codegen/native_functions.yaml in torch-rb-0.12.2 vs codegen/native_functions.yaml in torch-rb-0.13.0

- old
+ new

@@ -242,18 +242,19 @@ variants: function dispatch: CPU: native_dropout_cpu CUDA: native_dropout_cuda NestedTensorCPU, NestedTensorCUDA: native_dropout_nested - tags: nondeterministic_seeded + tags: [nondeterministic_seeded, core] autogen: native_dropout.out - func: native_dropout_backward(Tensor grad_output, Tensor mask, float scale) -> Tensor dispatch: CPU, NestedTensorCPU, NestedTensorCUDA: native_dropout_backward CUDA: native_dropout_backward_cuda autogen: native_dropout_backward.out + tags: pointwise - func: _sobol_engine_draw(Tensor quasi, int n, Tensor sobolstate, int dimension, int num_generated, ScalarType? dtype) -> (Tensor, Tensor) - func: _sobol_engine_ff_(Tensor(a!) self, int n, Tensor sobolstate, int dimension, int num_generated) -> Tensor(a!) @@ -294,10 +295,11 @@ variants: function, method dispatch: CompositeExplicitAutograd: abs SparseCPU, SparseCUDA: abs_sparse SparseCsrCPU, SparseCsrCUDA: abs_sparse_csr + tags: [core, pointwise] - func: abs_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method dispatch: @@ -310,10 +312,11 @@ dispatch: CPU, CUDA: abs_out MPS: abs_out_mps SparseCPU, SparseCUDA: abs_sparse_out SparseCsrCPU, SparseCsrCUDA: abs_sparse_csr_out + tags: pointwise # Note [Adding an alias] # To add an alias do the following: # # 1) Copy the original functions native_functions.yaml entry, but replace the @@ -333,12 +336,12 @@ # 5) Update the alias_map in torch/csrc/jit/passes/normalize_ops.cpp. # 6) Add aliases argument to existing OpInfo/UnaryUfuncInfo or create new OpInfo/UnaryUfuncInfo entry # in op_db list in torch/testing/_internal/common_methods_invocations.py # # See torch.absolute, an alias for torch.abs, as an example. - # Absolute, alias for abs + - func: absolute(Tensor self) -> Tensor device_check: NoCheck # TensorIterator variants: function, method - func: absolute_(Tensor(a!) self) -> Tensor(a!) @@ -352,16 +355,18 @@ device_check: NoCheck # TensorIterator variants: function, method dispatch: CPU, CUDA: angle SparseCsrCPU, SparseCsrCUDA: angle_sparse_csr + tags: pointwise - func: angle.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: angle_out SparseCsrCPU, SparseCsrCUDA: angle_sparse_csr_out + tags: pointwise - func: view_as_real(Tensor(a) self) -> Tensor(a) variants: function dispatch: CPU, CUDA, MPS, Meta: view_as_real @@ -375,25 +380,28 @@ variants: function, method structured_delegate: sgn.out dispatch: SparseCPU, SparseCUDA: sgn_sparse SparseCsrCPU, SparseCsrCUDA: sgn_sparse_csr + tags: pointwise - func: sgn_(Tensor(a!) self) -> Tensor(a!) variants: method structured_delegate: sgn.out dispatch: SparseCPU, SparseCUDA: sgn_sparse_ SparseCsrCPU, SparseCsrCUDA: sgn_sparse_csr_ + tags: pointwise - func: sgn.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: sgn_out SparseCPU, SparseCUDA: sgn_sparse_out SparseCsrCPU, SparseCsrCUDA: sgn_sparse_csr_out + tags: pointwise - func: chalf(Tensor self, *, MemoryFormat? memory_format=None) -> Tensor variants: method - func: real(Tensor(a) self) -> Tensor(a) @@ -420,22 +428,25 @@ SparseCsrCPU, SparseCsrCUDA: conj_physical_sparse_csr autogen: _conj_physical.out - func: conj_physical(Tensor self) -> Tensor variants: function, method + tags: pointwise - func: conj_physical.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: conj_physical_out SparseCPU, SparseCUDA: conj_physical_out_sparse SparseCsrCPU, SparseCsrCUDA: conj_physical_sparse_csr_out + tags: pointwise - func: conj_physical_(Tensor(a!) self) -> Tensor(a!) variants: function, method dispatch: CompositeExplicitAutograd: conj_physical_ SparseCsrCPU, SparseCsrCUDA: conj_physical_sparse_csr_ + tags: pointwise - func: resolve_conj(Tensor(a) self) -> Tensor(a) variants: function, method - func: resolve_neg(Tensor(a) self) -> Tensor(a) @@ -448,23 +459,26 @@ - func: acos(Tensor self) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: acos.out + tags: [core, pointwise] - func: acos_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method structured_delegate: acos.out + tags: pointwise - func: acos.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: acos_out MPS: acos_out_mps + tags: pointwise # arccos, alias of acos - func: arccos(Tensor self) -> Tensor variants: function, method @@ -488,20 +502,22 @@ SparseCPU, SparseCUDA: add_sparse SparseCsrCPU, SparseCsrCUDA: add_sparse_csr MkldnnCPU: mkldnn_add ZeroTensor: add_zerotensor NestedTensorCPU, NestedTensorCUDA: NestedTensor_add_Tensor + tags: [core, pointwise] - func: add_.Tensor(Tensor(a!) self, Tensor other, *, Scalar alpha=1) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: add.out dispatch: SparseCPU, SparseCUDA: add_sparse_ SparseCsrCPU, SparseCsrCUDA: add_sparse_csr_ MkldnnCPU: mkldnn_add_ NestedTensorCPU, NestedTensorCUDA: NestedTensor_add__Tensor + tags: pointwise - func: add.out(Tensor self, Tensor other, *, Scalar alpha=1, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase @@ -513,10 +529,11 @@ SparseCUDA: add_out_sparse_cuda SparseCsrCPU: add_out_sparse_csr_cpu SparseCsrCUDA: add_out_sparse_csr_cuda MkldnnCPU: mkldnn_add_out MPS: add_out_mps + tags: pointwise - func: _add_relu.Tensor(Tensor self, Tensor other, *, Scalar alpha=1) -> Tensor variants: function dispatch: CPU: add_relu @@ -546,17 +563,19 @@ - func: add.Scalar(Tensor self, Scalar other, Scalar alpha=1) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: add + tags: [core, pointwise] - func: add_.Scalar(Tensor(a!) self, Scalar other, Scalar alpha=1) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: add_ autogen: add.Scalar_out + tags: pointwise - func: addmv(Tensor self, Tensor mat, Tensor vec, *, Scalar beta=1, Scalar alpha=1) -> Tensor structured_delegate: addmv.out variants: function, method @@ -575,20 +594,22 @@ - func: addr(Tensor self, Tensor vec1, Tensor vec2, *, Scalar beta=1, Scalar alpha=1) -> Tensor variants: function, method dispatch: CPU, CUDA: addr + MPS: addr_mps CompositeExplicitAutograd: math_addr - func: addr_(Tensor(a!) self, Tensor vec1, Tensor vec2, *, Scalar beta=1, Scalar alpha=1) -> Tensor(a!) variants: method dispatch: CompositeExplicitAutograd: addr_ - func: addr.out(Tensor self, Tensor vec1, Tensor vec2, *, Scalar beta=1, Scalar alpha=1, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: addr_out + MPS: addr_out_mps CompositeExplicitAutograd: math_addr_out - func: affine_grid_generator(Tensor theta, int[] size, bool align_corners) -> Tensor variants: function dispatch: @@ -596,10 +617,24 @@ autogen: affine_grid_generator.out - func: affine_grid_generator_backward(Tensor grad, int[] size, bool align_corners) -> Tensor variants: function +- func: _is_all_true(Tensor self) -> Tensor + variants: function, method + dispatch: + CompositeExplicitAutograd: _is_all_true + +- func: _is_any_true(Tensor self) -> Tensor + variants: function, method + dispatch: + CompositeExplicitAutograd: _is_any_true + +# Note: this function is only for testing. +- func: _test_check_tensor(Tensor self) -> Tensor + variants: function + - func: all.dim(Tensor self, int dim, bool keepdim=False) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: all.out variants: function, method @@ -663,10 +698,11 @@ - func: arange.start_step(Scalar start, Scalar end, Scalar step=1, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor dispatch: CompositeExplicitAutograd: arange cpp_no_default_args: ['step'] + tags: core - func: arange.out(Scalar end, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: arange_out @@ -686,10 +722,11 @@ - func: argmax(Tensor self, int? dim=None, bool keepdim=False) -> Tensor structured_delegate: argmax.out device_check: NoCheck # TensorIterator variants: function, method + tags: core - func: argmax.out(Tensor self, int? dim=None, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!) structured: True dispatch: CPU, CUDA: argmax_out @@ -697,33 +734,37 @@ - func: argmin(Tensor self, int? dim=None, bool keepdim=False) -> Tensor structured_delegate: argmin.out device_check: NoCheck # TensorIterator variants: function, method + tags: core - func: argmin.out(Tensor self, int? dim=None, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!) structured: True dispatch: CPU, CUDA: argmin_out MPS: argmin_out_mps - func: acosh(Tensor self) -> Tensor variants: function, method structured_delegate: acosh.out + tags: [core, pointwise] - func: acosh_(Tensor(a!) self) -> Tensor(a!) variants: function, method structured_delegate: acosh.out + tags: pointwise - func: acosh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: acosh_out MPS: acosh_out_mps - + tags: pointwise # arccosh, alias for acosh + - func: arccosh(Tensor self) -> Tensor variants: function, method - func: arccosh_(Tensor(a!) self) -> Tensor(a!) variants: function, method @@ -734,26 +775,29 @@ variants: function, method structured_delegate: asinh.out dispatch: SparseCPU, SparseCUDA: asinh_sparse SparseCsrCPU, SparseCsrCUDA: asinh_sparse_csr + tags: [core, pointwise] - func: asinh_(Tensor(a!) self) -> Tensor(a!) variants: function, method structured_delegate: asinh.out dispatch: SparseCPU, SparseCUDA: asinh_sparse_ SparseCsrCPU, SparseCsrCUDA: asinh_sparse_csr_ + tags: pointwise - func: asinh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: asinh_out MPS: asinh_out_mps SparseCPU, SparseCUDA: asinh_sparse_out SparseCsrCPU, SparseCsrCUDA: asinh_sparse_csr_out + tags: pointwise # arcsinh, alias for asinh - func: arcsinh(Tensor self) -> Tensor variants: function, method @@ -766,28 +810,31 @@ structured_delegate: atanh.out variants: function, method dispatch: SparseCPU, SparseCUDA: atanh_sparse SparseCsrCPU, SparseCsrCUDA: atanh_sparse_csr + tags: [core, pointwise] - func: atanh_(Tensor(a!) self) -> Tensor(a!) structured_delegate: atanh.out variants: function, method dispatch: SparseCPU, SparseCUDA: atanh_sparse_ SparseCsrCPU, SparseCsrCUDA: atanh_sparse_csr_ + tags: pointwise - func: atanh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: atanh_out MPS: atanh_out_mps SparseCPU, SparseCUDA: atanh_sparse_out SparseCsrCPU, SparseCsrCUDA: atanh_sparse_csr_out - + tags: pointwise # arctanh, alias for atanh + - func: arctanh(Tensor self) -> Tensor variants: function, method - func: arctanh_(Tensor(a!) self) -> Tensor(a!) variants: function, method @@ -801,45 +848,49 @@ Meta: as_strided_tensorimpl_meta_symint MPS: as_strided_tensorimpl_mps QuantizedCPU, QuantizedCUDA: as_strided_qtensorimpl device_check: NoCheck device_guard: False + tags: core - func: as_strided_(Tensor(a!) self, SymInt[] size, SymInt[] stride, SymInt? storage_offset=None) -> Tensor(a!) use_const_ref_for_mutable_tensors: True variants: function, method device_check: NoCheck device_guard: False tags: inplace_view dispatch: - CompositeExplicitAutogradNonFunctional: as_strided_ + CompositeExplicitAutogradNonFunctional: as_strided__symint - func: asin(Tensor self) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: asin.out dispatch: SparseCPU, SparseCUDA: asin_sparse SparseCsrCPU, SparseCsrCUDA: asin_sparse_csr + tags: [core, pointwise] - func: asin_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method structured_delegate: asin.out dispatch: SparseCPU, SparseCUDA: asin_sparse_ SparseCsrCPU, SparseCsrCUDA: asin_sparse_csr_ + tags: pointwise - func: asin.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: asin_out MPS: asin_out_mps SparseCPU, SparseCUDA: asin_sparse_out SparseCsrCPU, SparseCsrCUDA: asin_sparse_csr_out + tags: pointwise # arcsin, alias of asin - func: arcsin(Tensor self) -> Tensor variants: function, method @@ -853,28 +904,31 @@ structured_delegate: atan.out variants: function, method dispatch: SparseCPU, SparseCUDA: atan_sparse SparseCsrCPU, SparseCsrCUDA: atan_sparse_csr + tags: [core, pointwise] - func: atan_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: atan.out variants: function, method dispatch: SparseCPU, SparseCUDA: atan_sparse_ SparseCsrCPU, SparseCsrCUDA: atan_sparse_csr_ + tags: pointwise - func: atan.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: atan_out MPS: atan_out_mps SparseCPU, SparseCUDA: atan_sparse_out SparseCsrCPU, SparseCsrCUDA: atan_sparse_csr_out + tags: pointwise # arctan, alias of atan - func: arctan(Tensor self) -> Tensor variants: function, method @@ -979,10 +1033,12 @@ # with `bernoulli(Tensor self, *, Generator? generator=None)` declaration. - func: bernoulli.p(Tensor self, float p, *, Generator? generator=None) -> Tensor device_check: NoCheck # TensorIterator variants: function, method tags: nondeterministic_seeded + dispatch: + CompositeExplicitAutogradNonFunctional: bernoulli - func: bilinear(Tensor input1, Tensor input2, Tensor weight, Tensor? bias=None) -> Tensor - func: binary_cross_entropy(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean) -> Tensor device_check: NoCheck # TensorIterator @@ -1028,132 +1084,152 @@ - func: bincount(Tensor self, Tensor? weights=None, int minlength=0) -> Tensor variants: function, method dispatch: CPU: _bincount_cpu CUDA: _bincount_cuda + MPS: _bincount_mps tags: dynamic_output_shape autogen: bincount.out - func: bitwise_not(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: bitwise_not.out variants: function, method + tags: [core, pointwise] - func: bitwise_not_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: bitwise_not.out variants: method + tags: pointwise - func: bitwise_not.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: bitwise_not_out + tags: pointwise - func: copysign.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: copysign_out + tags: pointwise - func: copysign.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: copysign.out + tags: pointwise - func: copysign_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: copysign.out - func: copysign.Scalar(Tensor self, Scalar other) -> Tensor variants: function, method dispatch: CompositeExplicitAutograd: copysign + tags: pointwise - func: copysign_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) variants: method dispatch: CompositeExplicitAutograd: copysign_ - func: copysign.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: copysign_out + tags: pointwise - func: logical_not(Tensor self) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: logical_not + tags: [core, pointwise] - func: logical_not_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: logical_not_ + tags: pointwise - func: logical_not.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: logical_not_out MPS: logical_not_out_mps + tags: pointwise - func: logical_xor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: logical_xor + tags: pointwise - func: logical_xor_(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: logical_xor_ + tags: pointwise - func: logical_xor.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: logical_xor_out MPS: logical_xor_out_mps + tags: pointwise - func: logical_and(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: logical_and + tags: [core, pointwise] - func: logical_and_(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: logical_and_ + tags: pointwise - func: logical_and.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: logical_and_out MPS: logical_and_out_mps + tags: pointwise - func: logical_or(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: logical_or + tags: [core, pointwise] - func: logical_or_(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: logical_or_ + tags: pointwise - func: logical_or.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: logical_or_out MPS: logical_or_out_mps + tags: pointwise - func: blackman_window(int window_length, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor dispatch: CompositeExplicitAutograd: blackman_window autogen: blackman_window.out @@ -1167,11 +1243,13 @@ structured_delegate: bmm.out variants: function, method dispatch: SparseCPU: bmm_sparse_cpu SparseCUDA: bmm_sparse_cuda - NestedTensorCPU, NestedTensorCUDA: bmm_nested + NestedTensorCPU: bmm_nested + NestedTensorCUDA: bmm_nested_cuda + tags: core - func: bmm.out(Tensor self, Tensor mat2, *, Tensor(a!) out) -> Tensor(a!) structured: True variants: function dispatch: @@ -1184,12 +1262,14 @@ - func: broadcast_tensors(Tensor[] tensors) -> Tensor[] device_check: NoCheck device_guard: False -- func: broadcast_to(Tensor(a) self, int[] size) -> Tensor(a) +- func: broadcast_to(Tensor(a) self, SymInt[] size) -> Tensor(a) variants: function, method + dispatch: + CompositeImplicitAutograd: broadcast_to_symint - func: _sparse_broadcast_to(Tensor(a) self, int[] size) -> Tensor(a) variants: function dispatch: SparseCPU, SparseCUDA: sparse_broadcast_to @@ -1197,10 +1277,11 @@ - func: cat(Tensor[] tensors, int dim=0) -> Tensor structured_delegate: cat.out dispatch: SparseCPU, SparseCUDA: cat_sparse QuantizedCPU: cat_quantized_cpu + tags: core - func: cat.out(Tensor[] tensors, int dim=0, *, Tensor(a!) out) -> Tensor(a!) structured: True precomputed: - dim -> int dim, int valid, bool all_contiguous, bool all_same_dtype, bool all_same_sizes_and_stride, MemoryFormat memory_format @@ -1243,28 +1324,31 @@ structured_delegate: ceil.out variants: function, method dispatch: SparseCPU, SparseCUDA: ceil_sparse SparseCsrCPU, SparseCsrCUDA: ceil_sparse_csr + tags: pointwise - func: ceil_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: ceil.out variants: function, method dispatch: SparseCPU, SparseCUDA: ceil_sparse_ SparseCsrCPU, SparseCsrCUDA: ceil_sparse_csr_ + tags: pointwise - func: ceil.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: ceil_out MPS: ceil_out_mps SparseCPU, SparseCUDA: ceil_sparse_out SparseCsrCPU, SparseCsrCUDA: ceil_sparse_csr_out + tags: pointwise # alias for torch.linalg.multi_dot - func: chain_matmul(Tensor[] matrices) -> Tensor variants: function @@ -1278,16 +1362,23 @@ - func: chunk(Tensor(a -> *) self, int chunks, int dim=0) -> Tensor(a)[] variants: function, method device_check: NoCheck device_guard: False + dispatch: + CompositeImplicitAutograd: chunk + NestedTensorCPU, NestedTensorCUDA: chunk_nested_tensor -- func: tensor_split.sections(Tensor(a -> *) self, int sections, int dim=0) -> Tensor(a)[] +- func: tensor_split.sections(Tensor(a -> *) self, SymInt sections, int dim=0) -> Tensor(a)[] variants: function, method + dispatch: + CompositeImplicitAutograd: tensor_split_sections_symint -- func: tensor_split.indices(Tensor(a -> *) self, int[] indices, int dim=0) -> Tensor(a)[] +- func: tensor_split.indices(Tensor(a -> *) self, SymInt[] indices, int dim=0) -> Tensor(a)[] variants: function, method + dispatch: + CompositeImplicitAutograd: tensor_split_indices_symint - func: tensor_split.tensor_indices_or_sections(Tensor(a -> *) self, Tensor tensor_indices_or_sections, int dim=0) -> Tensor(a)[] variants: function, method - func: clamp(Tensor self, Scalar? min=None, Scalar? max=None) -> Tensor @@ -1295,127 +1386,150 @@ variants: function, method cpp_no_default_args: ['min'] structured_delegate: clamp.out dispatch: QuantizedCPU: clamp_quantized_cpu + tags: [core, pointwise] - func: clamp.Tensor(Tensor self, Tensor? min=None, Tensor? max=None) -> Tensor variants: function, method structured_delegate: clamp.Tensor_out + tags: pointwise - func: clamp_(Tensor(a!) self, Scalar? min=None, Scalar? max=None) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method cpp_no_default_args: ['min'] structured_delegate: clamp.out + tags: pointwise - func: clamp_.Tensor(Tensor(a!) self, Tensor? min=None, Tensor? max=None) -> Tensor(a!) variants: function, method structured_delegate: clamp.Tensor_out + tags: pointwise - func: clamp.out(Tensor self, Scalar? min=None, Scalar? max=None, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator cpp_no_default_args: ['min'] structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: clamp_out MPS: clamp_out_mps + tags: pointwise - func: clamp.Tensor_out(Tensor self, Tensor? min=None, Tensor? max=None, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: clamp_Tensor_out MPS: clamp_Tensor_out_mps + tags: pointwise - func: clamp_max(Tensor self, Scalar max) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: clamp_max.out + tags: pointwise - func: clamp_max.Tensor(Tensor self, Tensor max) -> Tensor variants: function, method structured_delegate: clamp_max.Tensor_out + tags: pointwise - func: clamp_max_(Tensor(a!) self, Scalar max) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method structured_delegate: clamp_max.out + tags: pointwise - func: clamp_max_.Tensor(Tensor(a!) self, Tensor max) -> Tensor(a!) variants: function, method structured_delegate: clamp_max.Tensor_out + tags: pointwise - func: clamp_max.out(Tensor self, Scalar max, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: clamp_max_out MPS: clamp_max_out_mps + tags: pointwise - func: clamp_max.Tensor_out(Tensor self, Tensor max, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: clamp_max_Tensor_out MPS: clamp_max_Tensor_out_mps + tags: pointwise - func: clamp_min(Tensor self, Scalar min) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: clamp_min.out + tags: pointwise - func: clamp_min.Tensor(Tensor self, Tensor min) -> Tensor variants: function, method structured_delegate: clamp_min.Tensor_out + tags: pointwise - func: clamp_min_(Tensor(a!) self, Scalar min) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method structured_delegate: clamp_min.out + tags: pointwise - func: clamp_min_.Tensor(Tensor(a!) self, Tensor min) -> Tensor(a!) variants: function, method structured_delegate: clamp_min.Tensor_out + tags: pointwise - func: clamp_min.out(Tensor self, Scalar min, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: clamp_min_out MPS: clamp_min_out_mps + tags: pointwise - func: clamp_min.Tensor_out(Tensor self, Tensor min, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: clamp_min_Tensor_out MPS: clamp_min_Tensor_out_mps + tags: pointwise # clip is an alias for clamp - func: clip(Tensor self, Scalar? min=None, Scalar? max=None) -> Tensor cpp_no_default_args: ['min'] variants: function, method + tags: pointwise - func: clip.Tensor(Tensor self, Tensor? min=None, Tensor? max=None) -> Tensor variants: function, method + tags: pointwise - func: clip_(Tensor(a!) self, Scalar? min=None, Scalar? max=None) -> Tensor(a!) cpp_no_default_args: ['min'] variants: function, method + tags: pointwise - func: clip_.Tensor(Tensor(a!) self, Tensor? min=None, Tensor? max=None) -> Tensor(a!) variants: function, method + tags: pointwise - func: clip.out(Tensor self, Scalar? min=None, Scalar? max=None, *, Tensor(a!) out) -> Tensor(a!) cpp_no_default_args: ['min'] + tags: pointwise - func: clip.Tensor_out(Tensor self, Tensor? min=None, Tensor? max=None, *, Tensor(a!) out) -> Tensor(a!) - func: cudnn_is_acceptable(Tensor self) -> bool device_check: NoCheck @@ -1437,30 +1551,33 @@ - func: polar.out(Tensor abs, Tensor angle, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: polar_out -- func: constant_pad_nd(Tensor self, int[] pad, Scalar value=0) -> Tensor +- func: constant_pad_nd(Tensor self, SymInt[] pad, Scalar value=0) -> Tensor variants: function dispatch: CompositeExplicitAutograd: constant_pad_nd MPS: constant_pad_nd_mps autogen: constant_pad_nd.out + tags: core - func: contiguous(Tensor(a) self, *, MemoryFormat memory_format=contiguous_format) -> Tensor(a) variants: method manual_cpp_binding: True -- func: convolution(Tensor input, Tensor weight, Tensor? bias, int[] stride, int[] padding, int[] dilation, bool transposed, int[] output_padding, int groups) -> Tensor +- func: convolution(Tensor input, Tensor weight, Tensor? bias, int[] stride, SymInt[] padding, int[] dilation, bool transposed, SymInt[] output_padding, int groups) -> Tensor dispatch: CompositeExplicitAutograd: convolution autogen: convolution.out + tags: core -- func: convolution_backward(Tensor grad_output, Tensor input, Tensor weight, SymInt[]? bias_sizes, int[] stride, int[] padding, int[] dilation, bool transposed, int[] output_padding, int groups, bool[3] output_mask) -> (Tensor, Tensor, Tensor) +- func: convolution_backward(Tensor grad_output, Tensor input, Tensor weight, SymInt[]? bias_sizes, int[] stride, SymInt[] padding, int[] dilation, bool transposed, SymInt[] output_padding, int groups, bool[3] output_mask) -> (Tensor, Tensor, Tensor) dispatch: CompositeExplicitAutograd, CUDA: convolution_backward autogen: convolution_backward.out + tags: core - func: convolution_overrideable(Tensor input, Tensor weight, Tensor? bias, int[] stride, int[] padding, int[] dilation, bool transposed, int[] output_padding, int groups) -> Tensor dispatch: CompositeExplicitAutograd: convolution_overrideable autogen: convolution_overrideable.out @@ -1468,20 +1585,20 @@ - func: convolution_backward_overrideable(Tensor grad_output, Tensor input, Tensor weight, int[] stride, int[] padding, int[] dilation, bool transposed, int[] output_padding, int groups, bool[3] output_mask) -> (Tensor grad_input, Tensor grad_weight, Tensor grad_bias) dispatch: CompositeExplicitAutograd: convolution_backward_overrideable autogen: convolution_backward_overrideable.out -- func: _convolution(Tensor input, Tensor weight, Tensor? bias, int[] stride, int[] padding, int[] dilation, bool transposed, int[] output_padding, int groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32) -> Tensor +- func: _convolution(Tensor input, Tensor weight, Tensor? bias, int[] stride, SymInt[] padding, int[] dilation, bool transposed, SymInt[] output_padding, int groups, bool benchmark, bool deterministic, bool cudnn_enabled, bool allow_tf32) -> Tensor dispatch: CompositeExplicitAutograd: _convolution autogen: _convolution.out - func: _convolution.deprecated(Tensor input, Tensor weight, Tensor? bias, int[] stride, int[] padding, int[] dilation, bool transposed, int[] output_padding, int groups, bool benchmark, bool deterministic, bool cudnn_enabled) -> Tensor - func: _convolution_mode(Tensor input, Tensor weight, Tensor? bias, int[] stride, str padding, int[] dilation, int groups) -> Tensor -- func: _convolution_double_backward(Tensor? ggI, Tensor? ggW, Tensor? ggb, Tensor gO, Tensor weight, Tensor self, int[] stride, int[] padding, int[] dilation, bool transposed, int[] output_padding, int groups, bool[3] output_mask) -> (Tensor, Tensor, Tensor) +- func: _convolution_double_backward(Tensor? ggI, Tensor? ggW, Tensor? ggb, Tensor gO, Tensor weight, Tensor self, int[] stride, SymInt[] padding, int[] dilation, bool transposed, SymInt[] output_padding, int groups, bool[3] output_mask) -> (Tensor, Tensor, Tensor) - func: conv1d(Tensor input, Tensor weight, Tensor? bias=None, int[1] stride=1, int[1] padding=0, int[1] dilation=1, int groups=1) -> Tensor - func: conv2d(Tensor input, Tensor weight, Tensor? bias=None, int[2] stride=1, int[2] padding=0, int[2] dilation=1, int groups=1) -> Tensor @@ -1510,20 +1627,23 @@ - func: conv_transpose3d.input(Tensor input, Tensor weight, Tensor? bias=None, int[3] stride=1, int[3] padding=0, int[3] output_padding=0, int groups=1, int[3] dilation=1) -> Tensor - func: copy(Tensor self, Tensor src, bool non_blocking=False) -> Tensor variants: function + dispatch: + CompositeExplicitAutogradNonFunctional: copy - func: copy_(Tensor(a!) self, Tensor src, bool non_blocking=False) -> Tensor(a!) variants: method device_check: NoCheck device_guard: False dispatch: MkldnnCPU: copy_mkldnn_ SparseCPU, SparseCUDA: copy_sparse_wrapper_ CompositeExplicitAutograd: copy_ SparseCsrCPU, SparseCsrCUDA: copy_sparse_compressed_ + NestedTensorCPU, NestedTensorCUDA: copy_nested_ autogen: copy.out - func: _copy_from(Tensor self, Tensor dst, bool non_blocking=False) -> Tensor dispatch: MPS: _copy_from_mps @@ -1538,41 +1658,47 @@ - func: cos(Tensor self) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: cos.out + tags: [core, pointwise] - func: cos_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method structured_delegate: cos.out + tags: pointwise - func: cos.