define AT_MAGMA_ENABLED() 1
13
+
14
+ // Needed for hipMAGMA to correctly identify implementation
15
+ #if (AT_ROCM_ENABLED() && AT_MAGMA_ENABLED())
16
+ #define HAVE_HIP 1
17
+ #endif
18
+
19
+ #define NVCC_FLAGS_EXTRA "-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90"
.venv/lib/python3.11/site-packages/torch/include/ATen/cuda/CUDAContext.h ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ #pragma once
2
+
3
+ #include <ATen/cuda/CUDAContextLight.h>
4
+
5
+ // Preserved for BC, as many files depend on these includes
6
+ #include <ATen/Context.h>
7
+ #include <c10/cuda/CUDAStream.h>
8
+ #include <c10/util/Logging.h>
9
+ #include <ATen/cuda/Exceptions.h>