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: cos_out MPS: cos_out_mps + tags: pointwise - func: cosh(Tensor self) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: cosh.out + tags: [core, pointwise] - func: cosh_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method structured_delegate: cosh.out + tags: pointwise - func: cosh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: cosh_out MPS: cosh_out_mps + tags: pointwise - func: cosine_embedding_loss(Tensor input1, Tensor input2, Tensor target, float margin=0.0, int reduction=Mean) -> Tensor - func: count_nonzero.dim_IntList(Tensor self, int[] dim) -> Tensor variants: function, method @@ -1752,10 +1878,11 @@ - func: cumsum.out(Tensor self, int dim, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!) structured: True device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: cumsum_out + MPS: cumsum_out_mps - func: cumsum.dimname(Tensor self, Dimname dim, *, ScalarType? dtype=None) -> Tensor device_check: NoCheck # TensorIterator variants: function, method @@ -1777,14 +1904,17 @@ - func: _ctc_loss(Tensor log_probs, Tensor targets, int[] input_lengths, int[] target_lengths, int blank=0, bool zero_infinity=False) -> (Tensor, Tensor) dispatch: CPU: ctc_loss_cpu CUDA: ctc_loss_gpu autogen: _ctc_loss.out + tags: dynamic_output_shape # the shape of second output is data dependent - func: _ctc_loss.Tensor(Tensor log_probs, Tensor targets, Tensor input_lengths, Tensor target_lengths, int blank=0, bool zero_infinity=False) -> (Tensor, Tensor) dispatch: CPU, CUDA: ctc_loss_tensor + autogen: _ctc_loss.Tensor_out + tags: dynamic_output_shape # the shape of second output is data dependent - func: _ctc_loss_backward(Tensor grad, Tensor log_probs, Tensor targets, int[] input_lengths, int[] target_lengths, Tensor neg_log_likelihood, Tensor log_alpha, int blank, bool zero_infinity=False) -> Tensor dispatch: CPU: ctc_loss_backward_cpu CUDA: ctc_loss_backward_gpu @@ -1795,11 +1925,11 @@ CPU, CUDA: ctc_loss_backward_tensor - func: diag_embed(Tensor self, int offset=0, int dim1=-2, int dim2=-1) -> Tensor variants: function, method dispatch: - CompositeExplicitAutograd: diag_embed + CompositeExplicitAutogradNonFunctional: diag_embed autogen: diag_embed.out - func: diagflat(Tensor self, int offset=0) -> Tensor variants: function, method @@ -1858,74 +1988,86 @@ variants: function, method structured_delegate: div.out dispatch: SparseCPU, SparseCUDA: div_sparse ZeroTensor: div_zerotensor + NestedTensorCPU, NestedTensorCUDA: NestedTensor_div_Tensor + tags: [core, pointwise] - func: div_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: div.out dispatch: SparseCPU, SparseCUDA: div_sparse_ + tags: pointwise - func: div.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: div_out MPS: div_out_mps SparseCPU, SparseCUDA: div_out_sparse_zerodim + tags: pointwise - func: div.Tensor_mode(Tensor self, Tensor other, *, str? rounding_mode) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: div.out_mode dispatch: SparseCPU, SparseCUDA: div_sparse + tags: pointwise - func: div_.Tensor_mode(Tensor(a!) self, Tensor other, *, str? rounding_mode) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: div.out_mode dispatch: SparseCPU, SparseCUDA: div_sparse_ + tags: pointwise - func: div.out_mode(Tensor self, Tensor other, *, str? rounding_mode, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: div_out_mode MPS: div_out_mode_mps SparseCPU, SparseCUDA: div_out_sparse_zerodim + tags: pointwise # For C++ only, until we have conversion from C++ numbers to Tensor - func: div.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: div + NestedTensorCPU, NestedTensorCUDA: NestedTensor_div_Scalar + tags: [core, pointwise] - func: div_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: div_ autogen: div.Scalar_out + tags: pointwise - func: div.Scalar_mode(Tensor self, Scalar other, *, str? rounding_mode) -> Tensor variants: function, method dispatch: CompositeExplicitAutograd: div + tags: pointwise - func: div_.Scalar_mode(Tensor(a!) self, Scalar other, *, str? rounding_mode) -> Tensor(a!) variants: method dispatch: CompositeExplicitAutograd: div_ autogen: div.Scalar_mode_out + tags: pointwise # divide, alias for div - func: divide.Tensor(Tensor self, Tensor other) -> Tensor variants: function, method @@ -1956,10 +2098,11 @@ # true_divide, an alias for div - func: true_divide.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method + tags: pointwise - func: true_divide_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method @@ -1995,26 +2138,27 @@ dispatch: CompositeExplicitAutograd: vdot_out - func: einsum(str equation, Tensor[] tensors, *, int[]? path=None) -> Tensor -- func: embedding(Tensor weight, Tensor indices, int padding_idx=-1, bool scale_grad_by_freq=False, bool sparse=False) -> Tensor +- func: embedding(Tensor weight, Tensor indices, SymInt padding_idx=-1, bool scale_grad_by_freq=False, bool sparse=False) -> Tensor dispatch: - CompositeExplicitAutograd: embedding + CompositeExplicitAutograd: embedding_symint NestedTensorCPU, NestedTensorCUDA: NestedTensor_embedding autogen: embedding.out -- func: embedding_backward(Tensor grad, Tensor indices, SymInt num_weights, int padding_idx, bool scale_grad_by_freq, bool sparse) -> Tensor +- func: embedding_backward(Tensor grad, Tensor indices, SymInt num_weights, SymInt padding_idx, bool scale_grad_by_freq, bool sparse) -> Tensor dispatch: CompositeImplicitAutograd: embedding_backward_symint -- func: embedding_dense_backward(Tensor grad_output, Tensor indices, SymInt num_weights, int padding_idx, bool scale_grad_by_freq) -> Tensor +- func: embedding_dense_backward(Tensor grad_output, Tensor indices, SymInt num_weights, SymInt padding_idx, bool scale_grad_by_freq) -> Tensor dispatch: CPU: embedding_dense_backward_cpu CUDA: embedding_dense_backward_cuda MPS: embedding_dense_backward_mps autogen: embedding_dense_backward.out + tags: core - func: embedding_renorm_(Tensor(a!) self, Tensor indices, float max_norm, float norm_type) -> Tensor(a!) dispatch: CPU: embedding_renorm_cpu_ CUDA: embedding_renorm_cuda_ @@ -2057,15 +2201,19 @@ dispatch: CPU: _embedding_bag_cpu CUDA: _embedding_bag_cuda autogen: _embedding_bag.out -- func: _embedding_bag_backward(Tensor grad, Tensor indices, Tensor offsets, Tensor offset2bag, Tensor bag_size, Tensor maximum_indices, int num_weights, bool scale_grad_by_freq, int mode, bool sparse, Tensor? per_sample_weights, int padding_idx=-1) -> Tensor +- func: _embedding_bag_backward(Tensor grad, Tensor indices, Tensor offsets, Tensor offset2bag, Tensor bag_size, Tensor maximum_indices, SymInt num_weights, bool scale_grad_by_freq, int mode, bool sparse, Tensor? per_sample_weights, int padding_idx=-1) -> Tensor + dispatch: + CompositeImplicitAutograd: _embedding_bag_backward_symint -- func: _embedding_bag_sparse_backward(Tensor grad, Tensor indices, Tensor offsets, Tensor offset2bag, Tensor bag_size, int num_weights, bool scale_grad_by_freq, int mode, Tensor? per_sample_weights, int padding_idx=-1) -> Tensor +- func: _embedding_bag_sparse_backward(Tensor grad, Tensor indices, Tensor offsets, Tensor offset2bag, Tensor bag_size, SymInt num_weights, bool scale_grad_by_freq, int mode, Tensor? per_sample_weights, int padding_idx=-1) -> Tensor + dispatch: + CompositeImplicitAutograd: _embedding_bag_sparse_backward_symint -- func: _embedding_bag_dense_backward(Tensor grad, Tensor indices, Tensor offset2bag, Tensor bag_size, Tensor maximum_indices, int num_weights, bool scale_grad_by_freq, int mode, Tensor? per_sample_weights, int padding_idx=-1) -> Tensor +- func: _embedding_bag_dense_backward(Tensor grad, Tensor indices, Tensor offset2bag, Tensor bag_size, Tensor maximum_indices, SymInt num_weights, bool scale_grad_by_freq, int mode, Tensor? per_sample_weights, int padding_idx=-1) -> Tensor dispatch: CPU: _embedding_bag_dense_backward_cpu CUDA: _embedding_bag_dense_backward_cuda autogen: _embedding_bag_dense_backward.out @@ -2152,11 +2300,12 @@ variants: method device_check: NoCheck device_guard: False tags: inplace_view dispatch: - CPU, Meta: resize_ + Meta: resize__symint + CPU: resize_ CUDA: resize_cuda_ MPS: resize_mps_ QuantizedCPU: quantized_resize_cpu_ SparseCsrCPU, SparseCsrCUDA: resize_sparse_csr_ autogen: resize, resize.out @@ -2188,128 +2337,147 @@ dispatch: CompositeExplicitAutograd: empty_like QuantizedCPU, QuantizedCUDA: empty_like_quantized SparseCPU, SparseCUDA, SparseMeta: empty_like_sparse_coo SparseCsrCPU, SparseCsrCUDA: empty_like_sparse_csr + NestedTensorCPU, NestedTensorCUDA: empty_like_nested autogen: empty_like.out - func: empty_strided(SymInt[] size, SymInt[] stride, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor dispatch: CPU: empty_strided_cpu CUDA: empty_strided_cuda MPS: empty_strided_mps Meta: empty_strided_meta_symint QuantizedCPU, QuantizedCUDA: empty_strided_unknown_quantized autogen: empty_strided.out + tags: core - func: erf(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: erf.out variants: function, method dispatch: SparseCPU, SparseCUDA: erf_sparse SparseCsrCPU, SparseCsrCUDA: erf_sparse_csr + tags: [core, pointwise] - func: erf_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: erf.out variants: function, method dispatch: SparseCPU, SparseCUDA: erf_sparse_ SparseCsrCPU, SparseCsrCUDA: erf_sparse_csr_ + tags: pointwise - func: erf.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: erf_out MPS: erf_out_mps SparseCPU, SparseCUDA: erf_sparse_out SparseCsrCPU, SparseCsrCUDA: erf_sparse_csr_out + tags: pointwise - func: erfc(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: erfc.out variants: function, method + tags: pointwise - func: erfc_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: erfc.out variants: function, method + tags: pointwise - func: erfc.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: erfc_out + tags: pointwise - func: exp(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: exp.out variants: function, method + tags: [core, pointwise] - func: exp_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: exp.out variants: function, method + tags: pointwise - func: exp.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: exp_out MPS: exp_out_mps + tags: pointwise - func: exp2(Tensor self) -> Tensor structured_delegate: exp2.out variants: function, method + tags: pointwise - func: exp2_(Tensor(a!) self) -> Tensor(a!) structured_delegate: exp2.out variants: function, method + tags: pointwise - func: exp2.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: exp2_out MPS: exp2_out_mps + tags: pointwise - func: expm1(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: expm1.out variants: function, method dispatch: SparseCPU, SparseCUDA: expm1_sparse SparseCsrCPU, SparseCsrCUDA: expm1_sparse_csr + tags: pointwise - func: expm1_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: expm1.out variants: function, method dispatch: SparseCPU, SparseCUDA: expm1_sparse_ SparseCsrCPU, SparseCsrCUDA: expm1_sparse_csr_ + tags: pointwise - func: expm1.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: expm1_out + MPS: expm1_out_mps SparseCPU, SparseCUDA: expm1_sparse_out SparseCsrCPU, SparseCsrCUDA: expm1_sparse_csr_out + tags: pointwise - func: expand(Tensor(a) self, SymInt[] size, *, bool implicit=False) -> Tensor(a) variants: method # This is method-only to match the previous tensor API. In the future we could make this a function too. device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: expand + tags: core - func: expand_as(Tensor(a) self, Tensor other) -> Tensor(a) variants: method # This is method-only to match the previous tensor API. In the future we could make this a function too. device_check: NoCheck device_guard: False @@ -2355,10 +2523,11 @@ - func: fill.Scalar(Tensor self, Scalar value) -> Tensor variants: function dispatch: CompositeExplicitAutograd: fill + tags: core - func: fill.Tensor(Tensor self, Tensor value) -> Tensor variants: function dispatch: CompositeExplicitAutograd: fill @@ -2370,66 +2539,74 @@ CPU, CUDA: fill_ MPS: fill_scalar_mps QuantizedCPU, QuantizedCUDA: fill_quantized_ Meta: fill_meta_ SparseCsrCPU, SparseCsrCUDA: fill_sparse_csr_ + NestedTensorCPU, NestedTensorCUDA: fill_nested_ autogen: fill.Scalar_out - func: fill_.Tensor(Tensor(a!) self, Tensor value) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method dispatch: CPU, CUDA: fill_ MPS: fill_tensor_mps_ QuantizedCPU, QuantizedCUDA: fill_quantized_ Meta: fill_meta_ + NestedTensorCPU, NestedTensorCUDA: fill_nested_ autogen: fill.Tensor_out - func: floor(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: floor.out variants: function, method dispatch: SparseCPU, SparseCUDA: floor_sparse SparseCsrCPU, SparseCsrCUDA: floor_sparse_csr + tags: [core, pointwise] - func: floor_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: floor.out variants: function, method dispatch: SparseCPU, SparseCUDA: floor_sparse_ SparseCsrCPU, SparseCsrCUDA: floor_sparse_csr_ + tags: pointwise - func: floor.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: floor_out MPS: floor_out_mps SparseCPU, SparseCUDA: floor_sparse_out SparseCsrCPU, SparseCsrCUDA: floor_sparse_csr_out + tags: pointwise - func: floor_divide(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CPU, CUDA: floor_divide + MPS: floor_divide_mps SparseCPU, SparseCUDA: floor_divide_sparse - func: floor_divide_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CPU, CUDA: floor_divide_ + MPS: floor_divide_mps_ SparseCPU, SparseCUDA: floor_divide_sparse_ - func: floor_divide.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: floor_divide_out + MPS: floor_divide_out_mps SparseCPU, SparseCUDA: floor_divide_out_sparse_zerodim - func: floor_divide.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method @@ -2440,22 +2617,34 @@ - func: frac(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: frac.out variants: function, method + dispatch: + SparseCPU, SparseCUDA: frac_sparse + SparseCsrCPU, SparseCsrCUDA: frac_sparse_csr + tags: pointwise - func: frac_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: frac.out variants: function, method + dispatch: + SparseCPU, SparseCUDA: frac_sparse_ + SparseCsrCPU, SparseCsrCUDA: frac_sparse_csr_ + tags: pointwise - func: frac.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: frac_out + MPS: frac_out_mps + SparseCPU, SparseCUDA: frac_sparse_out + SparseCsrCPU, SparseCsrCUDA: frac_sparse_csr_out + tags: pointwise - func: full.names(int[] size, Scalar fill_value, *, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor device_check: NoCheck device_guard: False dispatch: @@ -2463,10 +2652,11 @@ autogen: full.names_out - func: full(SymInt[] size, Scalar fill_value, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor dispatch: CompositeExplicitAutograd: full + tags: core - func: full.out(SymInt[] size, Scalar fill_value, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: full_out @@ -2485,28 +2675,32 @@ - func: gcd.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: gcd_out + tags: pointwise - func: gcd(Tensor self, Tensor other) -> Tensor structured_delegate: gcd.out variants: function, method + tags: pointwise - func: gcd_(Tensor(a!) self, Tensor other) -> Tensor(a!) structured_delegate: gcd.out variants: function, method - func: lcm.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: lcm_out + tags: pointwise - func: lcm(Tensor self, Tensor other) -> Tensor structured_delegate: lcm.out variants: function, method + tags: pointwise - func: lcm_(Tensor(a!) self, Tensor other) -> Tensor(a!) structured_delegate: lcm.out variants: function, method @@ -2531,11 +2725,13 @@ - func: grid_sampler_2d(Tensor input, Tensor grid, int interpolation_mode, int padding_mode, bool align_corners) -> Tensor dispatch: CPU, QuantizedCPU: grid_sampler_2d_cpu CUDA: grid_sampler_2d_cuda + MPS: grid_sampler_2d_mps autogen: grid_sampler_2d.out + tags: core # `grid_sampler_2d_backward` takes in `output_mask` to optimize performance for # the case where `input` doesn't require gradient. Gradient for `grid` is always # computed (only `output_mask[0]` is checked by the implementations). - func: grid_sampler_2d_backward(Tensor grad_output, Tensor input, Tensor grid, int interpolation_mode, int padding_mode, bool align_corners, bool[2] output_mask) -> (Tensor, Tensor) @@ -2619,15 +2815,17 @@ - func: native_group_norm(Tensor input, Tensor? weight, Tensor? bias, SymInt N, SymInt C, SymInt HxW, int group, float eps) -> (Tensor, Tensor, Tensor) dispatch: CPU, CUDA: native_group_norm CompositeExplicitAutograd: math_group_norm autogen: native_group_norm.out + tags: core - func: native_group_norm_backward(Tensor grad_out, Tensor input, Tensor mean, Tensor rstd, Tensor? weight, SymInt N, SymInt C, SymInt HxW, int group, bool[3] output_mask) -> (Tensor, Tensor, Tensor) dispatch: CPU, CUDA: native_group_norm_backward autogen: native_group_norm_backward.out + tags: core # Real to complex forward FFT - func: _fft_r2c(Tensor self, int[] dim, int normalization, bool onesided) -> Tensor variants: function dispatch: @@ -2652,17 +2850,17 @@ dispatch: CPU: _fft_c2r_mkl_out CUDA: _fft_c2r_cufft_out # Standard complex to complex FFT (forward or backward) -- func: _fft_c2c(Tensor self, int[] dim, int normalization, bool forward) -> Tensor +- func: _fft_c2c(Tensor self, SymInt[] dim, int normalization, bool forward) -> Tensor variants: function dispatch: CPU: _fft_c2c_mkl CUDA: _fft_c2c_cufft -- func: _fft_c2c.out(Tensor self, int[] dim, int normalization, bool forward, *, Tensor(a!) out) -> Tensor(a!) +- func: _fft_c2c.out(Tensor self, SymInt[] dim, int normalization, bool forward, *, Tensor(a!) out) -> Tensor(a!) variants: function dispatch: CPU: _fft_c2c_mkl_out CUDA: _fft_c2c_cufft_out @@ -2794,10 +2992,11 @@ dispatch: CPU, CUDA, MPS: isnan SparseCPU, SparseCUDA: isnan_sparse SparseCsrCPU, SparseCsrCUDA: isnan_sparse_csr autogen: isnan.out + tags: [core, pointwise] - func: is_distributed(Tensor self) -> bool variants: function, method device_check: NoCheck device_guard: False @@ -2877,43 +3076,52 @@ - func: kthvalue.dimname(Tensor self, int k, Dimname dim, bool keepdim=False) -> (Tensor values, Tensor indices) variants: function, method - func: kthvalue.dimname_out(Tensor self, int k, Dimname dim, bool keepdim=False, *, Tensor(a!) values, Tensor(b!) indices) -> (Tensor(a!) values, Tensor(b!) indices) -- func: layer_norm(Tensor input, int[] normalized_shape, Tensor? weight=None, Tensor? bias=None, float eps=1e-05, bool cudnn_enable=True) -> Tensor +- func: layer_norm(Tensor input, SymInt[] normalized_shape, Tensor? weight=None, Tensor? bias=None, float eps=1e-05, bool cudnn_enable=True) -> Tensor + dispatch: + CompositeImplicitAutograd: layer_norm_symint - func: native_layer_norm(Tensor input, SymInt[] normalized_shape, Tensor? weight, Tensor? bias, float eps) -> (Tensor, Tensor, Tensor) dispatch: CPU: layer_norm_cpu CUDA: layer_norm_cuda MPS: layer_norm_mps CompositeExplicitAutograd: math_native_layer_norm + NestedTensorCPU, NestedTensorCUDA: nested_layer_norm autogen: native_layer_norm.out + tags: core - func: native_layer_norm_backward(Tensor grad_out, Tensor input, SymInt[] normalized_shape, Tensor mean, Tensor rstd, Tensor? weight, Tensor? bias, bool[3] output_mask) -> (Tensor, Tensor, Tensor) dispatch: CPU: layer_norm_backward_cpu CUDA: layer_norm_backward_cuda MPS: layer_norm_backward_mps autogen: native_layer_norm_backward.out + tags: core - func: nan_to_num(Tensor self, float? nan=None, float? posinf=None, float? neginf=None) -> Tensor variants: function, method dispatch: CompositeExplicitAutograd: nan_to_num SparseCPU, SparseCUDA: nan_to_num_sparse + tags: pointwise - func: nan_to_num_(Tensor(a!) self, float? nan=None, float? posinf=None, float? neginf=None) -> Tensor(a!) variants: function, method dispatch: CompositeExplicitAutograd: nan_to_num_ SparseCPU, SparseCUDA: nan_to_num_sparse_ + tags: pointwise - func: nan_to_num.out(Tensor self, float? nan=None, float? posinf=None, float? neginf=None, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: nan_to_num_out + MPS: nan_to_num_out_mps SparseCPU, SparseCUDA: nan_to_num_sparse_out + tags: pointwise - func: linear(Tensor input, Tensor weight, Tensor? bias=None) -> Tensor python_module: nn dispatch: CompositeImplicitAutograd: linear @@ -2971,12 +3179,14 @@ - func: ldexp.Tensor(Tensor self, Tensor other) -> Tensor variants: function, method - func: ldexp_(Tensor(a!) self, Tensor other) -> Tensor(a!) variants: function, method + tags: pointwise - func: ldexp.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) + tags: pointwise - func: linspace(Scalar start, Scalar end, int steps, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor dispatch: CompositeExplicitAutograd: linspace @@ -2988,130 +3198,150 @@ - func: log(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: log.out variants: function, method + tags: [core, pointwise] - func: log_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: log.out variants: function, method + tags: pointwise - func: log.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: log_out MPS: log_out_mps + tags: pointwise - func: log10(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: log10.out variants: function, method + tags: pointwise - func: log10_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: log10.out variants: function, method + tags: pointwise - func: log10.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: log10_out MPS: log10_out_mps + tags: pointwise - func: log1p(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: log1p.out variants: function, method dispatch: SparseCPU, SparseCUDA: log1p_sparse SparseCsrCPU, SparseCsrCUDA: log1p_sparse_csr + tags: pointwise - func: log1p_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: log1p.out variants: function, method dispatch: SparseCPU, SparseCUDA: log1p_sparse_ SparseCsrCPU, SparseCsrCUDA: log1p_sparse_csr_ + tags: pointwise - func: log1p.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: log1p_out MPS: log1p_out_mps SparseCPU, SparseCUDA: log1p_sparse_out SparseCsrCPU, SparseCsrCUDA: log1p_sparse_csr_out + tags: pointwise - func: log2(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: log2.out variants: function, method + tags: pointwise - func: log2_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: log2.out variants: function, method + tags: pointwise - func: log2.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: log2_out MPS: log2_out_mps + tags: pointwise - func: logaddexp.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: logaddexp_out MPS: logaddexp_out_mps + tags: pointwise - func: logaddexp(Tensor self, Tensor other) -> Tensor variants: method, function structured_delegate: logaddexp.out + tags: pointwise - func: logaddexp2.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: logaddexp2_out MPS: logaddexp2_out_mps + tags: pointwise - func: logaddexp2(Tensor self, Tensor other) -> Tensor variants: method, function structured_delegate: logaddexp2.out + tags: pointwise - func: xlogy.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: xlogy.OutTensor variants: function, method + tags: pointwise - func: xlogy.Scalar_Self(Scalar self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: xlogy + tags: pointwise - func: xlogy.Scalar_Other(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: xlogy + tags: pointwise # xlogy: inplace variant - func: xlogy_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method structured_delegate: xlogy.OutTensor + tags: pointwise - func: xlogy_.Scalar_Other(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method dispatch: @@ -3123,22 +3353,25 @@ structured: True structured_inherits: TensorIteratorBase variants: function dispatch: CPU, CUDA: xlogy_out + tags: pointwise - func: xlogy.OutScalar_Self(Scalar self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: xlogy_out + tags: pointwise - func: xlogy.OutScalar_Other(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: xlogy_out + tags: pointwise - func: logspace(Scalar start, Scalar end, int steps, float base=10.0, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor dispatch: CompositeExplicitAutograd: logspace @@ -3159,10 +3392,11 @@ - func: log_softmax.Dimname(Tensor self, Dimname dim, *, ScalarType? dtype=None) -> Tensor variants: function, method - func: _log_softmax(Tensor self, int dim, bool half_to_float) -> Tensor structured_delegate: _log_softmax.out + tags: core - func: _log_softmax.out(Tensor self, int dim, bool half_to_float, *, Tensor(a!) out) -> Tensor(a!) structured: True dispatch: CPU: log_softmax_cpu_out @@ -3289,10 +3523,11 @@ device_check: NoCheck # TensorIterator structured_delegate: max.dim_max variants: function, method dispatch: QuantizedCPU, QuantizedCUDA: qmax + tags: core - func: max.dim_max(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) max, Tensor(b!) max_values) -> (Tensor(a!) values, Tensor(b!) indices) device_check: NoCheck # TensorIterator structured: True precomputed: @@ -3306,18 +3541,21 @@ variants: function, method - func: max.names_dim_max(Tensor self, Dimname dim, bool keepdim=False, *, Tensor(a!) max, Tensor(b!) max_values) -> (Tensor(a!) values, Tensor(b!) indices) device_check: NoCheck # TensorIterator -- func: value_selecting_reduction_backward(Tensor grad, int dim, Tensor indices, int[] sizes, bool keepdim) -> Tensor +- func: value_selecting_reduction_backward(Tensor grad, int dim, Tensor indices, SymInt[] sizes, bool keepdim) -> Tensor variants: function device_check: NoCheck device_guard: False + dispatch: + CompositeImplicitAutograd: value_selecting_reduction_backward_symint - func: amax(Tensor self, int[1] dim=[], bool keepdim=False) -> Tensor variants: function, method structured_delegate: amax.out + tags: core - func: amax.out(Tensor self, int[1] dim=[], bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!) structured: True dispatch: CPU, CUDA: amax_out @@ -3327,23 +3565,18 @@ - func: max_pool1d_with_indices(Tensor self, int[1] kernel_size, int[1] stride=[], int[1] padding=0, int[1] dilation=1, bool ceil_mode=False) -> (Tensor, Tensor) - func: max_pool1d(Tensor self, int[1] kernel_size, int[1] stride=[], int[1] padding=0, int[1] dilation=1, bool ceil_mode=False) -> Tensor - func: max_pool2d(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False) -> Tensor - -# TODO: Add this function to MPS dispatch key so that we avoid declaring it in -# native_functions.yaml -# https://github.com/pytorch/pytorch/issues/77394 -- func: _mps_max_pool2d(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False) -> Tensor dispatch: - MPS: _mps_max_pool2d - autogen: _mps_max_pool2d.out + CompositeImplicitAutograd: max_pool2d + MPS: mps_max_pool2d -- func: mps_max_pool2d_backward(Tensor grad_output, Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False) -> Tensor +- func: max_pool2d_backward(Tensor grad_output, Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False) -> Tensor dispatch: MPS: mps_max_pool2d_backward - autogen: mps_max_pool2d_backward.out + autogen: max_pool2d_backward.out - func: mkldnn_max_pool2d(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False) -> Tensor dispatch: MkldnnCPU: mkldnn_max_pool2d autogen: mkldnn_max_pool2d.out @@ -3395,10 +3628,11 @@ structured_delegate: mean.out device_check: NoCheck # TensorIterator variants: function, method dispatch: QuantizedCPU: mean_quantized_cpu + tags: core - func: mean.out(Tensor self, int[1]? dim, bool keepdim=False, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!) structured: True device_check: NoCheck # TensorIterator dispatch: @@ -3423,10 +3657,11 @@ - func: median(Tensor self) -> Tensor variants: function, method dispatch: CPU: median_cpu CUDA: median_cuda + MPS: median_mps autogen: median.out - func: median.dim(Tensor self, int dim, bool keepdim=False) -> (Tensor values, Tensor indices) variants: function, method dispatch: @@ -3434,10 +3669,11 @@ - func: median.dim_values(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) values, Tensor(b!) indices) -> (Tensor(a!) values, Tensor(b!) indices) dispatch: CPU: median_out_cpu CUDA: median_out_cuda + MPS: median_out_mps - func: median.names_dim(Tensor self, Dimname dim, bool keepdim=False) -> (Tensor values, Tensor indices) variants: function, method - func: median.names_dim_values(Tensor self, Dimname dim, bool keepdim=False, *, Tensor(a!) values, Tensor(b!) indices) -> (Tensor(a!) values, Tensor(b!) indices) @@ -3468,10 +3704,11 @@ device_check: NoCheck # TensorIterator structured_delegate: min.dim_min variants: function, method dispatch: QuantizedCPU, QuantizedCUDA: qmin + tags: core - func: min.dim_min(Tensor self, int dim, bool keepdim=False, *, Tensor(a!) min, Tensor(b!) min_indices) -> (Tensor(a!) values, Tensor(b!) indices) device_check: NoCheck # TensorIterator structured: True precomputed: @@ -3488,10 +3725,11 @@ device_check: NoCheck # TensorIterator - func: amin(Tensor self, int[1] dim=[], bool keepdim=False) -> Tensor variants: function, method structured_delegate: amin.out + tags: core - func: amin.out(Tensor self, int[1] dim=[], bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!) structured: True dispatch: CPU, CUDA: amin_out @@ -3508,36 +3746,46 @@ - func: mps_convolution_backward(Tensor self, Tensor grad_output, Tensor weight, int[] padding, int[] stride, int[] dilation, int groups, bool[3] output_mask) -> (Tensor, Tensor, Tensor) dispatch: MPS: mps_convolution_backward autogen: mps_convolution_backward.out -- func: mkldnn_convolution(Tensor self, Tensor weight, Tensor? bias, int[] padding, int[] stride, int[] dilation, int groups) -> Tensor +- func: mkldnn_convolution(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, int[] stride, int[] dilation, int groups) -> Tensor dispatch: CompositeExplicitAutograd: mkldnn_convolution autogen: mkldnn_convolution.out +- func: mkldnn_rnn_layer(Tensor input, Tensor weight0, Tensor weight1, Tensor weight2, Tensor weight3, Tensor hx_, Tensor cx_, bool reverse, int[] batch_sizes, int mode, int hidden_size, int num_layers, bool has_biases, bool bidirectional, bool batch_first, bool train) -> (Tensor, Tensor, Tensor, Tensor) + dispatch: + CPU: mkldnn_rnn_layer + autogen: mkldnn_rnn_layer.out + +- func: mkldnn_rnn_layer_backward(Tensor input, Tensor weight1, Tensor weight2, Tensor weight3, Tensor weight4, Tensor hx_, Tensor cx_tmp, Tensor output, Tensor hy_, Tensor cy_, Tensor? grad_output, Tensor? grad_hy, Tensor? grad_cy, bool reverse, int mode, int hidden_size, int num_layers, bool has_biases, bool train, bool bidirectional, int[] batch_sizes, bool batch_first, Tensor workspace) -> (Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, Tensor) + dispatch: + CPU: mkldnn_rnn_layer_backward + autogen: mkldnn_rnn_layer_backward.out + - func: miopen_batch_norm(Tensor input, Tensor weight, Tensor? bias, Tensor? running_mean, Tensor? running_var, bool training, float exponential_average_factor, float epsilon) -> (Tensor, Tensor, Tensor) dispatch: CUDA: miopen_batch_norm autogen: miopen_batch_norm.out - func: miopen_batch_norm_backward(Tensor input, Tensor grad_output, Tensor weight, Tensor? running_mean, Tensor? running_var, Tensor? save_mean, Tensor? save_var, float epsilon) -> (Tensor, Tensor, Tensor) dispatch: CUDA: miopen_batch_norm_backward autogen: miopen_batch_norm_backward.out -- func: miopen_convolution(Tensor self, Tensor weight, Tensor? bias, int[] padding, int[] stride, int[] dilation, int groups, bool benchmark, bool deterministic) -> Tensor +- func: miopen_convolution(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, int[] stride, int[] dilation, int groups, bool benchmark, bool deterministic) -> Tensor dispatch: CUDA: miopen_convolution autogen: miopen_convolution.out -- func: miopen_convolution_transpose(Tensor self, Tensor weight, Tensor? bias, int[] padding, int[] output_padding, int[] stride, int[] dilation, int groups, bool benchmark, bool deterministic) -> Tensor +- func: miopen_convolution_transpose(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, SymInt[] output_padding, int[] stride, int[] dilation, int groups, bool benchmark, bool deterministic) -> Tensor dispatch: CUDA: miopen_convolution_transpose autogen: miopen_convolution_transpose.out -- func: miopen_depthwise_convolution(Tensor self, Tensor weight, Tensor? bias, int[] padding, int[] stride, int[] dilation, int groups, bool benchmark, bool deterministic) -> Tensor +- func: miopen_depthwise_convolution(Tensor self, Tensor weight, Tensor? bias, SymInt[] padding, int[] stride, int[] dilation, int groups, bool benchmark, bool deterministic) -> Tensor dispatch: CUDA: miopen_depthwise_convolution autogen: miopen_depthwise_convolution.out - func: miopen_convolution_relu(Tensor self, Tensor weight, Tensor? bias, int[] stride, int[] padding, int[] dilation, int groups) -> Tensor @@ -3562,10 +3810,11 @@ structured_delegate: mm.out variants: function, method dispatch: SparseCPU, SparseCUDA: _sparse_mm SparseCsrCPU, SparseCsrCUDA: _sparse_csr_mm + tags: core - func: mm.out(Tensor self, Tensor mat2, *, Tensor(a!) out) -> Tensor(a!) structured: True dispatch: CPU: mm_out_cpu @@ -3575,22 +3824,19 @@ SparseCsrCPU, SparseCsrCUDA: _sparse_csr_mm_out - func: _sparse_mm(Tensor sparse, Tensor dense) -> Tensor python_module: sparse +- func: _sparse_mm.reduce(Tensor sparse, Tensor dense, str reduce) -> Tensor + python_module: sparse + - func: _sparse_sparse_matmul(Tensor self, Tensor other) -> Tensor dispatch: SparseCPU: sparse_sparse_matmul_cpu SparseCUDA: sparse_sparse_matmul_cuda autogen: _sparse_sparse_matmul.out -- func: _sparse_mask_helper(Tensor t, Tensor mask_indices) -> Tensor - dispatch: - SparseCPU: sparse_mask_helper_cpu - SparseCUDA: sparse_mask_helper_cuda - autogen: _sparse_mask_helper.out - - func: mode(Tensor self, int dim=-1, bool keepdim=False) -> (Tensor values, Tensor indices) variants: function, method dispatch: CPU, CUDA: mode @@ -3611,20 +3857,22 @@ SparseCPU, SparseCUDA: mul_sparse SparseCsrCPU, SparseCsrCUDA: mul_sparse_csr MkldnnCPU: mkldnn_mul ZeroTensor: mul_zerotensor NestedTensorCPU, NestedTensorCUDA: NestedTensor_mul_Tensor + tags: [core, pointwise] - func: mul_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: mul.out variants: method dispatch: SparseCPU, SparseCUDA: mul_sparse_ SparseCsrCPU, SparseCsrCUDA: mul_sparse_csr_ MkldnnCPU: mkldnn_mul_ NestedTensorCPU, NestedTensorCUDA: NestedTensor_mul__Tensor + tags: pointwise - func: mul.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase @@ -3633,30 +3881,33 @@ MPS: mul_out_mps SparseCPU: mul_out_sparse_cpu SparseCUDA: mul_out_sparse_cuda SparseCsrCPU, SparseCsrCUDA: mul_out_sparse_csr MkldnnCPU: mkldnn_mul_out - + tags: pointwise # For C++ only, until we have conversion from C++ numbers to Tensor + - func: mul.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: mul SparseCsrCPU, SparseCsrCUDA: mul_scalar_sparse_csr NestedTensorCPU, NestedTensorCUDA: NestedTensor_mul_Scalar + tags: [core, pointwise] - func: mul_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: mul_ SparseCsrCPU, SparseCsrCUDA: mul__scalar_sparse_csr NestedTensorCPU, NestedTensorCUDA: NestedTensor_mul__Scalar autogen: mul.Scalar_out - + tags: pointwise # multiply, alias for mul + - func: multiply.Tensor(Tensor self, Tensor other) -> Tensor variants: function, method - func: multiply_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) variants: method @@ -3680,57 +3931,95 @@ CompositeExplicitAutograd: mv_out - func: mvlgamma.out(Tensor self, int p, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: mvlgamma_out + tags: pointwise - func: mvlgamma(Tensor self, int p) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: mvlgamma + tags: pointwise - func: mvlgamma_(Tensor(a!) self, int p) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: mvlgamma_ + tags: pointwise - func: narrow_copy(Tensor self, int dim, SymInt start, SymInt length) -> Tensor variants: function, method dispatch: CPU: narrow_copy_dense_cpu SparseCPU, SparseCUDA: narrow_copy_sparse - CompositeExplicitAutogradNonFunctional: narrow_copy_dense + CompositeExplicitAutogradNonFunctional: narrow_copy_dense_symint tags: view_copy - func: narrow_copy.out(Tensor self, int dim, SymInt start, SymInt length, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU: narrow_copy_dense_cpu_out -- func: narrow(Tensor(a) self, int dim, int start, int length) -> Tensor(a) +- func: narrow(Tensor(a) self, int dim, SymInt start, SymInt length) -> Tensor(a) variants: function, method device_check: NoCheck device_guard: False + dispatch: + CompositeImplicitAutograd: narrow_symint -- func: narrow.Tensor(Tensor(a) self, int dim, Tensor start, int length) -> Tensor(a) +- func: narrow.Tensor(Tensor(a) self, int dim, Tensor start, SymInt length) -> Tensor(a) variants: function, method device_check: NoCheck device_guard: False + dispatch: + CompositeImplicitAutograd: narrow_tensor_symint - func: native_batch_norm(Tensor input, Tensor? weight, Tensor? bias, Tensor? running_mean, Tensor? running_var, bool training, float momentum, float eps) -> (Tensor, Tensor, Tensor) dispatch: CPU: batch_norm_cpu CUDA: batch_norm_cuda MPS: batch_norm_mps MkldnnCPU: mkldnn_batch_norm + tags: core - func: native_batch_norm.out(Tensor input, Tensor? weight, Tensor? bias, Tensor? running_mean, Tensor? running_var, bool training, float momentum, float eps, *, Tensor(a!) out, Tensor(b!) save_mean, Tensor(c!) save_invstd) -> (Tensor(a!), Tensor(b!), Tensor(c!)) dispatch: CUDA: batch_norm_cuda_out MPS: batch_norm_mps_out + CPU: batch_norm_cpu_out +# TODO: In 2 weeks, we should make native_batch_norm composite implicit so that this correct schema percolates correctly through our dispatching +- func: _native_batch_norm_legit(Tensor input, Tensor? weight, Tensor? bias, Tensor(a!) running_mean, Tensor(b!) running_var, bool training, float momentum, float eps) -> (Tensor, Tensor, Tensor) + dispatch: + CPU: _batch_norm_legit_cpu + CUDA: _batch_norm_legit_cuda + MPS: _batch_norm_legit_mps + MkldnnCPU: _mkldnn_batch_norm_legit + autogen: _native_batch_norm_legit_functional + +- func: _native_batch_norm_legit.out(Tensor input, Tensor? weight, Tensor? bias, Tensor(a!) running_mean, Tensor(b!) running_var, bool training, float momentum, float eps, *, Tensor(d!) out, Tensor(e!) save_mean, Tensor(f!) save_invstd) -> (Tensor(d!), Tensor(e!), Tensor(f!)) + dispatch: + CPU: _batch_norm_legit_cpu_out + CUDA: _batch_norm_legit_cuda_out + MPS: _batch_norm_legit_mps_out + +- func: _native_batch_norm_legit.no_stats(Tensor input, Tensor? weight, Tensor? bias, bool training, float momentum, float eps) -> (Tensor, Tensor, Tensor) + dispatch: + CPU: _batch_norm_legit_no_stats_cpu + CUDA: _batch_norm_legit_no_stats_cuda + MPS: _batch_norm_legit_no_stats_mps + MkldnnCPU: _mkldnn_batch_norm_legit_no_stats + tags: core + +- func: _native_batch_norm_legit.no_stats_out(Tensor input, Tensor? weight, Tensor? bias, bool training, float momentum, float eps, *, Tensor(a!) out, Tensor(b!) save_mean, Tensor(c!) save_invstd) -> (Tensor(a!), Tensor(b!), Tensor(c!)) + dispatch: + CPU: _batch_norm_legit_no_stats_cpu_out + CUDA: _batch_norm_legit_no_stats_cuda_out + MPS: _batch_norm_legit_no_stats_mps_out + - func: batch_norm_stats(Tensor input, float eps) -> (Tensor, Tensor) dispatch: CUDA: batch_norm_stats_cuda autogen: batch_norm_stats.out @@ -3779,11 +4068,11 @@ - func: is_vulkan_available() -> bool - func: _nnpack_available() -> bool -- func: _nnpack_spatial_convolution(Tensor input, Tensor weight, Tensor? bias, int[2] padding, int[2] stride=1) -> Tensor +- func: _nnpack_spatial_convolution(Tensor input, Tensor weight, Tensor? bias, SymInt[2] padding, int[2] stride=1) -> Tensor variants: function dispatch: CompositeExplicitAutograd: _nnpack_spatial_convolution autogen: _nnpack_spatial_convolution.out @@ -3805,10 +4094,11 @@ - func: ones_like(Tensor self, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor dispatch: # NB: Although this composite mutates on the inside, it is # non-differentiable so NonFunctional doesn't apply CompositeExplicitAutograd: ones_like + NestedTensorCPU, NestedTensorCUDA: ones_like autogen: ones_like.out - func: pairwise_distance(Tensor x1, Tensor x2, float p=2, float eps=1e-06, bool keepdim=False) -> Tensor - func: cdist(Tensor x1, Tensor x2, float p=2, int? compute_mode=None) -> Tensor @@ -3819,10 +4109,11 @@ autogen: _euclidean_dist.out - func: _cdist_forward(Tensor x1, Tensor x2, float p, int? compute_mode) -> Tensor dispatch: CPU, CUDA: _cdist_forward + MPS: _cdist_forward_mps autogen: _cdist_forward.out - func: _cdist_backward(Tensor grad, Tensor x1, Tensor x2, float p, Tensor cdist) -> Tensor dispatch: CPU, CUDA: _cdist_backward @@ -3847,10 +4138,11 @@ variants: function, method dispatch: CompositeExplicitAutograd: permute MPS: permute_mps SparseCPU, SparseCUDA: permute_sparse_coo + tags: core - func: movedim.intlist(Tensor(a) self, int[] source, int[] destination) -> Tensor(a) variants: function, method - func: movedim.int(Tensor(a) self, int source, int destination) -> Tensor(a) @@ -3938,120 +4230,133 @@ - func: rad2deg(Tensor self) -> Tensor variants: function, method dispatch: CompositeExplicitAutograd: rad2deg + SparseCPU, SparseCUDA: rad2deg_sparse SparseCsrCPU, SparseCsrCUDA: rad2deg_sparse_csr - func: rad2deg_(Tensor(a!) self) -> Tensor(a!) variants: function, method dispatch: CompositeExplicitAutograd: rad2deg_ + SparseCPU, SparseCUDA: rad2deg_sparse_ SparseCsrCPU, SparseCsrCUDA: rad2deg_sparse_csr_ - func: rad2deg.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: rad2deg_out + SparseCPU, SparseCUDA: rad2deg_sparse_out SparseCsrCPU, SparseCsrCUDA: rad2deg_sparse_csr_out - func: deg2rad(Tensor self) -> Tensor variants: function, method dispatch: CompositeExplicitAutograd: deg2rad + SparseCPU, SparseCUDA: deg2rad_sparse + SparseCsrCPU, SparseCsrCUDA: deg2rad_sparse_csr + tags: pointwise - func: deg2rad_(Tensor(a!) self) -> Tensor(a!) variants: function, method dispatch: CompositeExplicitAutograd: deg2rad_ + SparseCPU, SparseCUDA: deg2rad_sparse_ + SparseCsrCPU, SparseCsrCUDA: deg2rad_sparse_csr_ + tags: pointwise - func: deg2rad.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: deg2rad_out + SparseCPU, SparseCUDA: deg2rad_sparse_out + SparseCsrCPU, SparseCsrCUDA: deg2rad_sparse_csr_out + tags: pointwise - func: scalar_tensor(Scalar s, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor dispatch: CompositeExplicitAutograd: scalar_tensor autogen: scalar_tensor.out + tags: core -- func: rand.names(int[] size, *, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: rand.names(SymInt[] size, *, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: rand autogen: rand.names_out tags: nondeterministic_seeded -- func: rand.generator_with_names(int[] size, *, Generator? generator, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: rand.generator_with_names(SymInt[] size, *, Generator? generator, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor device_check: NoCheck device_guard: False tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: rand autogen: rand.generator_with_names_out -- func: rand(int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: rand(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: rand -- func: rand.generator(int[] size, *, Generator? generator, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: rand.generator(SymInt[] size, *, Generator? generator, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: rand -- func: rand.out(int[] size, *, Tensor(a!) out) -> Tensor(a!) +- func: rand.out(SymInt[] size, *, Tensor(a!) out) -> Tensor(a!) tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: rand_out -- func: rand.generator_out(int[] size, *, Generator? generator, Tensor(a!) out) -> Tensor(a!) +- func: rand.generator_out(SymInt[] size, *, Generator? generator, Tensor(a!) out) -> Tensor(a!) tags: nondeterministic_seeded - func: rand_like(Tensor self, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor tags: nondeterministic_seeded dispatch: # NB: Although this composite mutates on the inside, it is # non-differentiable so NonFunctional doesn't apply CompositeExplicitAutograd: rand_like autogen: rand_like.out -- func: randint(int high, int[] size, *, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: randint(int high, SymInt[] size, *, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: randint -- func: randint.generator(int high, int[] size, *, Generator? generator, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: randint.generator(int high, SymInt[] size, *, Generator? generator, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: randint -- func: randint.low(int low, int high, int[] size, *, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: randint.low(int low, int high, SymInt[] size, *, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: randint -- func: randint.low_generator(int low, int high, int[] size, *, Generator? generator, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: randint.low_generator(int low, int high, SymInt[] size, *, Generator? generator, ScalarType? dtype=long, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: randint -- func: randint.out(int high, int[] size, *, Tensor(a!) out) -> Tensor(a!) +- func: randint.out(int high, SymInt[] size, *, Tensor(a!) out) -> Tensor(a!) tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: randint_out -- func: randint.generator_out(int high, int[] size, *, Generator? generator, Tensor(a!) out) -> Tensor(a!) +- func: randint.generator_out(int high, SymInt[] size, *, Generator? generator, Tensor(a!) out) -> Tensor(a!) tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: randint_out -- func: randint.low_out(int low, int high, int[] size, *, Tensor(a!) out) -> Tensor(a!) +- func: randint.low_out(int low, int high, SymInt[] size, *, Tensor(a!) out) -> Tensor(a!) tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: randint_out -- func: randint.low_generator_out(int low, int high, int[] size, *, Generator? generator, Tensor(a!) out) -> Tensor(a!) +- func: randint.low_generator_out(int low, int high, SymInt[] size, *, Generator? generator, Tensor(a!) out) -> Tensor(a!) tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: randint_out - func: randint_like(Tensor self, int high, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor @@ -4068,40 +4373,40 @@ # NB: Although this composite mutates on the inside, it is # non-differentiable so NonFunctional doesn't apply CompositeExplicitAutograd: randint_like autogen: randint_like.low_dtype_out -- func: randn(int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: randn(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: randn -- func: randn.generator(int[] size, *, Generator? generator, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: randn.generator(SymInt[] size, *, Generator? generator, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor tags: nondeterministic_seeded dispatch: CompositeExplicitAutograd: randn -- func: randn.names(int[] size, *, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: randn.names(SymInt[] size, *, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor tags: nondeterministic_seeded device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: randn autogen: randn.names_out -- func: randn.generator_with_names(int[] size, *, Generator? generator, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: randn.generator_with_names(SymInt[] size, *, Generator? generator, Dimname[]? names, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor tags: nondeterministic_seeded device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: randn autogen: randn.generator_with_names_out -- func: randn.out(int[] size, *, Tensor(a!) out) -> Tensor(a!) +- func: randn.out(SymInt[] size, *, Tensor(a!) out) -> Tensor(a!) tags: nondeterministic_seeded -- func: randn.generator_out(int[] size, *, Generator? generator, Tensor(a!) out) -> Tensor(a!) +- func: randn.generator_out(SymInt[] size, *, Generator? generator, Tensor(a!) out) -> Tensor(a!) tags: nondeterministic_seeded - func: randn_like(Tensor self, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, MemoryFormat? memory_format=None) -> Tensor tags: nondeterministic_seeded dispatch: @@ -4128,10 +4433,11 @@ - func: randperm.generator_out(int n, *, Generator? generator, Tensor(a!) out) -> Tensor(a!) tags: nondeterministic_seeded dispatch: CPU: randperm_out_cpu CUDA: randperm_out_cuda + MPS: randperm_out_mps - func: range.step(Scalar start, Scalar end, Scalar step=1, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor dispatch: CompositeExplicitAutograd: range @@ -4145,60 +4451,69 @@ - func: range.out(Scalar start, Scalar end, Scalar step=1, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, Meta: range_out CUDA: range_cuda_out + MPS: range_mps_out cpp_no_default_args: ['step'] - func: ravel(Tensor(a) self) -> Tensor(a) variants: function, method - func: reciprocal(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: reciprocal.out variants: function, method + tags: [core, pointwise] - func: reciprocal_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: reciprocal.out variants: function, method + tags: pointwise - func: reciprocal.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: reciprocal_out MPS: reciprocal_out_mps + tags: pointwise - func: neg(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: neg.out variants: function, method dispatch: SparseCPU, SparseCUDA: neg_sparse SparseCsrCPU, SparseCsrCUDA: neg_sparse_csr + NestedTensorCPU, NestedTensorCUDA: NestedTensor_neg + tags: [core, pointwise] - func: neg_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: neg.out variants: function, method dispatch: SparseCPU, SparseCUDA: neg_sparse_ SparseCsrCPU, SparseCsrCUDA: neg_sparse_csr_ + NestedTensorCPU, NestedTensorCUDA: NestedTensor_neg_ + tags: pointwise - func: neg.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: neg_out MPS: neg_out_mps SparseCPU, SparseCUDA: neg_out_sparse SparseCsrCPU, SparseCsrCUDA: neg_sparse_csr_out - + tags: pointwise # Alias for neg + - func: negative(Tensor self) -> Tensor variants: function, method - func: negative_(Tensor(a!) self) -> Tensor(a!) variants: function, method @@ -4209,33 +4524,42 @@ variants: method # This is method-only to match the previous tensor API. In the future we could make this a function too. dispatch: CompositeExplicitAutograd: repeat MPS: repeat_mps autogen: repeat.out + tags: core - func: repeat_interleave.Tensor(Tensor repeats, *, int? output_size=None) -> Tensor variants: function dispatch: CPU: repeat_interleave_cpu CUDA: repeat_interleave_cuda + MPS: repeat_interleave_mps tags: dynamic_output_shape autogen: repeat_interleave.Tensor_out - func: repeat_interleave.self_Tensor(Tensor self, Tensor repeats, int? dim=None, *, int? output_size=None) -> Tensor variants: function, method -- func: repeat_interleave.self_int(Tensor self, int repeats, int? dim=None, *, int? output_size=None) -> Tensor +- func: repeat_interleave.self_int(Tensor self, SymInt repeats, int? dim=None, *, int? output_size=None) -> Tensor variants: function, method + dispatch: + CompositeImplicitAutograd: repeat_interleave_symint - func: reshape(Tensor(a) self, SymInt[] shape) -> Tensor(a) variants: function, method device_check: NoCheck device_guard: False dispatch: CompositeImplicitAutograd: reshape_symint CompositeImplicitAutogradNestedTensor: reshape_nested +- func: _reshape_copy(Tensor self, SymInt[] size) -> Tensor + variants: function + dispatch: + CompositeExplicitAutograd: _reshape_copy_symint + # NOTE [ _reshape_alias ] is meant to be used in the implementation of reshape. # They are not user-facing, hence the leading underscore. Please don't use it # anywhere else. - func: _reshape_alias(Tensor(a) self, SymInt[] size, SymInt[] stride) -> Tensor(a) variants: function, method @@ -4265,18 +4589,20 @@ structured_delegate: round.out variants: function, method dispatch: SparseCPU, SparseCUDA: round_sparse SparseCsrCPU, SparseCsrCUDA: round_sparse_csr + tags: pointwise - func: round_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: round.out variants: function, method dispatch: SparseCPU, SparseCUDA: round_sparse_ SparseCsrCPU, SparseCsrCUDA: round_sparse_csr_ + tags: pointwise - func: round.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase @@ -4284,28 +4610,32 @@ CPU: round_out CUDA: round_out MPS: round_out_mps SparseCPU, SparseCUDA: round_sparse_out SparseCsrCPU, SparseCsrCUDA: round_sparse_csr_out + tags: pointwise - func: round.decimals(Tensor self, *, int decimals) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: round.decimals_out variants: function, method + tags: pointwise - func: round_.decimals(Tensor(a!) self, *, int decimals) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: round.decimals_out variants: function, method + tags: pointwise - func: round.decimals_out(Tensor self, *, int decimals, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU: round_decimals_out CUDA: round_decimals_out + tags: pointwise - func: rrelu(Tensor self, Scalar lower=0.125, Scalar upper=0.3333333333333333, bool training=False, Generator? generator=None) -> Tensor device_check: NoCheck # TensorIterator tags: nondeterministic_seeded @@ -4321,10 +4651,13 @@ MPS: relu_mps MkldnnCPU: mkldnn_relu QuantizedCPU: relu_quantized_cpu QuantizedCUDA: relu_quantized_cuda NestedTensorCPU, NestedTensorCUDA: NestedTensor_relu + SparseCPU, SparseCUDA: relu_sparse + SparseCsrCPU, SparseCsrCUDA: relu_sparse_csr + tags: [core, pointwise] - func: relu_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method dispatch: @@ -4332,36 +4665,37 @@ MPS: relu_mps_ MkldnnCPU: mkldnn_relu_ QuantizedCPU: relu_quantized_cpu_ QuantizedCUDA: relu_quantized_cuda_ NestedTensorCPU, NestedTensorCUDA: NestedTensor_relu_ + SparseCPU, SparseCUDA: relu_sparse_ + SparseCsrCPU, SparseCsrCUDA: relu_sparse_csr_ autogen: relu.out + tags: pointwise - func: relu6(Tensor self) -> Tensor python_module: nn - func: relu6_(Tensor(a!) self) -> Tensor(a!) python_module: nn - func: prelu(Tensor self, Tensor weight) -> Tensor variants: function, method + autogen: prelu.out + +- func: _prelu_kernel(Tensor self, Tensor weight) -> Tensor dispatch: + CPU, CUDA: _prelu_kernel + QuantizedCPU: _prelu_kernel_quantized_cpu MkldnnCPU: mkldnn_prelu - CPU: prelu_cpu - CUDA: prelu_cuda MPS: prelu_mps - QuantizedCPU: prelu_quantized_cpu - autogen: prelu.out -- func: prelu_backward(Tensor grad_output, Tensor self, Tensor weight) -> (Tensor, Tensor) - variants: function, method +- func: _prelu_kernel_backward(Tensor grad_output, Tensor self, Tensor weight) -> (Tensor, Tensor) dispatch: + CPU, CUDA: _prelu_kernel_backward MkldnnCPU: mkldnn_prelu_backward - CPU: prelu_backward_cpu - CUDA: prelu_backward_cuda MPS: prelu_backward_mps - autogen: prelu_backward.out - func: gelu.out(Tensor self, *, str approximate='none', Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator @@ -4385,10 +4719,11 @@ dispatch: MkldnnCPU: mkldnn_gelu QuantizedCPU: gelu_quantized_cpu QuantizedCUDA: gelu_quantized_cuda NestedTensorCPU, NestedTensorCUDA: NestedTensor_gelu + tags: [core, pointwise] - func: gelu_backward.grad_input(Tensor grad_output, Tensor self, *, str approximate='none', Tensor(a!) grad_input) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase python_module: nn @@ -4400,10 +4735,12 @@ - func: gelu_backward(Tensor grad_output, Tensor self, *, str approximate='none') -> Tensor structured_delegate: gelu_backward.grad_input python_module: nn dispatch: MkldnnCPU: mkldnn_gelu_backward + NestedTensorCPU, NestedTensorCUDA: gelu_backwards_nested + tags: pointwise - func: infinitely_differentiable_gelu_backward(Tensor grad, Tensor self) -> Tensor variants: function python_module: nn device_check: NoCheck @@ -4433,52 +4770,56 @@ - func: rsqrt(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: rsqrt.out variants: function, method + tags: [core, pointwise] - func: rsqrt_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: rsqrt.out variants: function, method + tags: pointwise - func: rsqrt.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: rsqrt_out MPS: rsqrt_out_mps + tags: pointwise - func: select.Dimname(Tensor(a) self, Dimname dim, int index) -> Tensor(a) variants: function, method device_check: NoCheck device_guard: False -- func: select.int(Tensor(a) self, int dim, int index) -> Tensor(a) +- func: select.int(Tensor(a) self, int dim, SymInt index) -> Tensor(a) variants: function, method device_check: NoCheck device_guard: False dispatch: - CompositeExplicitAutograd: select + CompositeExplicitAutograd: select_symint SparseCsrCPU, SparseCsrCUDA: select_sparse_csr NestedTensorCPU, NestedTensorCUDA: select_nested + tags: core -- func: select_backward(Tensor grad_output, SymInt[] input_sizes, int dim, int index) -> Tensor +- func: select_backward(Tensor grad_output, SymInt[] input_sizes, int dim, SymInt index) -> Tensor variants: function device_check: NoCheck device_guard: False dispatch: - CompositeExplicitAutogradNonFunctional: select_backward + CompositeExplicitAutogradNonFunctional: select_backward_symint autogen: select_backward.out -- func: _nested_select_backward(Tensor grad_output, Tensor self, int dim, int index) -> Tensor +- func: _nested_select_backward(Tensor grad_output, Tensor self, int dim, SymInt index) -> Tensor variants: function device_check: NoCheck device_guard: False dispatch: - NestedTensorCPU, NestedTensorCUDA: _nested_select_backward + NestedTensorCPU, NestedTensorCUDA: _nested_select_backward_symint - func: selu(Tensor self) -> Tensor device_check: NoCheck # TensorIterator - func: selu_(Tensor(a!) self) -> Tensor(a!) @@ -4551,95 +4892,109 @@ structured_delegate: sigmoid.out variants: function, method dispatch: QuantizedCPU: sigmoid_quantized_cpu MkldnnCPU: mkldnn_sigmoid + tags: [core, pointwise] - func: sigmoid_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: sigmoid.out variants: function, method dispatch: MkldnnCPU: mkldnn_sigmoid_ + tags: pointwise - func: sigmoid.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: sigmoid_out MPS: sigmoid_out_mps + tags: pointwise - func: logit(Tensor self, float? eps=None) -> Tensor variants: function, method dispatch: CPU, CUDA: logit + tags: pointwise - func: logit_(Tensor(a!) self, float? eps=None) -> Tensor(a!) variants: function, method dispatch: CPU, CUDA: logit_ + tags: pointwise - func: logit.out(Tensor self, float? eps=None, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: logit_out + tags: pointwise - func: sin(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: sin.out variants: function, method dispatch: SparseCsrCPU, SparseCsrCUDA: sin_sparse_csr SparseCPU, SparseCUDA: sin_sparse + tags: [core, pointwise] - func: sin_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: sin.out variants: function, method dispatch: SparseCsrCPU, SparseCsrCUDA: sin_sparse_csr_ SparseCPU, SparseCUDA: sin_sparse_ + tags: pointwise - func: sin.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: sin_out MPS: sin_out_mps SparseCsrCPU, SparseCsrCUDA: sin_sparse_csr_out SparseCPU, SparseCUDA: sin_sparse_out + tags: pointwise - func: sinc(Tensor self) -> Tensor structured_delegate: sinc.out variants: function, method + tags: pointwise - func: sinc_(Tensor(a!) self) -> Tensor(a!) structured_delegate: sinc.out variants: function, method + tags: pointwise - func: sinc.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: sinc_out + tags: pointwise - func: sinh(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: sinh.out variants: function, method dispatch: SparseCPU, SparseCUDA: sinh_sparse SparseCsrCPU, SparseCsrCUDA: sinh_sparse_csr + tags: [core, pointwise] - func: sinh_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: sinh.out variants: function, method dispatch: SparseCPU, SparseCUDA: sinh_sparse_ SparseCsrCPU, SparseCsrCUDA: sinh_sparse_csr_ + tags: pointwise - func: sinh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase @@ -4658,10 +5013,11 @@ # those metadata changes to the detached tensor will not update the original tensor # anymore, and in the `detach()` function we need to set `allow_tensor_metadata_change_` # to false to make such changes explicitly illegal, in order to prevent users from # changing metadata of the detached tensor and expecting the original tensor to also # be updated. + tags: pointwise - func: detach(Tensor(a) self) -> Tensor(a) variants: function, method dispatch: CompositeExplicitAutograd: detach NestedTensorCPU, NestedTensorCUDA: detach @@ -4690,10 +5046,12 @@ variants: function, method device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: slice + tags: core + # NOTE: The implementation of split_with_sizes bypasses the dispatcher to call this; undo # that if adding specific implementations here! - func: slice_backward(Tensor grad_output, SymInt[] input_sizes, int dim, SymInt start, SymInt end, SymInt step) -> Tensor variants: function @@ -4708,17 +5066,18 @@ device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: slice_scatter autogen: slice_scatter.out + tags: core -- func: select_scatter(Tensor self, Tensor src, int dim, int index) -> Tensor +- func: select_scatter(Tensor self, Tensor src, int dim, SymInt index) -> Tensor variants: function, method device_check: NoCheck device_guard: False dispatch: - CompositeExplicitAutograd: select_scatter + CompositeExplicitAutograd: select_scatter_symint autogen: select_scatter.out - func: diagonal_scatter(Tensor self, Tensor src, int offset=0, int dim1=0, int dim2=1) -> Tensor variants: function, method device_check: NoCheck @@ -4753,10 +5112,11 @@ - func: _softmax(Tensor self, int dim, bool half_to_float) -> Tensor structured_delegate: _softmax.out dispatch: MkldnnCPU: mkldnn_softmax NestedTensorCPU, NestedTensorCUDA: softmax_nested + tags: core - func: _softmax.out(Tensor self, int dim, bool half_to_float, *, Tensor(a!) out) -> Tensor(a!) structured: True dispatch: CPU: softmax_cpu_out @@ -4773,38 +5133,40 @@ dispatch: CPU: softmax_backward_cpu_out CUDA: softmax_backward_cuda_out MPS: softmax_backward_mps_out -- func: unsafe_split.Tensor(Tensor self, int split_size, int dim=0) -> Tensor[] +- func: unsafe_split.Tensor(Tensor self, SymInt split_size, int dim=0) -> Tensor[] variants: function, method device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: unsafe_split autogen: unsafe_split.Tensor_out -- func: split.Tensor(Tensor(a -> *) self, int split_size, int dim=0) -> Tensor(a)[] +- func: split.Tensor(Tensor(a -> *) self, SymInt split_size, int dim=0) -> Tensor(a)[] variants: function, method device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: split -- func: split.sizes(Tensor(a -> *) self, int[] split_size, int dim=0) -> Tensor(a)[] +- func: split.sizes(Tensor(a -> *) self, SymInt[] split_size, int dim=0) -> Tensor(a)[] variants: function, method device_guard: False + dispatch: + CompositeImplicitAutograd: split_symint -- func: unsafe_split_with_sizes(Tensor self, int[] split_sizes, int dim=0) -> Tensor[] +- func: unsafe_split_with_sizes(Tensor self, SymInt[] split_sizes, int dim=0) -> Tensor[] variants: function, method device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: unsafe_split_with_sizes autogen: unsafe_split_with_sizes.out -- func: split_with_sizes(Tensor(a -> *) self, int[] split_sizes, int dim=0) -> Tensor(a)[] +- func: split_with_sizes(Tensor(a -> *) self, SymInt[] split_sizes, int dim=0) -> Tensor(a)[] variants: function, method device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: split_with_sizes @@ -4832,24 +5194,38 @@ device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: squeeze QuantizedCPU, QuantizedCUDA: squeeze_quantized + NestedTensorCPU, NestedTensorCUDA: squeeze_nested - func: squeeze.dim(Tensor(a) self, int dim) -> Tensor(a) variants: function, method device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: squeeze QuantizedCPU, QuantizedCUDA: squeeze_quantized + NestedTensorCPU, NestedTensorCUDA: squeeze_dim_nested + tags: core - func: squeeze.dimname(Tensor(a) self, Dimname dim) -> Tensor(a) variants: function, method device_check: NoCheck device_guard: False + +- func: squeeze.dims(Tensor(a) self, int[] dim) -> Tensor(a) + variants: function, method + device_check: NoCheck + device_guard: False + dispatch: + CompositeExplicitAutograd: squeeze + QuantizedCPU, QuantizedCUDA: squeeze_quantized + NestedTensorCPU, NestedTensorCUDA: squeeze_dim_nested + tags: core + - func: squeeze_(Tensor(a!) self) -> Tensor(a!) variants: method device_check: NoCheck device_guard: False tags: inplace_view @@ -4862,10 +5238,18 @@ device_guard: False tags: inplace_view dispatch: CompositeExplicitAutograd: squeeze_ +- func: squeeze_.dims(Tensor(a!) self, int[] dim) -> Tensor(a!) + variants: method + device_check: NoCheck + device_guard: False + tags: inplace_view + dispatch: + CompositeExplicitAutograd: squeeze_ + - func: squeeze_.dimname(Tensor(a!) self, Dimname dim) -> Tensor(a!) variants: method device_check: NoCheck device_guard: False tags: inplace_view @@ -4935,19 +5319,22 @@ - func: sum(Tensor self, *, ScalarType? dtype=None) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: sum + SparseCPU, SparseCUDA: sum_coo SparseCsrCPU, SparseCsrCUDA: sum_csr autogen: sum.out - func: sum.dim_IntList(Tensor self, int[1]? dim, bool keepdim=False, *, ScalarType? dtype=None) -> Tensor structured_delegate: sum.IntList_out device_check: NoCheck # TensorIterator variants: function, method dispatch: NestedTensorCPU: NestedTensor_sum_dim_CPU + SparseCPU, SparseCUDA: sum_sparse_coo + tags: core - func: sum.dim_DimnameList(Tensor self, Dimname[1] dim, bool keepdim=False, *, ScalarType? dtype=None) -> Tensor device_check: NoCheck # TensorIterator variants: function, method @@ -4968,14 +5355,16 @@ - func: nansum(Tensor self, int[1]? dim=None, bool keepdim=False, *, ScalarType? dtype=None) -> Tensor variants: function, method dispatch: CPU, CUDA: nansum + MPS: nansum_mps - func: nansum.out(Tensor self, int[1]? dim=None, bool keepdim=False, *, ScalarType? dtype=None, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: nansum_out + MPS: nansum_out_mps - func: sum_to_size(Tensor self, int[] size) -> Tensor variants: method device_check: NoCheck device_guard: False @@ -4985,99 +5374,113 @@ structured_delegate: sqrt.out variants: function, method dispatch: SparseCPU, SparseCUDA: sqrt_sparse SparseCsrCPU, SparseCsrCUDA: sqrt_sparse_csr + tags: [core, pointwise] - func: sqrt_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: sqrt.out variants: function, method dispatch: SparseCPU, SparseCUDA: sqrt_sparse_ SparseCsrCPU, SparseCsrCUDA: sqrt_sparse_csr_ + tags: pointwise - func: sqrt.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: sqrt_out MPS: sqrt_out_mps SparseCPU, SparseCUDA: sqrt_sparse_out SparseCsrCPU, SparseCsrCUDA: sqrt_sparse_csr_out + tags: pointwise - func: square(Tensor self) -> Tensor device_check: NoCheck # TensorIterator variants: function, method + tags: pointwise - func: square_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function, method + tags: pointwise - func: square.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) + tags: pointwise - func: std(Tensor self, bool unbiased=True) -> Tensor device_check: NoCheck # TensorIterator variants: function, method + cpp_no_default_args: ["unbiased"] - func: std.dim(Tensor self, int[1]? dim, bool unbiased=True, bool keepdim=False) -> Tensor device_check: NoCheck # TensorIterator variants: function, method + cpp_no_default_args: ["unbiased"] -- func: std.correction(Tensor self, int[1]? dim, *, int? correction, bool keepdim=False) -> Tensor +- func: std.correction(Tensor self, int[1]? dim=None, *, int? correction=None, bool keepdim=False) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CPU, CUDA: std MPS: std_mps QuantizedCPU: std_quantized_cpu - func: std_mean(Tensor self, bool unbiased=True) -> (Tensor, Tensor) device_check: NoCheck # TensorIterator variants: function + cpp_no_default_args: ["unbiased"] - func: std_mean.dim(Tensor self, int[1]? dim, bool unbiased=True, bool keepdim=False) -> (Tensor, Tensor) device_check: NoCheck # TensorIterator variants: function + cpp_no_default_args: ["unbiased"] -- func: std_mean.correction(Tensor self, int[1]? dim, *, int? correction, bool keepdim=False) -> (Tensor, Tensor) +- func: std_mean.correction(Tensor self, int[1]? dim=None, *, int? correction=None, bool keepdim=False) -> (Tensor, Tensor) device_check: NoCheck # TensorIterator variants: function dispatch: CPU, CUDA: std_mean autogen: std_mean.correction_out - func: std_mean.names_dim(Tensor self, Dimname[1] dim, bool unbiased=True, bool keepdim=False) -> (Tensor, Tensor) device_check: NoCheck # TensorIterator variants: function + cpp_no_default_args: ["unbiased"] -- func: std_mean.correction_names(Tensor self, Dimname[1] dim, *, int? correction, bool keepdim=False) -> (Tensor, Tensor) +- func: std_mean.correction_names(Tensor self, Dimname[1] dim, *, int? correction=None, bool keepdim=False) -> (Tensor, Tensor) device_check: NoCheck # TensorIterator variants: function - func: std.out(Tensor self, int[1]? dim, bool unbiased=True, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator + cpp_no_default_args: ["unbiased"] -- func: std.correction_out(Tensor self, int[1]? dim, *, int? correction, bool keepdim=False, Tensor(a!) out) -> Tensor(a!) +- func: std.correction_out(Tensor self, int[1]? dim=None, *, int? correction=None, bool keepdim=False, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: std_out QuantizedCPU: std_out_quantized_cpu - func: std.names_dim(Tensor self, Dimname[1] dim, bool unbiased=True, bool keepdim=False) -> Tensor device_check: NoCheck # TensorIterator variants: function, method + cpp_no_default_args: ["unbiased"] - func: std.names_out(Tensor self, Dimname[1] dim, bool unbiased=True, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator + cpp_no_default_args: ["unbiased"] -- func: std.correction_names(Tensor self, Dimname[1] dim, *, int? correction, bool keepdim=False) -> Tensor +- func: std.correction_names(Tensor self, Dimname[1] dim, *, int? correction=None, bool keepdim=False) -> Tensor device_check: NoCheck # TensorIterator variants: function, method -- func: std.correction_names_out(Tensor self, Dimname[1] dim, *, int? correction, bool keepdim=False, Tensor(a!) out) -> Tensor(a!) +- func: std.correction_names_out(Tensor self, Dimname[1] dim, *, int? correction=None, bool keepdim=False, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function - func: prod(Tensor self, *, ScalarType? dtype=None) -> Tensor device_check: NoCheck # TensorIterator @@ -5126,57 +5529,65 @@ structured_delegate: tan.out variants: function, method dispatch: SparseCPU, SparseCUDA: tan_sparse SparseCsrCPU, SparseCsrCUDA: tan_sparse_csr + tags: pointwise - func: tan_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: tan.out variants: function, method dispatch: SparseCPU, SparseCUDA: tan_sparse_ SparseCsrCPU, SparseCsrCUDA: tan_sparse_csr_ + tags: pointwise - func: tan.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: tan_out MPS: tan_out_mps SparseCPU, SparseCUDA: tan_sparse_out SparseCsrCPU, SparseCsrCUDA: tan_sparse_csr_out + tags: pointwise - func: tanh(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: tanh.out variants: function, method dispatch: QuantizedCPU: tanh_quantized_cpu MkldnnCPU: mkldnn_tanh SparseCPU, SparseCUDA: tanh_sparse SparseCsrCPU, SparseCsrCUDA: tanh_sparse_csr + NestedTensorCPU, NestedTensorCUDA: NestedTensor_tanh + tags: [core, pointwise] - func: tanh_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: tanh.out variants: function, method dispatch: MkldnnCPU: mkldnn_tanh_ SparseCPU, SparseCUDA: tanh_sparse_ SparseCsrCPU, SparseCsrCUDA: tanh_sparse_csr_ + NestedTensorCPU, NestedTensorCUDA: NestedTensor_tanh_ + tags: pointwise - func: tanh.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: tanh_out MPS: tanh_out_mps SparseCPU, SparseCUDA: tanh_sparse_out SparseCsrCPU, SparseCsrCUDA: tanh_sparse_csr_out + tags: pointwise - func: tensordot(Tensor self, Tensor other, int[] dims_self, int[] dims_other) -> Tensor variants: function - func: tensordot.out(Tensor self, Tensor other, int[] dims_self, int[] dims_other, *, Tensor(a!) out) -> Tensor(a!) @@ -5209,16 +5620,22 @@ structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: threshold_backward_out MPS: threshold_backward_out_mps + SparseCPU, SparseCUDA: threshold_backward_sparse_out + SparseCsrCPU, SparseCsrCUDA: threshold_backward_sparse_compressed_out - func: threshold_backward(Tensor grad_output, Tensor self, Scalar threshold) -> Tensor variants: function structured_delegate: threshold_backward.grad_input dispatch: MkldnnCPU: mkldnn_relu_backward + SparseCPU, SparseCUDA: threshold_backward_sparse + SparseCsrCPU, SparseCsrCUDA: threshold_backward_sparse_compressed + NestedTensorCPU, NestedTensorCUDA: threshold_backwards_nested + tags: pointwise - func: tile(Tensor self, int[] dims) -> Tensor variants: function, method - func: transpose.int(Tensor(a) self, int dim0, int dim1) -> Tensor(a) @@ -5264,10 +5681,11 @@ variants: function, method dispatch: CPU, QuantizedCPU, CUDA, QuantizedCUDA: flip MPS: flip_mps autogen: flip.out + tags: core - func: fliplr(Tensor self) -> Tensor variants: function, method - func: flipud(Tensor self) -> Tensor @@ -5373,30 +5791,33 @@ device_check: NoCheck # TensorIterator variants: function, method dispatch: SparseCPU, SparseCUDA: trunc_sparse SparseCsrCPU, SparseCsrCUDA: trunc_sparse_csr + tags: pointwise - func: trunc_(Tensor(a!) self) -> Tensor(a!) structured_delegate: trunc.out device_check: NoCheck # TensorIterator variants: function, method dispatch: SparseCPU, SparseCUDA: trunc_sparse_ SparseCsrCPU, SparseCsrCUDA: trunc_sparse_csr_ + tags: pointwise - func: trunc.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: trunc_out MPS: trunc_out_mps SparseCPU, SparseCUDA: trunc_sparse_out SparseCsrCPU, SparseCsrCUDA: trunc_sparse_csr_out - + tags: pointwise # Alias for trunc + - func: fix(Tensor self) -> Tensor variants: function, method - func: fix_(Tensor(a!) self) -> Tensor(a!) variants: function, method @@ -5427,18 +5848,20 @@ - func: unique_consecutive(Tensor self, bool return_inverse=False, bool return_counts=False, int? dim=None) -> (Tensor, Tensor, Tensor) variants: function dispatch: CPU: unique_consecutive_cpu CUDA: unique_consecutive_cuda + MPS: unique_consecutive_mps tags: dynamic_output_shape autogen: unique_consecutive.out - func: unique_dim_consecutive(Tensor self, int dim, bool return_inverse=False, bool return_counts=False) -> (Tensor, Tensor, Tensor) variants: function dispatch: CPU: unique_dim_consecutive_cpu CUDA: unique_dim_consecutive_cuda + MPS: unique_dim_consecutive_mps tags: dynamic_output_shape autogen: unique_dim_consecutive.out # _unique and _unique_dim are fragile and modifying them easily cause internal break # the below operator is a temporary hack for adding return_counts support @@ -5447,10 +5870,11 @@ - func: _unique2(Tensor self, bool sorted=True, bool return_inverse=False, bool return_counts=False) -> (Tensor, Tensor, Tensor) variants: function dispatch: CPU: _unique2_cpu CUDA: _unique2_cuda + MPS: _unique2_mps tags: dynamic_output_shape autogen: _unique2.out - func: _unsafe_view(Tensor self, SymInt[] size) -> Tensor dispatch: @@ -5463,10 +5887,12 @@ device_guard: False dispatch: CompositeExplicitAutograd: unsqueeze SparseCPU, SparseCUDA: unsqueeze_sparse QuantizedCPU, QuantizedCUDA: unsqueeze_quantized + NestedTensorCPU, NestedTensorCUDA: unsqueeze_nested + tags: core - func: unsqueeze_(Tensor(a!) self, int dim) -> Tensor(a!) variants: method device_check: NoCheck device_guard: False @@ -5477,65 +5903,74 @@ - func: vander(Tensor x, int? N=None, bool increasing=False) -> Tensor - func: var(Tensor self, bool unbiased=True) -> Tensor device_check: NoCheck # TensorIterator variants: function, method + cpp_no_default_args: ["unbiased"] - func: var.dim(Tensor self, int[1]? dim, bool unbiased=True, bool keepdim=False) -> Tensor device_check: NoCheck # TensorIterator variants: function, method + tags: core + cpp_no_default_args: ["unbiased"] -- func: var.correction(Tensor self, int[1]? dim, *, int? correction, bool keepdim=False) -> Tensor +- func: var.correction(Tensor self, int[1]? dim=None, *, int? correction=None, bool keepdim=False) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CPU, CUDA: var MPS: var_mps - func: var.out(Tensor self, int[1]? dim, bool unbiased=True, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator + cpp_no_default_args: ["unbiased"] -- func: var.correction_out(Tensor self, int[1]? dim, *, int? correction, bool keepdim=False, Tensor(a!) out) -> Tensor(a!) +- func: var.correction_out(Tensor self, int[1]? dim=None, *, int? correction=None, bool keepdim=False, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: var_out - func: var.names_dim(Tensor self, Dimname[1] dim, bool unbiased=True, bool keepdim=False) -> Tensor device_check: NoCheck # TensorIterator variants: function, method + cpp_no_default_args: ["unbiased"] - func: var.names_out(Tensor self, Dimname[1] dim, bool unbiased=True, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator + cpp_no_default_args: ["unbiased"] -- func: var.correction_names(Tensor self, Dimname[1] dim, *, int? correction, bool keepdim=False) -> Tensor +- func: var.correction_names(Tensor self, Dimname[1] dim, *, int? correction=None, bool keepdim=False) -> Tensor device_check: NoCheck # TensorIterator variants: function, method -- func: var.correction_names_out(Tensor self, Dimname[1] dim, *, int? correction, bool keepdim=False, Tensor(a!) out) -> Tensor(a!) +- func: var.correction_names_out(Tensor self, Dimname[1] dim, *, int? correction=None, bool keepdim=False, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function - func: var_mean(Tensor self, bool unbiased=True) -> (Tensor, Tensor) device_check: NoCheck # TensorIterator variants: function + cpp_no_default_args: ["unbiased"] - func: var_mean.dim(Tensor self, int[1]? dim, bool unbiased=True, bool keepdim=False) -> (Tensor, Tensor) device_check: NoCheck # TensorIterator variants: function + cpp_no_default_args: ["unbiased"] -- func: var_mean.correction(Tensor self, int[1]? dim, *, int? correction, bool keepdim=False) -> (Tensor, Tensor) +- func: var_mean.correction(Tensor self, int[1]? dim=None, *, int? correction=None, bool keepdim=False) -> (Tensor, Tensor) device_check: NoCheck # TensorIterator variants: function dispatch: CPU, CUDA: var_mean autogen: var_mean.correction_out - func: var_mean.names_dim(Tensor self, Dimname[1] dim, bool unbiased=True, bool keepdim=False) -> (Tensor, Tensor) device_check: NoCheck # TensorIterator variants: function + cpp_no_default_args: ["unbiased"] -- func: var_mean.correction_names(Tensor self, Dimname[1] dim, *, int? correction, bool keepdim=False) -> (Tensor, Tensor) +- func: var_mean.correction_names(Tensor self, Dimname[1] dim, *, int? correction=None, bool keepdim=False) -> (Tensor, Tensor) device_check: NoCheck # TensorIterator variants: function - func: view_as(Tensor(a) self, Tensor other) -> Tensor(a) variants: method @@ -5546,10 +5981,11 @@ device_check: NoCheck # TensorIterator variants: function, method dispatch: CPU, CUDA: where MPS: where_mps + tags: [core, pointwise] - func: where.self_out(Tensor condition, Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: where_self_out @@ -5557,11 +5993,11 @@ - func: where.ScalarSelf(Tensor condition, Scalar self, Tensor other) -> Tensor variants: function - func: where.ScalarOther(Tensor condition, Tensor self, Scalar other) -> Tensor - variants: function + variants: function, method - func: where.Scalar(Tensor condition, Scalar self, Scalar other) -> Tensor variants: function - func: where(Tensor condition) -> Tensor[] @@ -5602,10 +6038,11 @@ - func: _efficientzerotensor(int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor dispatch: CPU: _efficientzerotensor CUDA: _efficientzerotensor_cuda + Meta: _efficientzerotensor_meta autogen: _efficientzerotensor.out - func: zeros(SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor dispatch: CompositeExplicitAutograd: zeros_symint @@ -5789,10 +6226,11 @@ - func: norm.dtype_out(Tensor self, Scalar? p, int[1] dim, bool keepdim, *, ScalarType dtype, Tensor(a!) out) -> Tensor(a!) structured: True device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: norm_dtype_out + MPS: norm_dtype_out_mps - func: norm.out(Tensor self, Scalar? p, int[1] dim, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!) structured: True device_check: NoCheck # TensorIterator dispatch: @@ -5816,20 +6254,18 @@ - func: frexp.Tensor(Tensor self) -> (Tensor mantissa, Tensor exponent) variants: method, function dispatch: CompositeExplicitAutograd: frexp + tags: pointwise - func: frexp.Tensor_out(Tensor self, *, Tensor(a!) mantissa, Tensor(b!) exponent) -> (Tensor(a!) mantissa, Tensor(b!) exponent) dispatch: CPU, CUDA: frexp_out + tags: pointwise # Deprecated (v.1.12) -- func: frobenius_norm(Tensor self) -> Tensor - variants: function - -# Deprecated (v.1.12) - func: frobenius_norm.dim(Tensor self, int[1] dim, bool keepdim=False) -> Tensor variants: function # Deprecated (v.1.12) - func: frobenius_norm.out(Tensor self, int[1] dim, bool keepdim=False, *, Tensor(a!) out) -> Tensor(a!) @@ -5859,13 +6295,15 @@ SparseCsrCPU, SparseCsrCUDA: clone_sparse_compressed MkldnnCPU: mkldnn_clone QuantizedCPU, QuantizedCUDA: quantized_clone NestedTensorCPU, NestedTensorCUDA: clone_nested autogen: clone.out + tags: core - func: positive(Tensor(a) self) -> Tensor(a) variants: function, method + tags: pointwise - func: resize_as_(Tensor(a!) self, Tensor the_template, *, MemoryFormat? memory_format=None) -> Tensor(a!) use_const_ref_for_mutable_tensors: True variants: function, method dispatch: @@ -5898,41 +6336,46 @@ structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: sub_out MPS: sub_out_mps SparseCPU, SparseCUDA: sub_out_sparse + tags: pointwise - func: sub.Tensor(Tensor self, Tensor other, *, Scalar alpha=1) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: sub.out dispatch: SparseCPU, SparseCUDA: sub_sparse ZeroTensor: sub_zerotensor + tags: [core, pointwise] - func: sub_.Tensor(Tensor(a!) self, Tensor other, *, Scalar alpha=1) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: sub.out dispatch: SparseCPU, SparseCUDA: sub_sparse_ - + tags: pointwise # For C++ only, until we have conversion from C++ numbers to Tensor + - func: sub.Scalar(Tensor self, Scalar other, Scalar alpha=1) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: sub + tags: [core, pointwise] - func: sub_.Scalar(Tensor(a!) self, Scalar other, Scalar alpha=1) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: sub_ autogen: sub.Scalar_out - + tags: pointwise # subtract, alias for sub + - func: subtract.out(Tensor self, Tensor other, *, Scalar alpha=1, Tensor(a!) out) -> Tensor(a!) - func: subtract.Tensor(Tensor self, Tensor other, *, Scalar alpha=1) -> Tensor variants: function, method @@ -5957,15 +6400,17 @@ structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: heaviside_out + tags: pointwise - func: heaviside(Tensor self, Tensor values) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: heaviside.out + tags: pointwise - func: heaviside_(Tensor(a!) self, Tensor values) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: heaviside.out @@ -5978,10 +6423,11 @@ CompositeExplicitAutograd: rsub autogen: rsub.Scalar_out # Functionally the same as addmm, but we give it a different derivative formula # that doesn't propagate gradients to non-present entries on sparse. + tags: pointwise - func: _sparse_addmm(Tensor self, Tensor mat1, Tensor mat2, *, Scalar beta=1, Scalar alpha=1) -> Tensor python_module: sparse dispatch: CompositeExplicitAutograd: _sparse_addmm autogen: _sparse_addmm.out @@ -5996,10 +6442,20 @@ python_module: sparse dispatch: SparseCsrCUDA: sparse_sampled_addmm_sparse_csr_cuda SparseCsrCPU: sparse_sampled_addmm_sparse_csr_cpu +- func: _sparse_mm_reduce_impl(Tensor self, Tensor other, str reduce) -> (Tensor, Tensor) + python_module: sparse + dispatch: + SparseCsrCPU: _sparse_mm_reduce_impl_sparse_csr_cpu + +- func: _sparse_mm_reduce_impl_backward(Tensor self, Tensor grad_out, Tensor weight, str reduce, Tensor arg_out, bool[2] output_mask) -> (Tensor, Tensor) + python_module: sparse + dispatch: + SparseCsrCPU: _sparse_mm_reduce_impl_backward_sparse_csr_cpu + - func: addmm.out(Tensor self, Tensor mat1, Tensor mat2, *, Scalar beta=1, Scalar alpha=1, Tensor(a!) out) -> Tensor(a!) structured: True dispatch: CPU: addmm_out_cpu CUDA: addmm_out_cuda @@ -6014,10 +6470,11 @@ variants: function, method dispatch: SparseCPU: addmm_sparse_dense_cpu SparseCUDA: addmm_sparse_dense_cuda SparseCsrCPU, SparseCsrCUDA: addmm_sparse_compressed_dense + tags: core - func: addmm_(Tensor(a!) self, Tensor mat1, Tensor mat2, *, Scalar beta=1, Scalar alpha=1) -> Tensor(a!) structured_delegate: addmm.out variants: method dispatch: @@ -6172,11 +6629,13 @@ - func: sparse_coo_tensor.indices(Tensor indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor - func: sparse_coo_tensor.indices_size(Tensor indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor -- func: _sparse_coo_tensor_unsafe(Tensor indices, Tensor values, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: _sparse_coo_tensor_unsafe(Tensor indices, Tensor values, SymInt[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor + dispatch: + CompositeImplicitAutograd: _sparse_coo_tensor_unsafe_symint - func: _validate_sparse_coo_tensor_args(Tensor indices, Tensor values, int[] size) -> () - func: _validate_sparse_compressed_tensor_args(Tensor compressed_indices, Tensor plain_indices, Tensor values, int[] size, Layout layout) -> () - func: _validate_sparse_csr_tensor_args(Tensor crow_indices, Tensor col_indices, Tensor values, int[] size) -> () @@ -6187,13 +6646,13 @@ - func: _sparse_coo_tensor_with_dims(int sparse_dim, int dense_dim, int[] size, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor dispatch: SparseCPU, SparseCUDA, SparseMeta, Meta: new_with_dims_sparse autogen: _sparse_coo_tensor_with_dims.out -- func: _sparse_coo_tensor_with_dims_and_tensors(int sparse_dim, int dense_dim, int[] size, Tensor indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor +- func: _sparse_coo_tensor_with_dims_and_tensors(int sparse_dim, int dense_dim, SymInt[] size, Tensor indices, Tensor values, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=False) -> Tensor dispatch: - SparseCPU, SparseCUDA, SparseMeta, Meta: new_with_dims_and_tensor_sparse + SparseCPU, SparseCUDA, SparseMeta, Meta: new_with_dims_and_tensor_sparse_symint autogen: _sparse_coo_tensor_with_dims_and_tensors.out - func: sparse_resize_(Tensor(a!) self, int[] size, int sparse_dim, int dense_dim) -> Tensor(a!) use_const_ref_for_mutable_tensors: True variants: method @@ -6209,12 +6668,11 @@ autogen: sparse_resize_and_clear, sparse_resize_and_clear.out - func: sparse_mask(Tensor self, Tensor mask) -> Tensor variants: method dispatch: - SparseCPU: sparse_mask_cpu - SparseCUDA: sparse_mask_cuda + SparseCPU, SparseCUDA: sparse_mask SparseCsrCPU, SparseCsrCUDA: sparse_mask_sparse_csr autogen: sparse_mask.out - func: _to_cpu(Tensor[] tensors) -> Tensor[] variants: function @@ -6234,10 +6692,11 @@ - func: to_dense_backward(Tensor grad, Tensor input) -> Tensor - func: sparse_dim(Tensor self) -> int variants: method dispatch: + CPU, CUDA: sparse_dim_strided SparseCPU, SparseCUDA, SparseMeta: sparse_dim_sparse SparseCsrCPU, SparseCsrCUDA: sparse_dim_sparse_csr device_check: NoCheck device_guard: False @@ -6250,10 +6709,11 @@ device_guard: False - func: dense_dim(Tensor self) -> int variants: method dispatch: + CPU, CUDA: dense_dim_strided SparseCPU, SparseCUDA, SparseMeta: dense_dim_sparse SparseCsrCPU, SparseCsrCUDA: dense_dim_sparse_csr device_check: NoCheck device_guard: False @@ -6289,10 +6749,11 @@ - func: is_coalesced(Tensor self) -> bool variants: method dispatch: SparseCPU, SparseCUDA, SparseMeta: is_coalesced_sparse + CompositeExplicitAutograd: is_coalesced_default device_check: NoCheck device_guard: False - func: _indices(Tensor(a) self) -> Tensor(a) variants: method @@ -6321,47 +6782,53 @@ - func: indices(Tensor(a) self) -> Tensor(a) variants: method dispatch: SparseCPU, SparseCUDA, SparseMeta: indices_sparse + CompositeExplicitAutograd: indices_default device_check: NoCheck device_guard: False - func: values(Tensor(a) self) -> Tensor(a) variants: method dispatch: SparseCPU, SparseCUDA, SparseMeta: values_sparse SparseCsrCPU, SparseCsrCUDA: values_sparse_csr NestedTensorCPU, NestedTensorCUDA: values_nested + CompositeExplicitAutograd: values_default device_check: NoCheck device_guard: False - func: crow_indices(Tensor(a) self) -> Tensor(a) variants: method dispatch: SparseCsrCPU, SparseCsrCUDA: crow_indices_sparse_csr + CompositeExplicitAutograd: crow_indices_default device_check: NoCheck device_guard: False - func: col_indices(Tensor(a) self) -> Tensor(a) variants: method dispatch: SparseCsrCPU, SparseCsrCUDA: col_indices_sparse_csr + CompositeExplicitAutograd: col_indices_default device_check: NoCheck device_guard: False - func: ccol_indices(Tensor(a) self) -> Tensor(a) variants: method dispatch: SparseCsrCPU, SparseCsrCUDA: ccol_indices_sparse_csr + CompositeExplicitAutograd: ccol_indices_default device_check: NoCheck device_guard: False - func: row_indices(Tensor(a) self) -> Tensor(a) variants: method dispatch: SparseCsrCPU, SparseCsrCUDA: row_indices_sparse_csr + CompositeExplicitAutograd: row_indices_default device_check: NoCheck device_guard: False - func: hspmm.out(Tensor mat1, Tensor mat2, *, Tensor(a!) out) -> Tensor(a!) dispatch: @@ -6392,45 +6859,47 @@ - func: to_sparse.sparse_dim(Tensor self, int sparse_dim) -> Tensor variants: method dispatch: CPU, CUDA: dense_to_sparse + SparseCPU, SparseCUDA: sparse_coo_to_sparse SparseCsrCPU, SparseCsrCUDA: sparse_compressed_to_sparse autogen: to_sparse.sparse_dim_out -- func: to_sparse(Tensor self) -> Tensor +- func: to_sparse(Tensor self, *, Layout? layout=None, int[2]? blocksize=None, int? dense_dim=None) -> Tensor variants: method dispatch: CPU, CUDA: dense_to_sparse + SparseCPU, SparseCUDA: sparse_coo_to_sparse SparseCsrCPU, SparseCsrCUDA: sparse_compressed_to_sparse autogen: to_sparse.out -- func: to_sparse_csr(Tensor self) -> Tensor +- func: to_sparse_csr(Tensor self, int? dense_dim=None) -> Tensor variants: method dispatch: CPU, CUDA: dense_to_sparse_csr SparseCPU, SparseCUDA: coo_to_sparse_csr SparseCsrCPU, SparseCsrCUDA: sparse_compressed_to_sparse_csr autogen: to_sparse_csr.out -- func: to_sparse_csc(Tensor self) -> Tensor +- func: to_sparse_csc(Tensor self, int? dense_dim=None) -> Tensor variants: method dispatch: CPU, CUDA: dense_to_sparse_csc SparseCPU, SparseCUDA: coo_to_sparse_csc SparseCsrCPU, SparseCsrCUDA: sparse_compressed_to_sparse_csc autogen: to_sparse_csc.out -- func: to_sparse_bsr(Tensor self, int[2] blocksize) -> Tensor +- func: to_sparse_bsr(Tensor self, int[2] blocksize, int? dense_dim=None) -> Tensor variants: method dispatch: CPU, CUDA: dense_to_sparse_bsr SparseCPU, SparseCUDA: coo_to_sparse_bsr SparseCsrCPU, SparseCsrCUDA: sparse_compressed_to_sparse_bsr autogen: to_sparse_bsr.out -- func: to_sparse_bsc(Tensor self, int[2] blocksize) -> Tensor +- func: to_sparse_bsc(Tensor self, int[2] blocksize, int? dense_dim=None) -> Tensor variants: method dispatch: CPU, CUDA: dense_to_sparse_bsc SparseCPU, SparseCUDA: coo_to_sparse_bsc SparseCsrCPU, SparseCsrCUDA: sparse_compressed_to_sparse_bsc @@ -6440,11 +6909,11 @@ variants: method dispatch: CPU: dense_to_mkldnn autogen: to_mkldnn.out -- func: mkldnn_reorder_conv2d_weight(Tensor self, int[2] padding=0, int[2] stride=1, int[2] dilation=1, int groups=1) -> Tensor +- func: mkldnn_reorder_conv2d_weight(Tensor self, int[2] padding=0, int[2] stride=1, int[2] dilation=1, int groups=1, int[]? input_size=None) -> Tensor variants: function python_module: nn dispatch: MkldnnCPU: mkldnn_reorder_conv2d_weight autogen: mkldnn_reorder_conv2d_weight.out @@ -6640,11 +7109,13 @@ - func: _to_copy(Tensor self, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, bool non_blocking=False, MemoryFormat? memory_format=None) -> Tensor device_check: NoCheck device_guard: False dispatch: CompositeExplicitAutograd: _to_copy + NestedTensorCPU, NestedTensorCUDA: _to_copy_nested autogen: _to_copy.out + tags: core # to(Device) must not exist because all constructors of Device also works for # TensorOptions. Otherwise, an ambiguity error is thrown. # See NOTE [ TensorOptions Constructors ]. - func: to.dtype_layout(Tensor(a) self, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, bool non_blocking=False, bool copy=False, MemoryFormat? memory_format=None) -> Tensor(a) @@ -6710,16 +7181,16 @@ MPS: _local_scalar_dense_mps variants: function # MPS LSTM implementation -- func: _lstm_mps(Tensor input, Tensor[] hx, Tensor[] params, bool has_biases, int num_layers, float dropout, bool train, bool bidirectional, bool batch_first) -> (Tensor, Tensor, Tensor, Tensor, Tensor) +- func: _lstm_mps(Tensor input, Tensor[] hx, Tensor[] params, bool has_biases, int num_layers, float dropout, bool train, bool bidirectional, bool batch_first) -> (Tensor, Tensor, Tensor, Tensor, Tensor, Tensor) dispatch: MPS: _lstm_mps autogen: _lstm_mps.out -- func: lstm_mps_backward(Tensor grad_y, Tensor? grad_hy, Tensor? grad_cy, Tensor z_state, Tensor cell_state_fwd, Tensor input, Tensor[] hx, Tensor[] params, bool has_biases, int num_layers, float dropout, bool train, bool bidirectional, bool batch_first) -> (Tensor, Tensor[], Tensor[]) +- func: lstm_mps_backward(Tensor grad_y, Tensor? grad_hy, Tensor? grad_cy, Tensor z_state, Tensor cell_state_fwd, Tensor input, Tensor layersOutputs, Tensor[] hx, Tensor[] params, bool has_biases, int num_layers, float dropout, bool train, bool bidirectional, bool batch_first) -> (Tensor, Tensor[], Tensor[]) dispatch: MPS: lstm_mps_backward autogen: lstm_mps_backward.out @@ -6808,11 +7279,13 @@ - func: _pack_padded_sequence(Tensor input, Tensor lengths, bool batch_first) -> (Tensor, Tensor) dispatch: CompositeExplicitAutograd: _pack_padded_sequence autogen: _pack_padded_sequence.out -- func: _pack_padded_sequence_backward(Tensor grad, int[] input_size, Tensor batch_sizes, bool batch_first) -> Tensor +- func: _pack_padded_sequence_backward(Tensor grad, SymInt[] input_size, Tensor batch_sizes, bool batch_first) -> Tensor + dispatch: + CompositeImplicitAutograd: _pack_padded_sequence_backward_symint - func: _pad_packed_sequence(Tensor data, Tensor batch_sizes, bool batch_first, Scalar padding_value, int total_length) -> (Tensor, Tensor) # wrappers for legacy TH methods @@ -6881,11 +7354,11 @@ # Like lift, but it clones the input. - func: lift_fresh_copy(Tensor self) -> Tensor tags: view_copy dispatch: - CompositeExplicitAutograd: lift_fresh_copy + CompositeExplicitAutogradNonFunctional: lift_fresh_copy autogen: lift_fresh_copy.out - func: is_set_to(Tensor self, Tensor tensor) -> bool variants: method device_check: NoCheck @@ -6907,10 +7380,11 @@ - func: masked_fill.Scalar(Tensor self, Tensor mask, Scalar value) -> Tensor device_check: NoCheck # TensorIterator variants: function, method dispatch: CompositeExplicitAutograd: masked_fill + tags: pointwise - func: masked_fill_.Tensor(Tensor(a!) self, Tensor mask, Tensor value) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: @@ -6957,10 +7431,11 @@ device_guard: False dispatch: ZeroTensor, Meta, CPU, CUDA, QuantizedCPU, QuantizedCUDA, MPS: view MkldnnCPU: mkldnn_view NestedTensorCPU, NestedTensorCUDA: view_nested + tags: core # Warning: If you want to change the name or overload name of this # operator, you might also want to change the `isBlockListedSchema` # function in `torch/csrc/jit/frontend/schema_catching.cpp`. # The name and overload name of this operator is hardcoded in that @@ -6974,11 +7449,11 @@ CompositeExplicitAutograd: view_dtype - func: put_(Tensor(a!) self, Tensor index, Tensor source, bool accumulate=False) -> Tensor(a!) variants: method dispatch: - CPU, CUDA, MPS: put_ + CPU, CUDA: put_ autogen: put.out - func: put(Tensor self, Tensor index, Tensor source, bool accumulate=False) -> Tensor variants: function, method dispatch: @@ -7132,10 +7607,11 @@ variants: function, method - func: scatter_add(Tensor self, int dim, Tensor index, Tensor src) -> Tensor structured_delegate: scatter_add.out variants: function, method + tags: core - func: scatter_add_(Tensor(a!) self, int dim, Tensor index, Tensor src) -> Tensor(a!) structured_delegate: scatter_add.out variants: method @@ -7150,10 +7626,11 @@ variants: function, method - func: scatter_reduce.two(Tensor self, int dim, Tensor index, Tensor src, str reduce, *, bool include_self=True) -> Tensor structured_delegate: scatter_reduce.two_out variants: function, method + tags: core - func: scatter_reduce_.two(Tensor(a!) self, int dim, Tensor index, Tensor src, str reduce, *, bool include_self=True) -> Tensor(a!) structured_delegate: scatter_reduce.two_out variants: method @@ -7178,43 +7655,50 @@ structured: True structured_inherits: TensorIteratorBase variants: function dispatch: CPU, CUDA: bitwise_and_out + tags: pointwise - func: bitwise_and.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: bitwise_and_out + tags: pointwise - func: bitwise_and.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: CompositeExplicitAutograd: bitwise_and + tags: pointwise - func: bitwise_and.Scalar_Tensor(Scalar self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: bitwise_and autogen: bitwise_and.Scalar_Tensor_out + tags: pointwise - func: bitwise_and.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function structured_delegate: bitwise_and.Tensor_out + tags: [core, pointwise] - func: bitwise_and_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method + tags: pointwise - func: bitwise_and_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: bitwise_and.Tensor_out + tags: pointwise - func: __and__.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function @@ -7235,41 +7719,48 @@ structured: True structured_inherits: TensorIteratorBase variants: function dispatch: CPU, CUDA: bitwise_or_out + tags: pointwise - func: bitwise_or.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: bitwise_or_out + tags: pointwise - func: bitwise_or.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function + tags: pointwise - func: bitwise_or.Scalar_Tensor(Scalar self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: bitwise_or autogen: bitwise_or.Scalar_Tensor_out + tags: pointwise - func: bitwise_or.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function structured_delegate: bitwise_or.Tensor_out + tags: [core, pointwise] - func: bitwise_or_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method + tags: pointwise - func: bitwise_or_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: bitwise_or.Tensor_out + tags: pointwise - func: __or__.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function @@ -7290,137 +7781,161 @@ structured: True structured_inherits: TensorIteratorBase variants: function dispatch: CPU, CUDA: bitwise_xor_out + tags: pointwise - func: bitwise_xor.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: bitwise_xor_out + tags: pointwise - func: bitwise_xor.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function + tags: pointwise - func: bitwise_xor.Scalar_Tensor(Scalar self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: bitwise_xor autogen: bitwise_xor.Scalar_Tensor_out + tags: pointwise - func: bitwise_xor.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function structured_delegate: bitwise_xor.Tensor_out + tags: [core, pointwise] - func: bitwise_xor_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method + tags: pointwise - func: bitwise_xor_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: bitwise_xor.Tensor_out + tags: pointwise - func: __xor__.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function + tags: pointwise - func: __xor__.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function + tags: pointwise - func: __ixor__.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method + tags: pointwise - func: __ixor__.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method + tags: pointwise - func: __lshift__.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: CPU, CUDA: __lshift__ + tags: pointwise - func: __lshift__.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: CPU, CUDA: __lshift__ + tags: pointwise - func: __ilshift__.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CPU, CUDA: __ilshift__ autogen: __lshift__.Scalar_out + tags: pointwise - func: __ilshift__.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CPU, CUDA: __ilshift__ autogen: __lshift__.Tensor_out + tags: pointwise - func: bitwise_left_shift.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: bitwise_left_shift.Tensor_out + tags: pointwise - func: bitwise_left_shift_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: bitwise_left_shift.Tensor_out + tags: pointwise - func: bitwise_left_shift.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: bitwise_left_shift_out + tags: pointwise - func: bitwise_left_shift.Tensor_Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: CompositeExplicitAutograd: bitwise_left_shift + tags: pointwise - func: bitwise_left_shift_.Tensor_Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: bitwise_left_shift_ + tags: pointwise - func: bitwise_left_shift.Tensor_Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: bitwise_left_shift_out + tags: pointwise - func: bitwise_left_shift.Scalar_Tensor(Scalar self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: bitwise_left_shift autogen: bitwise_left_shift.Scalar_Tensor_out + tags: pointwise - func: __rshift__.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: CPU, CUDA: __rshift__ + tags: pointwise - func: __rshift__.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: CPU, CUDA: __rshift__ + tags: pointwise - func: __irshift__.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: @@ -7436,47 +7951,54 @@ - func: bitwise_right_shift.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function, method structured_delegate: bitwise_right_shift.Tensor_out + tags: pointwise - func: bitwise_right_shift_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: bitwise_right_shift.Tensor_out + tags: pointwise - func: bitwise_right_shift.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: bitwise_right_shift_out + tags: pointwise - func: bitwise_right_shift.Tensor_Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: CompositeExplicitAutograd: bitwise_right_shift + tags: pointwise - func: bitwise_right_shift_.Tensor_Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: bitwise_right_shift_ + tags: pointwise - func: bitwise_right_shift.Tensor_Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: bitwise_right_shift_out + tags: pointwise - func: bitwise_right_shift.Scalar_Tensor(Scalar self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function dispatch: CompositeExplicitAutograd: bitwise_right_shift autogen: bitwise_right_shift.Scalar_Tensor_out + tags: pointwise - func: tril_(Tensor(a!) self, int diagonal=0) -> Tensor(a!) structured_delegate: tril.out variants: method @@ -7486,20 +8008,23 @@ - func: digamma_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: digamma.out variants: method + tags: pointwise - func: lerp_.Scalar(Tensor(a!) self, Tensor end, Scalar weight) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: lerp.Scalar_out + tags: pointwise - func: lerp_.Tensor(Tensor(a!) self, Tensor end, Tensor weight) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: lerp.Tensor_out + tags: pointwise - func: addbmm_(Tensor(a!) self, Tensor batch1, Tensor batch2, *, Scalar beta=1, Scalar alpha=1) -> Tensor(a!) variants: method dispatch: CPU, CUDA: addbmm_ @@ -7589,27 +8114,14 @@ # wrappers for TH functions autogen: geometric, geometric.out - func: diag.out(Tensor self, int diagonal=0, *, Tensor(a!) out) -> Tensor(a!) - dispatch: - CPU: diag_cpu_out - CUDA: diag_cuda_out - MPS: diag_mps_out - func: diag(Tensor self, int diagonal=0) -> Tensor variants: method, function - dispatch: - CompositeExplicitAutograd: diag -- func: diag_backward(Tensor grad, SymInt[] input_sizes, int diagonal) -> Tensor - variants: function - device_check: NoCheck - device_guard: False - dispatch: - CompositeImplicitAutograd: diag_backward_symint - - func: cross.out(Tensor self, Tensor other, int? dim=None, *, Tensor(a!) out) -> Tensor(a!) - func: cross(Tensor self, Tensor other, int? dim=None) -> Tensor variants: method, function @@ -7650,48 +8162,55 @@ - func: trace(Tensor self) -> Tensor variants: method, function dispatch: CPU: trace_cpu CUDA: trace_cuda + MPS: trace_mps_out autogen: trace.out -- func: trace_backward(Tensor grad, int[] sizes) -> Tensor +- func: trace_backward(Tensor grad, SymInt[] sizes) -> Tensor variants: function device_check: NoCheck device_guard: False + dispatch: + CompositeImplicitAutograd: trace_backward_symint - func: ne.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: ne_Scalar_out MPS: ne_scalar_out_mps QuantizedCPU: ne_out_quantized_cpu + tags: pointwise - func: ne.Scalar(Tensor self, Scalar other) -> Tensor structured_delegate: ne.Scalar_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: ne_quantized_cpu + tags: [core, pointwise] - func: ne.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: ne_Tensor_out MPS: ne_tensor_out_mps QuantizedCPU: ne_out_quantized_cpu + tags: pointwise - func: ne.Tensor(Tensor self, Tensor other) -> Tensor structured_delegate: ne.Tensor_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: ne_quantized_cpu + tags: [core, pointwise] - func: ne_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) structured_delegate: ne.Scalar_out device_check: NoCheck # TensorIterator variants: method @@ -7724,65 +8243,73 @@ device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: eq_Scalar_out MPS: eq_scalar_out_mps QuantizedCPU: eq_out_quantized_cpu + tags: pointwise - func: eq.Scalar(Tensor self, Scalar other) -> Tensor structured_delegate: eq.Scalar_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: eq_quantized_cpu + tags: [core, pointwise] - func: eq.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: eq_Tensor_out MPS: eq_tensor_out_mps QuantizedCPU: eq_out_quantized_cpu + tags: pointwise - func: eq.Tensor(Tensor self, Tensor other) -> Tensor structured_delegate: eq.Tensor_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: eq_quantized_cpu + tags: [core, pointwise] - func: ge.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: ge_Scalar_out MPS: ge_scalar_out_mps QuantizedCPU: ge_out_quantized_cpu + tags: pointwise - func: ge.Scalar(Tensor self, Scalar other) -> Tensor structured_delegate: ge.Scalar_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: ge_quantized_cpu + tags: [core, pointwise] - func: ge.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: ge_Tensor_out MPS: ge_tensor_out_mps QuantizedCPU: ge_out_quantized_cpu + tags: pointwise - func: ge.Tensor(Tensor self, Tensor other) -> Tensor structured_delegate: ge.Tensor_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: ge_quantized_cpu + tags: [core, pointwise] - func: ge_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) structured_delegate: ge.Scalar_out device_check: NoCheck # TensorIterator variants: method @@ -7815,33 +8342,37 @@ device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: le_Scalar_out MPS: le_scalar_out_mps QuantizedCPU: le_out_quantized_cpu + tags: pointwise - func: le.Scalar(Tensor self, Scalar other) -> Tensor structured_delegate: le.Scalar_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: le_quantized_cpu + tags: [core, pointwise] - func: le.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: le_Tensor_out MPS: le_tensor_out_mps QuantizedCPU: le_out_quantized_cpu + tags: pointwise - func: le.Tensor(Tensor self, Tensor other) -> Tensor structured_delegate: le.Tensor_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: le_quantized_cpu + tags: [core, pointwise] - func: le_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) structured_delegate: le.Scalar_out device_check: NoCheck # TensorIterator variants: method @@ -7874,33 +8405,37 @@ device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: gt_Scalar_out MPS: gt_scalar_out_mps QuantizedCPU: gt_out_quantized_cpu + tags: pointwise - func: gt.Scalar(Tensor self, Scalar other) -> Tensor structured_delegate: gt.Scalar_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: gt_quantized_cpu + tags: [core, pointwise] - func: gt.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: gt_Tensor_out MPS: gt_tensor_out_mps QuantizedCPU: gt_out_quantized_cpu + tags: pointwise - func: gt.Tensor(Tensor self, Tensor other) -> Tensor structured_delegate: gt.Tensor_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: gt_quantized_cpu + tags: [core, pointwise] - func: gt_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) structured_delegate: gt.Scalar_out device_check: NoCheck # TensorIterator variants: method @@ -7933,33 +8468,37 @@ device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: lt_Scalar_out MPS: lt_scalar_out_mps QuantizedCPU: lt_out_quantized_cpu + tags: pointwise - func: lt.Scalar(Tensor self, Scalar other) -> Tensor structured_delegate: lt.Scalar_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: lt_quantized_cpu + tags: [core, pointwise] - func: lt.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: lt_Tensor_out MPS: lt_tensor_out_mps QuantizedCPU: lt_out_quantized_cpu + tags: pointwise - func: lt.Tensor(Tensor self, Tensor other) -> Tensor structured_delegate: lt.Tensor_out device_check: NoCheck # TensorIterator variants: method, function dispatch: QuantizedCPU: lt_quantized_cpu + tags: [core, pointwise] - func: lt_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) structured_delegate: lt.Scalar_out device_check: NoCheck # TensorIterator variants: method @@ -8014,20 +8553,23 @@ CUDA: index_select_cuda QuantizedCUDA: index_select_quantized_cuda SparseCPU: index_select_sparse_cpu SparseCUDA: index_select_sparse_cuda MPS: index_select_mps + tags: core - func: index_select.dimname_out(Tensor self, Dimname dim, Tensor index, *, Tensor(a!) out) -> Tensor(a!) - func: index_select.dimname(Tensor self, Dimname dim, Tensor index) -> Tensor variants: method, function -- func: index_select_backward(Tensor grad, int[] self_sizes, int dim, Tensor index) -> Tensor +- func: index_select_backward(Tensor grad, SymInt[] self_sizes, int dim, Tensor index) -> Tensor variants: function device_check: NoCheck device_guard: False + dispatch: + CompositeImplicitAutograd: index_select_backward_symint - func: masked_select.out(Tensor self, Tensor mask, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU: masked_select_out_cpu CUDA: masked_select_out_cuda @@ -8049,18 +8591,20 @@ - func: nonzero.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU: nonzero_out_cpu CUDA: nonzero_out_cuda + MPS: nonzero_out_mps tags: dynamic_output_shape - func: nonzero(Tensor self) -> Tensor variants: method, function dispatch: CPU: nonzero_cpu CUDA: nonzero_cuda - tags: dynamic_output_shape + MPS: nonzero_mps + tags: [dynamic_output_shape, core] - func: nonzero_numpy(Tensor self) -> Tensor[] variants: method, function - func: argwhere(Tensor self) -> Tensor @@ -8074,10 +8618,11 @@ MPS: gather_out_mps - func: gather(Tensor self, int dim, Tensor index, *, bool sparse_grad=False) -> Tensor variants: method, function structured_delegate: gather.out + tags: core - func: gather_backward(Tensor grad, Tensor self, int dim, Tensor index, bool sparse_grad) -> Tensor variants: function device_check: NoCheck device_guard: False @@ -8094,46 +8639,55 @@ structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: addcmul_out MPS: addcmul_out_mps + tags: pointwise - func: addcmul(Tensor self, Tensor tensor1, Tensor tensor2, *, Scalar value=1) -> Tensor structured_delegate: addcmul.out device_check: NoCheck # TensorIterator variants: method, function + tags: pointwise - func: addcmul_(Tensor(a!) self, Tensor tensor1, Tensor tensor2, *, Scalar value=1) -> Tensor(a!) structured_delegate: addcmul.out device_check: NoCheck # TensorIterator variants: method + tags: pointwise - func: addcdiv.out(Tensor self, Tensor tensor1, Tensor tensor2, *, Scalar value=1, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: addcdiv_out MPS: addcdiv_out_mps + tags: pointwise - func: addcdiv(Tensor self, Tensor tensor1, Tensor tensor2, *, Scalar value=1) -> Tensor structured_delegate: addcdiv.out device_check: NoCheck # TensorIterator variants: method, function + tags: pointwise - func: addcdiv_(Tensor(a!) self, Tensor tensor1, Tensor tensor2, *, Scalar value=1) -> Tensor(a!) structured_delegate: addcdiv.out device_check: NoCheck # TensorIterator variants: method + tags: pointwise -- func: cross_entropy_loss(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, int ignore_index=-100, float label_smoothing=0.0) -> Tensor +- func: cross_entropy_loss(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100, float label_smoothing=0.0) -> Tensor python_module: nn + dispatch: + CompositeImplicitAutograd: cross_entropy_loss_symint - func: triangular_solve.X(Tensor self, Tensor A, bool upper=True, bool transpose=False, bool unitriangular=False, *, Tensor(a!) X, Tensor(b!) M) -> (Tensor(a!) solution, Tensor(b!) cloned_coefficient) structured: True dispatch: CPU, CUDA: triangular_solve_out + MPS: triangular_solve_mps_out SparseCsrCPU: triangular_solve_out_sparse_csr_cpu SparseCsrCUDA: triangular_solve_out_sparse_csr_cuda - func: triangular_solve(Tensor self, Tensor A, bool upper=True, bool transpose=False, bool unitriangular=False) -> (Tensor solution, Tensor cloned_coefficient) structured_delegate: triangular_solve.X @@ -8145,36 +8699,22 @@ - func: linalg_solve_triangular.out(Tensor self, Tensor B, *, bool upper, bool left=True, bool unitriangular=False, Tensor(a!) out) -> Tensor(a!) python_module: linalg dispatch: CPU, CUDA: linalg_solve_triangular_out + MPS: linalg_solve_triangular_mps_out - func: linalg_solve_triangular(Tensor self, Tensor B, *, bool upper, bool left=True, bool unitriangular=False) -> Tensor python_module: linalg variants: function dispatch: CPU, CUDA: linalg_solve_triangular + MPS: linalg_solve_triangular_mps - func: linalg_vander(Tensor x, *, int? N=None) -> Tensor python_module: linalg -- func: symeig.e(Tensor self, bool eigenvectors=False, bool upper=True, *, Tensor(a!) e, Tensor(b!) V) -> (Tensor(a!) eigenvalues, Tensor(b!) eigenvectors) - dispatch: - CompositeExplicitAutograd: symeig_out - -- func: symeig(Tensor self, bool eigenvectors=False, bool upper=True) -> (Tensor eigenvalues, Tensor eigenvectors) - variants: method, function - dispatch: - CompositeExplicitAutograd: symeig - -- func: _symeig_helper(Tensor self, bool eigenvectors, bool upper) -> (Tensor, Tensor) - variants: function - dispatch: - CPU: _symeig_helper_cpu - CUDA: _symeig_helper_cuda - autogen: _symeig_helper.out - - func: svd.U(Tensor self, bool some=True, bool compute_uv=True, *, Tensor(a!) U, Tensor(b!) S, Tensor(c!) V) -> (Tensor(a!) U, Tensor(b!) S, Tensor(c!) V) - func: svd(Tensor self, bool some=True, bool compute_uv=True) -> (Tensor U, Tensor S, Tensor V) variants: method, function @@ -8302,131 +8842,151 @@ device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: lgamma_out + tags: pointwise - func: lgamma_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: lgamma.out variants: method + tags: pointwise - func: lgamma(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: lgamma.out variants: method, function + tags: pointwise - func: digamma.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: digamma_out + tags: pointwise - func: digamma(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: digamma.out variants: method, function + tags: pointwise - func: polygamma.out(int n, Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: polygamma_out + tags: pointwise - func: polygamma(int n, Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: polygamma.out variants: method, function + tags: pointwise - func: polygamma_(Tensor(a!) self, int n) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: polygamma_ + tags: pointwise - func: erfinv(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: erfinv.out variants: method, function dispatch: SparseCPU, SparseCUDA: erfinv_sparse SparseCsrCPU, SparseCsrCUDA: erfinv_sparse_csr + tags: pointwise - func: erfinv_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: erfinv.out variants: method dispatch: SparseCPU, SparseCUDA: erfinv_sparse_ SparseCsrCPU, SparseCsrCUDA: erfinv_sparse_csr_ + tags: pointwise - func: erfinv.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: erfinv_out SparseCPU, SparseCUDA: erfinv_sparse_out SparseCsrCPU, SparseCsrCUDA: erfinv_sparse_csr_out + tags: pointwise - func: i0(Tensor self) -> Tensor structured_delegate: i0.out variants: function, method + tags: pointwise - func: i0_(Tensor(a!) self) -> Tensor(a!) structured_delegate: i0.out variants: function, method + tags: pointwise - func: i0.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: i0_out + tags: pointwise - func: sign(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: sign.out variants: function, method dispatch: SparseCPU, SparseCUDA: sign_sparse SparseCsrCPU, SparseCsrCUDA: sign_sparse_csr + tags: [core, pointwise] - func: sign_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: sign.out variants: method dispatch: SparseCPU, SparseCUDA: sign_sparse_ SparseCsrCPU, SparseCsrCUDA: sign_sparse_csr_ + tags: pointwise - func: sign.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: sign_out MPS: sign_out_mps SparseCPU, SparseCUDA: sign_sparse_out SparseCsrCPU, SparseCsrCUDA: sign_sparse_csr_out + tags: pointwise - func: signbit(Tensor self) -> Tensor variants: function, method structured_delegate: signbit.out dispatch: SparseCPU, SparseCUDA: signbit_sparse SparseCsrCPU, SparseCsrCUDA: signbit_sparse_csr + tags: pointwise - func: signbit.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU: signbit_out CUDA: signbit_out + MPS: signbit_out_mps SparseCPU, SparseCUDA: signbit_sparse_out SparseCsrCPU, SparseCsrCUDA: signbit_sparse_csr_out + tags: pointwise - func: dist(Tensor self, Tensor other, Scalar p=2) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: @@ -8438,22 +8998,25 @@ structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: atan2_out MPS: atan2_mps_out + tags: pointwise - func: atan2_(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: atan2.out variants: method + tags: pointwise - func: atan2(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: atan2.out variants: method, function - + tags: pointwise # arctan2, alias of atan2 + - func: arctan2(Tensor self, Tensor other) -> Tensor variants: method, function - func: arctan2.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator @@ -8465,27 +9028,31 @@ device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: lerp_Scalar + tags: pointwise - func: lerp.Tensor_out(Tensor self, Tensor end, Tensor weight, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: lerp_Tensor + tags: pointwise - func: lerp.Scalar(Tensor self, Tensor end, Scalar weight) -> Tensor device_check: NoCheck # TensorIterator variants: method, function structured_delegate: lerp.Scalar_out + tags: pointwise - func: lerp.Tensor(Tensor self, Tensor end, Tensor weight) -> Tensor device_check: NoCheck # TensorIterator variants: method, function structured_delegate: lerp.Tensor_out + tags: pointwise - func: histc.out(Tensor self, int bins=100, Scalar min=0, Scalar max=0, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU: histogram_histc_cpu_out CUDA: _histc_out_cuda @@ -8537,134 +9104,160 @@ - func: fmod.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator dispatch: CompositeExplicitAutograd: fmod_out + tags: pointwise - func: fmod.Scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: CompositeExplicitAutograd: fmod + tags: pointwise - func: fmod_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method dispatch: CompositeExplicitAutograd: fmod_ + tags: pointwise - func: fmod.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: fmod_out + MPS: fmod_mps_out + tags: pointwise - func: fmod.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: fmod.Tensor_out variants: method, function + tags: [core, pointwise] - - func: fmod_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator variants: method structured_delegate: fmod.Tensor_out + tags: pointwise - func: hypot.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: hypot_out + tags: pointwise - func: hypot(Tensor self, Tensor other) -> Tensor structured_delegate: hypot.out variants: method, function + tags: pointwise - func: hypot_(Tensor(a!) self, Tensor other) -> Tensor(a!) structured_delegate: hypot.out variants: method + tags: pointwise - func: igamma.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: igamma_out + tags: pointwise - func: igamma(Tensor self, Tensor other) -> Tensor structured_delegate: igamma.out variants: method, function + tags: pointwise - func: igamma_(Tensor(a!) self, Tensor other) -> Tensor(a!) structured_delegate: igamma.out variants: method + tags: pointwise - func: igammac.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: igammac_out + tags: pointwise - func: igammac(Tensor self, Tensor other) -> Tensor structured_delegate: igammac.out variants: method, function + tags: pointwise - func: igammac_(Tensor(a!) self, Tensor other) -> Tensor(a!) structured_delegate: igammac.out variants: method + tags: pointwise - func: nextafter.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: nextafter_out + tags: pointwise - func: nextafter(Tensor self, Tensor other) -> Tensor structured_delegate: nextafter.out variants: method, function + tags: pointwise - func: nextafter_(Tensor(a!) self, Tensor other) -> Tensor(a!) structured_delegate: nextafter.out variants: method + tags: pointwise - func: remainder.Scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: remainder_out + tags: pointwise - func: remainder.Scalar(Tensor self, Scalar other) -> Tensor variants: method, function dispatch: CompositeExplicitAutograd: remainder + tags: pointwise - func: remainder_.Scalar(Tensor(a!) self, Scalar other) -> Tensor(a!) variants: method dispatch: CompositeExplicitAutograd: remainder_ + tags: pointwise - func: remainder.Tensor_out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: remainder_out + MPS: remainder_out_mps + tags: pointwise - func: remainder.Tensor(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: remainder.Tensor_out variants: method, function + tags: [core, pointwise] - func: remainder_.Tensor(Tensor(a!) self, Tensor other) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: remainder.Tensor_out variants: method + tags: pointwise - func: remainder.Scalar_Tensor(Scalar self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: function dispatch: - CPU, CUDA: remainder + CPU, CUDA, MPS: remainder autogen: remainder.Scalar_Tensor_out + tags: pointwise - func: min(Tensor self) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: @@ -8681,88 +9274,99 @@ - func: fmin(Tensor self, Tensor other) -> Tensor structured_delegate: fmin.out device_check: NoCheck # TensorIterator variants: method, function + tags: pointwise - func: fmin.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: fmin_out + tags: pointwise - func: max(Tensor self) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: CPU, CUDA: max MPS: max_mps QuantizedCPU: max_quantized_cpu -# Not to be confused with binary op `max.out`. Commented because of failed CI -# FIXME: enable this -#- func: max.unary_out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) -# device_check: NoCheck # TensorIterator -# dispatch: -# CompositeExplicitAutograd: max_unary_out - - func: fmax(Tensor self, Tensor other) -> Tensor structured_delegate: fmax.out device_check: NoCheck # TensorIterator variants: method, function + tags: pointwise - func: fmax.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: fmax_out + tags: pointwise - func: maximum(Tensor self, Tensor other) -> Tensor structured_delegate: maximum.out device_check: NoCheck # TensorIterator variants: method, function + tags: [core, pointwise] - func: maximum.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: maximum_out MPS: maximum_out_mps + tags: pointwise # binary max, alias of maximum # NOTE: max is not an alias for maximum, since there is also unary max - func: max.other(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function + tags: pointwise - func: max.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator + tags: pointwise +- func: max.unary_out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) + device_check: NoCheck # TensorIterator + dispatch: + CPU, CUDA: max_unary_out + QuantizedCPU: max_quantized_unary_out + - func: minimum(Tensor self, Tensor other) -> Tensor structured_delegate: minimum.out device_check: NoCheck # TensorIterator variants: method, function + tags: [core, pointwise] - func: minimum.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase device_check: NoCheck # TensorIterator dispatch: CPU, CUDA: minimum_out MPS: minimum_out_mps + tags: pointwise # binary min, alias for minimum # NOTE: min is not an alias for minimum, since there is also unary min - func: min.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator + tags: pointwise - func: min.other(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator variants: method, function + tags: pointwise - func: quantile(Tensor self, Tensor q, int? dim=None, bool keepdim=False, *, str interpolation='linear') -> Tensor variants: method, function - func: quantile.out(Tensor self, Tensor q, int? dim=None, bool keepdim=False, *, str interpolation='linear', Tensor(a!) out) -> Tensor(a!) @@ -8789,10 +9393,11 @@ - func: sort.values_stable(Tensor self, *, bool? stable, int dim=-1, bool descending=False, Tensor(a!) values, Tensor(b!) indices) -> (Tensor(a!) values, Tensor(b!) indices) structured: True dispatch: CPU, CUDA: sort_stable_out + MPS: sort_stable_out_mps - func: sort(Tensor self, int dim=-1, bool descending=False) -> (Tensor values, Tensor indices) device_check: NoCheck # TensorIterator variants: method, function dispatch: @@ -8825,11 +9430,11 @@ - func: argsort.stable(Tensor self, *, bool stable, int dim=-1, bool descending=False) -> Tensor device_check: NoCheck # TensorIterator variants: method, function dispatch: - CPU, CUDA: argsort_stable + CPU, CUDA, MPS: argsort_stable autogen: argsort.stable_out - func: argsort.dimname(Tensor self, Dimname dim, bool descending=False) -> Tensor variants: method, function @@ -8843,10 +9448,11 @@ - func: topk(Tensor self, int k, int dim=-1, bool largest=True, bool sorted=True) -> (Tensor values, Tensor indices) variants: method, function structured_delegate: topk.values dispatch: QuantizedCPU: topk_quantized_cpu + tags: core - func: all(Tensor self) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: all.all_out variants: method, function @@ -8891,21 +9497,21 @@ - func: unfold(Tensor(a) self, int dimension, int size, int step) -> Tensor(a) variants: method device_check: NoCheck device_guard: False dispatch: - CPU, CUDA, Meta: unfold + CPU, CUDA, Meta, MPS: unfold QuantizedCPU, QuantizedCUDA: unfold -- func: unfold_backward(Tensor grad_in, int[] input_sizes, int dim, int size, int step) -> Tensor +- func: unfold_backward(Tensor grad_in, SymInt[] input_sizes, int dim, int size, int step) -> Tensor variants: function dispatch: CPU, CUDA: unfold_backward autogen: unfold_backward.out - func: equal(Tensor self, Tensor other) -> bool - tags: data_dependent_output + tags: [data_dependent_output, pointwise] variants: method, function dispatch: CPU: cpu_equal CUDA: cuda_equal MPS: mps_equal @@ -8916,71 +9522,87 @@ structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: pow_Tensor_Tensor_out MPS: pow_tensor_tensor_out_mps + tags: pointwise - func: pow.Tensor_Tensor(Tensor self, Tensor exponent) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: pow.Tensor_Tensor_out variants: method, function + tags: [core, pointwise] - func: pow.Scalar_out(Scalar self, Tensor exponent, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True dispatch: CPU, CUDA: pow_Scalar_out + tags: pointwise - func: pow.Scalar(Scalar self, Tensor exponent) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: pow.Scalar_out + tags: pointwise - func: pow.Tensor_Scalar_out(Tensor self, Scalar exponent, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: pow_Tensor_Scalar_out SparseCPU, SparseCUDA: pow_out_sparse_scalar MPS: pow_tensor_scalar_out_mps + tags: pointwise - func: pow.Tensor_Scalar(Tensor self, Scalar exponent) -> Tensor device_check: NoCheck # TensorIterator structured_delegate: pow.Tensor_Scalar_out variants: function, method dispatch: SparseCPU, SparseCUDA: pow_sparse_scalar + tags: [core, pointwise] - func: pow_.Scalar(Tensor(a!) self, Scalar exponent) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: pow.Tensor_Scalar_out variants: method + tags: pointwise - func: pow_.Tensor(Tensor(a!) self, Tensor exponent) -> Tensor(a!) device_check: NoCheck # TensorIterator structured_delegate: pow.Tensor_Tensor_out variants: method + tags: pointwise - func: float_power.Tensor_Tensor_out(Tensor self, Tensor exponent, *, Tensor(a!) out) -> Tensor(a!) + tags: pointwise - func: float_power.Tensor_Tensor(Tensor self, Tensor exponent) -> Tensor variants: function, method + tags: pointwise - func: float_power.Scalar_out(Scalar self, Tensor exponent, *, Tensor(a!) out) -> Tensor(a!) + tags: pointwise - func: float_power.Scalar(Scalar self, Tensor exponent) -> Tensor + tags: pointwise - func: float_power.Tensor_Scalar_out(Tensor self, Scalar exponent, *, Tensor(a!) out) -> Tensor(a!) + tags: pointwise - func: float_power.Tensor_Scalar(Tensor self, Scalar exponent) -> Tensor variants: function, method + tags: pointwise - func: float_power_.Scalar(Tensor(a!) self, Scalar exponent) -> Tensor(a!) variants: method + tags: pointwise - func: float_power_.Tensor(Tensor(a!) self, Tensor exponent) -> Tensor(a!) variants: method + tags: pointwise - func: normal_(Tensor(a!) self, float mean=0, float std=1, *, Generator? generator=None) -> Tensor(a!) device_check: NoCheck # TensorIterator tags: nondeterministic_seeded variants: method @@ -9040,24 +9662,25 @@ CPU, CUDA: normal MPS: normal_mps Meta: normal_meta tags: nondeterministic_seeded -- func: normal.float_float(float mean, float std, int[] size, *, Generator? generator=None, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor +- func: normal.float_float(float mean, float std, SymInt[] size, *, Generator? generator=None, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None) -> Tensor dispatch: CompositeExplicitAutograd: normal tags: nondeterministic_seeded -- func: normal.float_float_out(float mean, float std, int[] size, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!) +- func: normal.float_float_out(float mean, float std, SymInt[] size, *, Generator? generator=None, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: normal_out tags: nondeterministic_seeded - func: alias(Tensor(a) self) -> Tensor(a) variants: method, function dispatch: CompositeExplicitAutograd: alias + tags: core - func: _amp_foreach_non_finite_check_and_unscale_(Tensor(a!)[] self, Tensor(b!) found_inf, Tensor inv_scale) -> () variants: function dispatch: CUDA: _amp_foreach_non_finite_check_and_unscale_cuda_ @@ -9140,10 +9763,71 @@ dispatch: CPU: foreach_tensor_div_scalar_kernel_slow_ CUDA: foreach_tensor_div_scalar_kernel_cuda_ autogen: _foreach_div.Scalar_out +- func: _foreach_clamp_min.Scalar(Tensor[] self, Scalar scalar) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_scalar_kernel_slow + CUDA: foreach_tensor_clamp_min_scalar_kernel_cuda + +- func: _foreach_clamp_min_.Scalar(Tensor(a!)[] self, Scalar scalar) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_scalar_kernel_slow_ + CUDA: foreach_tensor_clamp_min_scalar_kernel_cuda_ + autogen: _foreach_clamp_min.Scalar_out + +- func: _foreach_clamp_max.Scalar(Tensor[] self, Scalar scalar) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_scalar_kernel_slow + CUDA: foreach_tensor_clamp_max_scalar_kernel_cuda + +- func: _foreach_clamp_max_.Scalar(Tensor(a!)[] self, Scalar scalar) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_scalar_kernel_slow_ + CUDA: foreach_tensor_clamp_max_scalar_kernel_cuda_ + autogen: _foreach_clamp_max.Scalar_out + +# foreach_minimum/maximum dispatches to clamp_max/min +- func: _foreach_maximum.Scalar(Tensor[] self, Scalar scalar) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_scalar_kernel_slow + CUDA: foreach_tensor_clamp_min_scalar_kernel_cuda + +- func: _foreach_maximum_.Scalar(Tensor(a!)[] self, Scalar scalar) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_scalar_kernel_slow_ + CUDA: foreach_tensor_clamp_min_scalar_kernel_cuda_ + autogen: _foreach_maximum.Scalar_out + +- func: _foreach_minimum.Scalar(Tensor[] self, Scalar scalar) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_scalar_kernel_slow + CUDA: foreach_tensor_clamp_max_scalar_kernel_cuda + +- func: _foreach_minimum_.Scalar(Tensor(a!)[] self, Scalar scalar) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_scalar_kernel_slow_ + CUDA: foreach_tensor_clamp_max_scalar_kernel_cuda_ + autogen: _foreach_minimum.Scalar_out + - func: _foreach_add.List(Tensor[] self, Tensor[] other, *, Scalar alpha=1) -> Tensor[] device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices variants: function dispatch: CPU: foreach_tensor_add_list_kernel_slow @@ -9200,10 +9884,72 @@ dispatch: CPU: foreach_tensor_div_list_kernel_slow_ CUDA: foreach_tensor_div_list_kernel_cuda_ autogen: _foreach_div.List_out +- func: _foreach_clamp_min.List(Tensor[] self, Tensor[] other) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_list_kernel_slow + CUDA: foreach_tensor_clamp_min_list_kernel_cuda + +- func: _foreach_clamp_min_.List(Tensor(a!)[] self, Tensor[] other) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_list_kernel_slow_ + CUDA: foreach_tensor_clamp_min_list_kernel_cuda_ + autogen: _foreach_clamp_min.List_out + +- func: _foreach_clamp_max.List(Tensor[] self, Tensor[] other) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_list_kernel_slow + CUDA: foreach_tensor_clamp_max_list_kernel_cuda + +- func: _foreach_clamp_max_.List(Tensor(a!)[] self, Tensor[] other) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_list_kernel_slow_ + CUDA: foreach_tensor_clamp_max_list_kernel_cuda_ + autogen: _foreach_clamp_max.List_out + +# foreach_minimum/maximum dispatches to clamp_max/min +- func: _foreach_maximum.List(Tensor[] self, Tensor[] other) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_list_kernel_slow + CUDA: foreach_tensor_clamp_min_list_kernel_cuda + +- func: _foreach_maximum_.List(Tensor(a!)[] self, Tensor[] other) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_list_kernel_slow_ + CUDA: foreach_tensor_clamp_min_list_kernel_cuda_ + autogen: _foreach_maximum.List_out + +- func: _foreach_minimum.List(Tensor[] self, Tensor[] other) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_list_kernel_slow + CUDA: foreach_tensor_clamp_max_list_kernel_cuda + +- func: _foreach_minimum_.List(Tensor(a!)[] self, Tensor[] other) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_list_kernel_slow_ + CUDA: foreach_tensor_clamp_max_list_kernel_cuda_ + autogen: _foreach_minimum.List_out + + - func: _foreach_add.ScalarList(Tensor[] self, Scalar[] scalars) -> Tensor[] device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices variants: function dispatch: CPU: foreach_tensor_add_scalarlist_kernel_slow @@ -9260,10 +10006,71 @@ dispatch: CPU: foreach_tensor_mul_scalarlist_kernel_slow_ CUDA: foreach_tensor_mul_scalarlist_kernel_cuda_ autogen: _foreach_mul.ScalarList_out +- func: _foreach_clamp_min.ScalarList(Tensor[] self, Scalar[] scalars) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_scalarlist_kernel_slow + CUDA: foreach_tensor_clamp_min_scalarlist_kernel_cuda + +- func: _foreach_clamp_min_.ScalarList(Tensor(a!)[] self, Scalar[] scalars) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_scalarlist_kernel_slow_ + CUDA: foreach_tensor_clamp_min_scalarlist_kernel_cuda_ + autogen: _foreach_clamp_min.ScalarList_out + +- func: _foreach_clamp_max.ScalarList(Tensor[] self, Scalar[] scalars) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_scalarlist_kernel_slow + CUDA: foreach_tensor_clamp_max_scalarlist_kernel_cuda + +- func: _foreach_clamp_max_.ScalarList(Tensor(a!)[] self, Scalar[] scalars) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_scalarlist_kernel_slow_ + CUDA: foreach_tensor_clamp_max_scalarlist_kernel_cuda_ + autogen: _foreach_clamp_max.ScalarList_out + +# foreach_minimum/maximum dispatches to clamp_max/min +- func: _foreach_maximum.ScalarList(Tensor[] self, Scalar[] scalars) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_scalarlist_kernel_slow + CUDA: foreach_tensor_clamp_min_scalarlist_kernel_cuda + +- func: _foreach_maximum_.ScalarList(Tensor(a!)[] self, Scalar[] scalars) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_min_scalarlist_kernel_slow_ + CUDA: foreach_tensor_clamp_min_scalarlist_kernel_cuda_ + autogen: _foreach_maximum.ScalarList_out + +- func: _foreach_minimum.ScalarList(Tensor[] self, Scalar[] scalars) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_scalarlist_kernel_slow + CUDA: foreach_tensor_clamp_max_scalarlist_kernel_cuda + +- func: _foreach_minimum_.ScalarList(Tensor(a!)[] self, Scalar[] scalars) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_clamp_max_scalarlist_kernel_slow_ + CUDA: foreach_tensor_clamp_max_scalarlist_kernel_cuda_ + autogen: _foreach_minimum.ScalarList_out + - func: _foreach_exp(Tensor[] self) -> Tensor[] device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices variants: function dispatch: CPU: foreach_tensor_exp_slow @@ -9712,18 +10519,34 @@ dispatch: CPU: foreach_tensor_addcdiv_scalarlist_slow_ CUDA: foreach_tensor_addcdiv_scalarlist_cuda_ autogen: _foreach_addcdiv.ScalarList_out +- func: _foreach_addcdiv_.Tensor(Tensor(a!)[] self, Tensor[] tensor1, Tensor[] tensor2, Tensor scalars) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_addcdiv_tensor_slow_ + CUDA: foreach_tensor_addcdiv_tensor_cuda_ + autogen: _foreach_addcdiv.Tensor_out + - func: _foreach_addcmul_.ScalarList(Tensor(a!)[] self, Tensor[] tensor1, Tensor[] tensor2, Scalar[] scalars) -> () device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices variants: function dispatch: CPU: foreach_tensor_addcmul_scalarlist_slow_ CUDA: foreach_tensor_addcmul_scalarlist_cuda_ autogen: _foreach_addcmul.ScalarList_out +- func: _foreach_addcmul_.Tensor(Tensor(a!)[] self, Tensor[] tensor1, Tensor[] tensor2, Tensor scalars) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_addcmul_tensor_slow_ + CUDA: foreach_tensor_addcmul_tensor_cuda_ + autogen: _foreach_addcmul.Tensor_out + - func: _foreach_addcdiv.Scalar(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Scalar value=1) -> Tensor[] device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices variants: function dispatch: CPU: foreach_tensor_addcdiv_scalar_slow @@ -9741,55 +10564,71 @@ variants: function dispatch: CPU: foreach_tensor_addcdiv_scalarlist_slow CUDA: foreach_tensor_addcdiv_scalarlist_cuda +- func: _foreach_addcdiv.Tensor(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Tensor scalars) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices + variants: function + dispatch: + CPU: foreach_tensor_addcdiv_tensor_slow + CUDA: foreach_tensor_addcdiv_tensor_cuda + - func: _foreach_addcmul.ScalarList(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Scalar[] scalars) -> Tensor[] device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices variants: function dispatch: CPU: foreach_tensor_addcmul_scalarlist_slow CUDA: foreach_tensor_addcmul_scalarlist_cuda -- func: _foreach_maximum.List(Tensor[] self, Tensor[] other) -> Tensor[] +- func: _foreach_addcmul.Tensor(Tensor[] self, Tensor[] tensor1, Tensor[] tensor2, Tensor scalars) -> Tensor[] device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices variants: function dispatch: - CPU: foreach_tensor_maximum_slow - CUDA: foreach_tensor_maximum_cuda + CPU: foreach_tensor_addcmul_tensor_slow + CUDA: foreach_tensor_addcmul_tensor_cuda -- func: _foreach_maximum_.List(Tensor(a!)[] self, Tensor[] other) -> () +- func: _foreach_norm.Scalar(Tensor[] self, Scalar ord=2) -> Tensor[] device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices variants: function dispatch: - CPU: foreach_tensor_maximum_slow_ - CUDA: foreach_tensor_maximum_cuda_ - autogen: _foreach_maximum.List_out + CPU: foreach_tensor_norm_slow + CUDA: foreach_tensor_norm_cuda + autogen: _foreach_norm.Scalar_out -- func: _foreach_minimum.List(Tensor[] self, Tensor[] other) -> Tensor[] - device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices +- func: _foreach_lerp.List(Tensor[] self, Tensor[] tensors1, Tensor[] weights) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensors are on different devices variants: function dispatch: - CPU: foreach_tensor_minimum_slow - CUDA: foreach_tensor_minimum_cuda + CPU: foreach_tensor_ternary_lerp_slow + CUDA: foreach_tensor_lerp_ternary_cuda + autogen: _foreach_lerp.List_out -- func: _foreach_minimum_.List(Tensor(a!)[] self, Tensor[] other) -> () - device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices +- func: _foreach_lerp_.List(Tensor(a!)[] self, Tensor[] tensors1, Tensor[] weights) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensors are on different devices variants: function dispatch: - CPU: foreach_tensor_minimum_slow_ - CUDA: foreach_tensor_minimum_cuda_ - autogen: _foreach_minimum.List_out + CPU: foreach_tensor_ternary_lerp_slow_ + CUDA: foreach_tensor_lerp_ternary_cuda_ + autogen: _foreach_lerp.List_out -- func: _foreach_norm.Scalar(Tensor[] self, Scalar ord=2) -> Tensor[] - device_check: NoCheck # foreach kernels fall back to slow path when tensor are on different devices +- func: _foreach_lerp.Scalar(Tensor[] self, Tensor[] tensors1, Scalar weight) -> Tensor[] + device_check: NoCheck # foreach kernels fall back to slow path when tensors are on different devices variants: function dispatch: - CPU: foreach_tensor_norm_slow - CUDA: foreach_tensor_norm_cuda - autogen: _foreach_norm.Scalar_out + CPU: foreach_tensor_lerp_list_kernel_slow + CUDA: foreach_tensor_lerp_list_cuda + autogen: _foreach_lerp.Scalar_out +- func: _foreach_lerp_.Scalar(Tensor(a!)[] self, Tensor[] tensors1, Scalar weight) -> () + device_check: NoCheck # foreach kernels fall back to slow path when tensors are on different devices + variants: function + dispatch: + CPU: foreach_tensor_lerp_list_kernel_slow_ + CUDA: foreach_tensor_lerp_list_cuda_ + autogen: _foreach_lerp.Scalar_out + - func: bucketize.Tensor(Tensor self, Tensor boundaries, *, bool out_int32=False, bool right=False) -> Tensor dispatch: CPU: bucketize_cpu CUDA: bucketize_cuda @@ -9807,21 +10646,10 @@ - func: searchsorted.Tensor(Tensor sorted_sequence, Tensor self, *, bool out_int32=False, bool right=False, str? side=None, Tensor? sorter=None) -> Tensor dispatch: CPU: searchsorted_cpu CUDA: searchsorted_cuda -# [Note about _torch_cuda_cu_linker_symbol_op and torch_cuda_cu] -# This is a DUMMY function to force the linking against torch_cuda_cu on Windows. -# Otherwise, the Windows linker will optimize and not include torch_cuda_cu even when we -# want it to be included. This is similar to what we do with warp_size for torch_cuda_cpp, -# described as the solution to this issue: https://github.com/pytorch/pytorch/issues/31611 -# This op should NOT be used or exposed or edited or else Windows builds (with BUILD_SPLIT_CUDA) will break. -- func: _torch_cuda_cu_linker_symbol_op(Tensor self) -> Tensor - dispatch: - CUDA: _torch_cuda_cu_linker_symbol_op_cuda - autogen: _torch_cuda_cu_linker_symbol_op.out - - func: searchsorted.Tensor_out(Tensor sorted_sequence, Tensor self, *, bool out_int32=False, bool right=False, str? side=None, Tensor? sorter=None, Tensor(a!) out) -> Tensor(a!) dispatch: CPU: searchsorted_out_cpu CUDA: searchsorted_out_cuda @@ -9932,71 +10760,77 @@ python_module: nn dispatch: CPU: multilabel_margin_loss_backward_cpu CUDA: multilabel_margin_loss_backward_cuda -- func: nll_loss.out(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, int ignore_index=-100, *, Tensor(a!) out) -> Tensor(a!) +- func: nll_loss.out(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100, *, Tensor(a!) out) -> Tensor(a!) python_module: nn -- func: nll_loss_nd(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, int ignore_index=-100) -> Tensor +- func: nll_loss_nd(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100) -> Tensor python_module: nn + dispatch: + CompositeImplicitAutograd: nll_loss_nd_symint -- func: nll_loss(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, int ignore_index=-100) -> Tensor +- func: nll_loss(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100) -> Tensor python_module: nn + dispatch: + CompositeImplicitAutograd: nll_loss_symint -- func: nll_loss_forward.output(Tensor self, Tensor target, Tensor? weight, int reduction, int ignore_index, *, Tensor(a!) output, Tensor(b!) total_weight) -> (Tensor(a!), Tensor(b!)) +- func: nll_loss_forward.output(Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, *, Tensor(a!) output, Tensor(b!) total_weight) -> (Tensor(a!), Tensor(b!)) python_module: nn structured: True dispatch: CPU: nll_loss_forward_out_cpu CUDA: nll_loss_forward_out_cuda MPS: nll_loss_forward_out_mps -- func: nll_loss_forward(Tensor self, Tensor target, Tensor? weight, int reduction, int ignore_index) -> (Tensor output, Tensor total_weight) +- func: nll_loss_forward(Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index) -> (Tensor output, Tensor total_weight) python_module: nn structured_delegate: nll_loss_forward.output -- func: nll_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, int ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!) +- func: nll_loss_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn structured: True dispatch: CPU: nll_loss_backward_out_cpu CUDA: nll_loss_backward_out_cuda MPS: nll_loss_backward_out_mps -- func: nll_loss_backward(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, int ignore_index, Tensor total_weight) -> Tensor +- func: nll_loss_backward(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight) -> Tensor python_module: nn structured_delegate: nll_loss_backward.grad_input -- func: nll_loss2d.out(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, int ignore_index=-100, *, Tensor(a!) out) -> Tensor(a!) +- func: nll_loss2d.out(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100, *, Tensor(a!) out) -> Tensor(a!) python_module: nn -- func: nll_loss2d(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, int ignore_index=-100) -> Tensor +- func: nll_loss2d(Tensor self, Tensor target, Tensor? weight=None, int reduction=Mean, SymInt ignore_index=-100) -> Tensor python_module: nn + dispatch: + CompositeImplicitAutograd: nll_loss2d_symint -- func: nll_loss2d_forward.output(Tensor self, Tensor target, Tensor? weight, int reduction, int ignore_index, *, Tensor(a!) output, Tensor(b!) total_weight) -> (Tensor(a!), Tensor(b!)) +- func: nll_loss2d_forward.output(Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, *, Tensor(a!) output, Tensor(b!) total_weight) -> (Tensor(a!), Tensor(b!)) python_module: nn dispatch: CPU: nll_loss2d_forward_out_cpu CUDA: nll_loss2d_forward_out_cuda MPS: nll_loss2d_forward_out_mps -- func: nll_loss2d_forward(Tensor self, Tensor target, Tensor? weight, int reduction, int ignore_index) -> (Tensor output, Tensor total_weight) +- func: nll_loss2d_forward(Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index) -> (Tensor output, Tensor total_weight) python_module: nn dispatch: CPU: nll_loss2d_forward_cpu CUDA: nll_loss2d_forward_cuda MPS: nll_loss2d_forward_mps -- func: nll_loss2d_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, int ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!) +- func: nll_loss2d_backward.grad_input(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn dispatch: CPU: nll_loss2d_backward_out_cpu CUDA: nll_loss2d_backward_out_cuda MPS: nll_loss2d_backward_out_mps -- func: nll_loss2d_backward(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, int ignore_index, Tensor total_weight) -> Tensor +- func: nll_loss2d_backward(Tensor grad_output, Tensor self, Tensor target, Tensor? weight, int reduction, SymInt ignore_index, Tensor total_weight) -> Tensor python_module: nn dispatch: CPU: nll_loss2d_backward_cpu CUDA: nll_loss2d_backward_cuda MPS: nll_loss2d_backward_mps @@ -10183,10 +11017,11 @@ device_check: NoCheck # TensorIterator python_module: nn dispatch: CPU, CUDA, MPS: hardtanh QuantizedCPU: hardtanh_quantized_cpu + tags: core - func: hardtanh_backward.grad_input(Tensor grad_output, Tensor self, Scalar min_val, Scalar max_val, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn dispatch: CPU, CUDA: hardtanh_backward_out @@ -10208,27 +11043,31 @@ - func: hardswish.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator python_module: nn dispatch: CPU, CUDA: hardswish_out + MPS: hardswish_out_mps - func: hardswish(Tensor self) -> Tensor device_check: NoCheck # TensorIterator python_module: nn dispatch: CPU, CUDA: hardswish + MPS: hardswish_mps - func: hardswish_(Tensor(a!) self) -> Tensor(a!) device_check: NoCheck # TensorIterator python_module: nn dispatch: CPU, CUDA: hardswish_ + MPS: hardswish_mps_ - func: hardswish_backward(Tensor grad_output, Tensor self) -> Tensor python_module: nn dispatch: CPU, CUDA: hardswish_backward + MPS: hardswish_backward_mps autogen: hardswish_backward.out - func: leaky_relu.out(Tensor self, Scalar negative_slope=0.01, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase @@ -10243,10 +11082,11 @@ structured_delegate: leaky_relu.out device_check: NoCheck # TensorIterator python_module: nn dispatch: QuantizedCPU: leaky_relu_quantized_cpu + tags: core - func: leaky_relu_backward.grad_input(Tensor grad_output, Tensor self, Scalar negative_slope, bool self_is_result, *, Tensor(a!) grad_input) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase python_module: nn @@ -10408,30 +11248,34 @@ CUDA: adaptive_avg_pool2d_cuda MPS: adaptive_avg_pool2d_mps QuantizedCPU: adaptive_avg_pool2d_quantized_cpu QuantizedCUDA: adaptive_avg_pool2d_quantized_cuda autogen: _adaptive_avg_pool2d.out + tags: core - func: _adaptive_avg_pool2d_backward(Tensor grad_output, Tensor self) -> Tensor python_module: nn dispatch: CPU: adaptive_avg_pool2d_backward_cpu CUDA: adaptive_avg_pool2d_backward_cuda MPS: adaptive_avg_pool2d_backward_mps autogen: _adaptive_avg_pool2d_backward.out + tags: core -- func: adaptive_avg_pool3d.out(Tensor self, int[3] output_size, *, Tensor(a!) out) -> Tensor(a!) +- func: adaptive_avg_pool3d.out(Tensor self, SymInt[3] output_size, *, Tensor(a!) out) -> Tensor(a!) python_module: nn dispatch: CPU: adaptive_avg_pool3d_out_cpu CUDA: adaptive_avg_pool3d_out_cuda QuantizedCPU: adaptive_avg_pool3d_out_quantized_cpu -- func: adaptive_avg_pool3d(Tensor self, int[3] output_size) -> Tensor +- func: adaptive_avg_pool3d(Tensor self, SymInt[3] output_size) -> Tensor python_module: nn + dispatch: + CompositeImplicitAutograd: adaptive_avg_pool3d_symint -- func: _adaptive_avg_pool3d(Tensor self, int[3] output_size) -> Tensor +- func: _adaptive_avg_pool3d(Tensor self, SymInt[3] output_size) -> Tensor dispatch: CPU: adaptive_avg_pool3d_cpu CUDA: adaptive_avg_pool3d_cuda QuantizedCPU: adaptive_avg_pool3d_quantized_cpu autogen: _adaptive_avg_pool3d.out @@ -10516,10 +11360,11 @@ python_module: nn structured_delegate: avg_pool2d.out dispatch: MkldnnCPU: mkldnn_avg_pool2d QuantizedCPU: avg_pool2d_quantized_cpu + tags: core - func: avg_pool2d_backward.grad_input(Tensor grad_output, Tensor self, int[2] kernel_size, int[2] stride, int[2] padding, bool ceil_mode, bool count_include_pad, int? divisor_override, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn structured: True dispatch: @@ -10531,10 +11376,11 @@ - func: avg_pool2d_backward(Tensor grad_output, Tensor self, int[2] kernel_size, int[2] stride, int[2] padding, bool ceil_mode, bool count_include_pad, int? divisor_override) -> Tensor python_module: nn structured_delegate: avg_pool2d_backward.grad_input dispatch: MkldnnCPU: mkldnn_avg_pool2d_backward + tags: core - func: avg_pool3d.out(Tensor self, int[3] kernel_size, int[3] stride=[], int[3] padding=0, bool ceil_mode=False, bool count_include_pad=True, int? divisor_override=None, *, Tensor(a!) out) -> Tensor(a!) python_module: nn structured: True dispatch: @@ -10627,10 +11473,11 @@ # Return: (Tensor output, Tensor indices) - func: max_pool2d_with_indices(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=0, int[2] dilation=1, bool ceil_mode=False) -> (Tensor, Tensor) python_module: nn structured_delegate: max_pool2d_with_indices.out + tags: core - func: max_pool2d_with_indices_backward.grad_input(Tensor grad_output, Tensor self, int[2] kernel_size, int[2] stride, int[2] padding, int[2] dilation, bool ceil_mode, Tensor indices, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn structured: True dispatch: @@ -10639,10 +11486,11 @@ MPS: max_pool2d_with_indices_backward_out_mps - func: max_pool2d_with_indices_backward(Tensor grad_output, Tensor self, int[2] kernel_size, int[2] stride, int[2] padding, int[2] dilation, bool ceil_mode, Tensor indices) -> Tensor python_module: nn structured_delegate: max_pool2d_with_indices_backward.grad_input + tags: core # Return: (Tensor output, Tensor indices) - func: max_pool3d_with_indices.out(Tensor self, int[3] kernel_size, int[3] stride=[], int[3] padding=0, int[3] dilation=1, bool ceil_mode=False, *, Tensor(a!) out, Tensor(b!) indices) -> (Tensor(a!), Tensor(b!)) python_module: nn dispatch: @@ -10653,10 +11501,11 @@ - func: max_pool3d_with_indices(Tensor self, int[3] kernel_size, int[3] stride=[], int[3] padding=0, int[3] dilation=1, bool ceil_mode=False) -> (Tensor, Tensor) python_module: nn dispatch: CPU: max_pool3d_with_indices_cpu CUDA: max_pool3d_with_indices_cuda + tags: core - func: max_pool3d_with_indices_backward.grad_input(Tensor grad_output, Tensor self, int[3] kernel_size, int[3] stride, int[3] padding, int[3] dilation, bool ceil_mode, Tensor indices, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn dispatch: CPU: max_pool3d_with_indices_backward_out_cpu @@ -10690,323 +11539,233 @@ python_module: nn dispatch: CPU: max_unpooling3d_forward_cpu CUDA: max_unpooling3d_forward_cuda -- func: reflection_pad1d.out(Tensor self, int[2] padding, *, Tensor(a!) out) -> Tensor(a!) +- func: reflection_pad1d.out(Tensor self, SymInt[2] padding, *, Tensor(a!) out) -> Tensor(a!) python_module: nn structured: True dispatch: CPU: reflection_pad1d_out_cpu QuantizedCPU: reflection_pad1d_out_quantized_cpu CUDA: reflection_pad1d_out_cuda MPS: reflection_pad1d_out_mps -- func: reflection_pad1d(Tensor self, int[2] padding) -> Tensor +- func: reflection_pad1d(Tensor self, SymInt[2] padding) -> Tensor python_module: nn structured_delegate: reflection_pad1d.out -- func: reflection_pad1d_backward.grad_input(Tensor grad_output, Tensor self, int[2] padding, *, Tensor(a!) grad_input) -> Tensor(a!) +- func: reflection_pad1d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[2] padding, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn structured: True dispatch: CPU: reflection_pad1d_backward_out_cpu CUDA: reflection_pad1d_backward_out_cuda MPS: reflection_pad1d_backward_out_mps -- func: reflection_pad1d_backward(Tensor grad_output, Tensor self, int[2] padding) -> Tensor +- func: reflection_pad1d_backward(Tensor grad_output, Tensor self, SymInt[2] padding) -> Tensor python_module: nn structured_delegate: reflection_pad1d_backward.grad_input -- func: reflection_pad2d.out(Tensor self, int[4] padding, *, Tensor(a!) out) -> Tensor(a!) +- func: reflection_pad2d.out(Tensor self, SymInt[4] padding, *, Tensor(a!) out) -> Tensor(a!) python_module: nn dispatch: CPU, QuantizedCPU: reflection_pad2d_out_cpu CUDA: reflection_pad2d_out_cuda MPS: reflection_pad2d_out_mps -- func: reflection_pad2d(Tensor self, int[4] padding) -> Tensor +- func: reflection_pad2d(Tensor self, SymInt[4] padding) -> Tensor python_module: nn dispatch: CPU: reflection_pad2d_cpu QuantizedCPU: reflection_pad2d_quantized_cpu CUDA: reflection_pad2d_cuda MPS: reflection_pad2d_mps + tags: core -- func: reflection_pad2d_backward.grad_input(Tensor grad_output, Tensor self, int[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!) +- func: reflection_pad2d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn dispatch: CPU: reflection_pad2d_backward_out_cpu CUDA: reflection_pad2d_backward_out_cuda MPS: reflection_pad2d_backward_out_mps -- func: reflection_pad2d_backward(Tensor grad_output, Tensor self, int[4] padding) -> Tensor +- func: reflection_pad2d_backward(Tensor grad_output, Tensor self, SymInt[4] padding) -> Tensor python_module: nn dispatch: CPU: reflection_pad2d_backward_cpu CUDA: reflection_pad2d_backward_cuda MPS: reflection_pad2d_backward_mps -- func: reflection_pad3d.out(Tensor self, int[6] padding, *, Tensor(a!) out) -> Tensor(a!) +- func: reflection_pad3d.out(Tensor self, SymInt[6] padding, *, Tensor(a!) out) -> Tensor(a!) python_module: nn structured: True dispatch: CPU: reflection_pad3d_out_cpu CUDA: reflection_pad3d_out_cuda MPS: reflection_pad3d_out_mps -- func: reflection_pad3d(Tensor self, int[6] padding) -> Tensor +- func: reflection_pad3d(Tensor self, SymInt[6] padding) -> Tensor python_module: nn structured_delegate: reflection_pad3d.out -- func: reflection_pad3d_backward.grad_input(Tensor grad_output, Tensor self, int[6] padding, *, Tensor(a!) grad_input) -> Tensor(a!) +- func: reflection_pad3d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[6] padding, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn structured: True dispatch: CPU: reflection_pad3d_backward_out_cpu CUDA: reflection_pad3d_backward_out_cuda MPS: reflection_pad3d_backward_out_mps -- func: reflection_pad3d_backward(Tensor grad_output, Tensor self, int[6] padding) -> Tensor +- func: reflection_pad3d_backward(Tensor grad_output, Tensor self, SymInt[6] padding) -> Tensor python_module: nn structured_delegate: reflection_pad3d_backward.grad_input -- func: replication_pad1d.out(Tensor self, int[2] padding, *, Tensor(a!) out) -> Tensor(a!) +- func: replication_pad1d.out(Tensor self, SymInt[2] padding, *, Tensor(a!) out) -> Tensor(a!) python_module: nn structured: True dispatch: CPU: replication_pad1d_out_cpu CUDA: replication_pad1d_out_cuda MPS: replication_pad1d_out_mps -- func: replication_pad1d(Tensor self, int[2] padding) -> Tensor +- func: replication_pad1d(Tensor self, SymInt[2] padding) -> Tensor python_module: nn structured_delegate: replication_pad1d.out -- func: replication_pad1d_backward.grad_input(Tensor grad_output, Tensor self, int[2] padding, *, Tensor(a!) grad_input) -> Tensor(a!) +- func: replication_pad1d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[2] padding, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn structured: True dispatch: CPU: replication_pad1d_backward_out_cpu CUDA: replication_pad1d_backward_out_cuda MPS: replication_pad1d_backward_out_mps -- func: replication_pad1d_backward(Tensor grad_output, Tensor self, int[2] padding) -> Tensor +- func: replication_pad1d_backward(Tensor grad_output, Tensor self, SymInt[2] padding) -> Tensor python_module: nn structured_delegate: replication_pad1d_backward.grad_input -- func: replication_pad2d.out(Tensor self, int[4] padding, *, Tensor(a!) out) -> Tensor(a!) +- func: replication_pad2d.out(Tensor self, SymInt[4] padding, *, Tensor(a!) out) -> Tensor(a!) python_module: nn structured: True dispatch: CPU: replication_pad2d_out_cpu CUDA: replication_pad2d_out_cuda MPS: replication_pad2d_out_mps -- func: replication_pad2d(Tensor self, int[4] padding) -> Tensor +- func: replication_pad2d(Tensor self, SymInt[4] padding) -> Tensor python_module: nn structured_delegate: replication_pad2d.out + tags: core -- func: replication_pad2d_backward.grad_input(Tensor grad_output, Tensor self, int[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!) +- func: replication_pad2d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[4] padding, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn dispatch: CPU: replication_pad2d_backward_out_cpu CUDA: replication_pad2d_backward_out_cuda MPS: replication_pad2d_backward_out_mps -- func: replication_pad2d_backward(Tensor grad_output, Tensor self, int[4] padding) -> Tensor +- func: replication_pad2d_backward(Tensor grad_output, Tensor self, SymInt[4] padding) -> Tensor python_module: nn dispatch: CPU: replication_pad2d_backward_cpu CUDA: replication_pad2d_backward_cuda MPS: replication_pad2d_backward_mps -- func: replication_pad3d.out(Tensor self, int[6] padding, *, Tensor(a!) out) -> Tensor(a!) +- func: replication_pad3d.out(Tensor self, SymInt[6] padding, *, Tensor(a!) out) -> Tensor(a!) python_module: nn structured: True dispatch: CPU: replication_pad3d_out_cpu CUDA: replication_pad3d_out_cuda MPS: replication_pad3d_out_mps -- func: replication_pad3d(Tensor self, int[6] padding) -> Tensor +- func: replication_pad3d(Tensor self, SymInt[6] padding) -> Tensor python_module: nn structured_delegate: replication_pad3d.out + tags: core -- func: replication_pad3d_backward.grad_input(Tensor grad_output, Tensor self, int[6] padding, *, Tensor(a!) grad_input) -> Tensor(a!) + +- func: replication_pad3d_backward.grad_input(Tensor grad_output, Tensor self, SymInt[6] padding, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn dispatch: CPU: replication_pad3d_backward_out_cpu CUDA: replication_pad3d_backward_out_cuda MPS: replication_pad3d_backward_out_mps -- func: replication_pad3d_backward(Tensor grad_output, Tensor self, int[6] padding) -> Tensor +- func: replication_pad3d_backward(Tensor grad_output, Tensor self, SymInt[6] padding) -> Tensor python_module: nn dispatch: CPU: replication_pad3d_backward_cpu CUDA: replication_pad3d_backward_cuda MPS: replication_pad3d_backward_mps -- func: _pad_circular(Tensor self, int[] pad) -> Tensor +- func: _pad_circular(Tensor self, SymInt[] pad) -> Tensor python_module: nn + dispatch: + CompositeImplicitAutograd: _pad_circular_symint -- func: _pad_enum(Tensor self, int[] pad, int mode, float? value=None) -> Tensor +- func: _pad_enum(Tensor self, SymInt[] pad, int mode, float? value=None) -> Tensor python_module: nn + dispatch: + CompositeImplicitAutograd: _pad_enum_symint -- func: pad(Tensor self, int[] pad, str mode="constant", float? value=None) -> Tensor +- func: pad(Tensor self, SymInt[] pad, str mode="constant", float? value=None) -> Tensor python_module: nn + dispatch: + CompositeImplicitAutograd: pad_symint - func: upsample_linear1d.vec(Tensor input, SymInt[]? output_size, bool align_corners, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_linear1d autogen: upsample_linear1d.vec_out -- func: upsample_linear1d_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, bool align_corners, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_linear1d_backward - autogen: upsample_linear1d_backward.vec_out - - func: upsample_bilinear2d.vec(Tensor input, SymInt[]? output_size, bool align_corners, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_bilinear2d autogen: upsample_bilinear2d.vec_out + tags: core -- func: upsample_bilinear2d_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, bool align_corners, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_bilinear2d_backward - autogen: upsample_bilinear2d_backward.vec_out - - func: _upsample_bilinear2d_aa.vec(Tensor input, SymInt[]? output_size, bool align_corners, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CompositeExplicitAutograd: _upsample_bilinear2d_aa autogen: _upsample_bilinear2d_aa.vec_out -- func: _upsample_bilinear2d_aa_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, bool align_corners, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CompositeExplicitAutograd: _upsample_bilinear2d_aa_backward - autogen: _upsample_bilinear2d_aa_backward.vec_out - - func: upsample_trilinear3d.vec(Tensor input, SymInt[]? output_size, bool align_corners, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_trilinear3d autogen: upsample_trilinear3d.vec_out -- func: upsample_trilinear3d_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, bool align_corners, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_trilinear3d_backward - autogen: upsample_trilinear3d_backward.vec_out - - func: upsample_bicubic2d.vec(Tensor input, SymInt[]? output_size, bool align_corners, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_bicubic2d autogen: upsample_bicubic2d.vec_out -- func: upsample_bicubic2d_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, bool align_corners, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_bicubic2d_backward - autogen: upsample_bicubic2d_backward.vec_out - - func: _upsample_bicubic2d_aa.vec(Tensor input, SymInt[]? output_size, bool align_corners, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CompositeExplicitAutograd: _upsample_bicubic2d_aa autogen: _upsample_bicubic2d_aa.vec_out -- func: _upsample_bicubic2d_aa_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, bool align_corners, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CompositeExplicitAutograd: _upsample_bicubic2d_aa_backward - autogen: _upsample_bicubic2d_aa_backward.vec_out - - func: upsample_nearest1d.vec(Tensor input, SymInt[]? output_size, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_nearest1d autogen: upsample_nearest1d.vec_out - func: _upsample_nearest_exact1d.vec(Tensor input, SymInt[]? output_size, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CompositeExplicitAutograd: _upsample_nearest_exact1d autogen: _upsample_nearest_exact1d.vec_out -- func: upsample_nearest1d_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_nearest1d_backward - autogen: upsample_nearest1d_backward.vec_out - -- func: _upsample_nearest_exact1d_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CompositeExplicitAutograd: _upsample_nearest_exact1d_backward - autogen: _upsample_nearest_exact1d_backward.vec_out - - func: upsample_nearest2d.vec(Tensor input, SymInt[]? output_size, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_nearest2d autogen: upsample_nearest2d.vec_out + tags: core - func: _upsample_nearest_exact2d.vec(Tensor input, SymInt[]? output_size, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CompositeExplicitAutograd: _upsample_nearest_exact2d autogen: _upsample_nearest_exact2d.vec_out -- func: upsample_nearest2d_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CompositeExplicitAutograd: upsample_nearest2d_backward - autogen: upsample_nearest2d_backward.vec_out - -- func: _upsample_nearest_exact2d_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CompositeExplicitAutograd: _upsample_nearest_exact2d_backward - autogen: _upsample_nearest_exact2d_backward.vec_out - - func: upsample_nearest3d.vec(Tensor input, SymInt[]? output_size, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CPU: upsample_nearest3d_cpu - CUDA: upsample_nearest3d_cuda - QuantizedCPU: upsample_nearest3d_quantized_cpu autogen: upsample_nearest3d.vec_out - func: _upsample_nearest_exact3d.vec(Tensor input, SymInt[]? output_size, float[]? scale_factors) -> Tensor python_module: nn - dispatch: - CPU: _upsample_nearest_exact3d_cpu - CUDA: _upsample_nearest_exact3d_cuda - QuantizedCPU: _upsample_nearest_exact3d_quantized_cpu autogen: _upsample_nearest_exact3d.vec_out -- func: upsample_nearest3d_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CPU: upsample_nearest3d_backward_cpu - CUDA: upsample_nearest3d_backward_cuda - autogen: upsample_nearest3d_backward.vec_out - -- func: _upsample_nearest_exact3d_backward.vec(Tensor grad_output, SymInt[]? output_size, SymInt[] input_size, float[]? scale_factors) -> Tensor - python_module: nn - dispatch: - CPU: _upsample_nearest_exact3d_backward_cpu - CUDA: _upsample_nearest_exact3d_backward_cuda - autogen: _upsample_nearest_exact3d_backward.vec_out - # NOTE: all of the non-"vec" upsample overloads are only kept for backward compatibility. - func: upsample_linear1d.out(Tensor self, SymInt[1] output_size, bool align_corners, float? scales=None, *, Tensor(a!) out) -> Tensor(a!) python_module: nn structured: True dispatch: @@ -11154,10 +11913,11 @@ python_module: nn structured: True dispatch: CPU: _upsample_nearest_exact1d_out_cpu CUDA: _upsample_nearest_exact1d_out_cuda + MPS: _upsample_nearest_exact1d_out_mps - func: upsample_nearest1d(Tensor self, SymInt[1] output_size, float? scales=None) -> Tensor python_module: nn structured_delegate: upsample_nearest1d.out @@ -11169,17 +11929,19 @@ python_module: nn structured: True dispatch: CPU: upsample_nearest1d_backward_out_cpu CUDA: upsample_nearest1d_backward_out_cuda + MPS: upsample_nearest1d_backward_out_mps - func: _upsample_nearest_exact1d_backward.grad_input(Tensor grad_output, SymInt[1] output_size, SymInt[3] input_size, float? scales=None, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn structured: True dispatch: CPU: _upsample_nearest_exact1d_backward_out_cpu CUDA: _upsample_nearest_exact1d_backward_out_cuda + MPS: _upsample_nearest_exact1d_backward_out_mps - func: upsample_nearest1d_backward(Tensor grad_output, SymInt[1] output_size, SymInt[3] input_size, float? scales=None) -> Tensor python_module: nn structured_delegate: upsample_nearest1d_backward.grad_input @@ -11292,33 +12054,38 @@ structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: sigmoid_backward_out MPS: sigmoid_backward_out_mps + tags: pointwise - func: sigmoid_backward(Tensor grad_output, Tensor output) -> Tensor python_module: nn structured_delegate: sigmoid_backward.grad_input + tags: pointwise - func: logit_backward.grad_input(Tensor grad_output, Tensor self, float? eps=None, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: logit_backward_out + tags: pointwise - func: logit_backward(Tensor grad_output, Tensor self, float? eps=None) -> Tensor python_module: nn structured_delegate: logit_backward.grad_input + tags: pointwise - func: tanh_backward.grad_input(Tensor grad_output, Tensor output, *, Tensor(a!) grad_input) -> Tensor(a!) python_module: nn structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: tanh_backward_out MPS: tanh_backward_out_mps + tags: pointwise - func: tanh_backward(Tensor grad_output, Tensor output) -> Tensor python_module: nn structured_delegate: tanh_backward.grad_input @@ -11337,29 +12104,30 @@ # to a convolution that is still written in the "legacy" style; that is, # C code in the THNN/ or THCUNN/ directory. A slow_ convolution is # one that is written in the native style: modern C++. Algorithmically, # these are the same thing, but we give them different prefixes to # make the operational distinction clear. + tags: pointwise -- func: slow_conv_transpose2d.out(Tensor self, Tensor weight, int[2] kernel_size, Tensor? bias=None, int[2] stride=1, int[2] padding=0, int[2] output_padding=0, int[2] dilation=1, *, Tensor(a!) out) -> Tensor(a!) +- func: slow_conv_transpose2d.out(Tensor self, Tensor weight, int[2] kernel_size, Tensor? bias=None, int[2] stride=1, SymInt[2] padding=0, SymInt[2] output_padding=0, int[2] dilation=1, *, Tensor(a!) out) -> Tensor(a!) python_module: nn structured: True dispatch: CPU: slow_conv_transpose2d_structured_cpu CUDA: slow_conv_transpose2d_structured_cuda -- func: slow_conv_transpose2d(Tensor self, Tensor weight, int[2] kernel_size, Tensor? bias=None, int[2] stride=1, int[2] padding=0, int[2] output_padding=0, int[2] dilation=1) -> Tensor +- func: slow_conv_transpose2d(Tensor self, Tensor weight, int[2] kernel_size, Tensor? bias=None, int[2] stride=1, SymInt[2] padding=0, SymInt[2] output_padding=0, int[2] dilation=1) -> Tensor python_module: nn structured_delegate: slow_conv_transpose2d.out -- func: slow_conv_transpose3d.out(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias=None, int[3] stride=1, int[3] padding=0, int[3] output_padding=0, int[3] dilation=1, *, Tensor(a!) out) -> Tensor(a!) +- func: slow_conv_transpose3d.out(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias=None, int[3] stride=1, SymInt[3] padding=0, SymInt[3] output_padding=0, int[3] dilation=1, *, Tensor(a!) out) -> Tensor(a!) python_module: nn dispatch: CPU: slow_conv_transpose3d_out_cpu CUDA: slow_conv_transpose3d_out_cuda -- func: slow_conv_transpose3d(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias=None, int[3] stride=1, int[3] padding=0, int[3] output_padding=0, int[3] dilation=1) -> Tensor +- func: slow_conv_transpose3d(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias=None, int[3] stride=1, SymInt[3] padding=0, SymInt[3] output_padding=0, int[3] dilation=1) -> Tensor python_module: nn dispatch: CPU: slow_conv_transpose3d_cpu CUDA: slow_conv_transpose3d_cuda @@ -11392,51 +12160,51 @@ dispatch: CPU: slow_conv2d_backward_cpu CUDA: slow_conv2d_backward_cuda autogen: _slow_conv2d_backward.output_mask_out -- func: _conv_depthwise2d.out(Tensor self, Tensor weight, int[2] kernel_size, Tensor? bias, int[2] stride, int[2] padding, int[2] dilation, *, Tensor(a!) out) -> Tensor(a!) +- func: _conv_depthwise2d.out(Tensor self, Tensor weight, int[2] kernel_size, Tensor? bias, int[2] stride, SymInt[2] padding, int[2] dilation, *, Tensor(a!) out) -> Tensor(a!) use_const_ref_for_mutable_tensors: True python_module: nn dispatch: CUDA: conv_depthwise2d_cuda_out -- func: _conv_depthwise2d(Tensor self, Tensor weight, int[2] kernel_size, Tensor? bias, int[2] stride, int[2] padding, int[2] dilation) -> Tensor +- func: _conv_depthwise2d(Tensor self, Tensor weight, int[2] kernel_size, Tensor? bias, int[2] stride, SymInt[2] padding, int[2] dilation) -> Tensor python_module: nn dispatch: CUDA: conv_depthwise2d_cuda -- func: conv_depthwise3d(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias, int[3] stride, int[3] padding, int[3] dilation) -> Tensor +- func: conv_depthwise3d(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias, int[3] stride, SymInt[3] padding, int[3] dilation) -> Tensor python_module: nn dispatch: CUDA: conv_depthwise3d_cuda autogen: conv_depthwise3d.out -- func: slow_conv3d.out(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias=None, int[3] stride=1, int[3] padding=0, *, Tensor(a!) out) -> Tensor(a!) +- func: slow_conv3d.out(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias=None, int[3] stride=1, SymInt[3] padding=0, *, Tensor(a!) out) -> Tensor(a!) python_module: nn -- func: slow_conv3d(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias=None, int[3] stride=1, int[3] padding=0) -> Tensor +- func: slow_conv3d(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias=None, int[3] stride=1, SymInt[3] padding=0) -> Tensor python_module: nn -- func: slow_conv3d_forward.output(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias, int[3] stride, int[3] padding, *, Tensor(a!) output) -> Tensor(a!) +- func: slow_conv3d_forward.output(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias, int[3] stride, SymInt[3] padding, *, Tensor(a!) output) -> Tensor(a!) python_module: nn dispatch: CPU: slow_conv3d_forward_out_cpu -- func: slow_conv3d_forward(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias, int[3] stride, int[3] padding) -> Tensor +- func: slow_conv3d_forward(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias, int[3] stride, SymInt[3] padding) -> Tensor python_module: nn dispatch: CPU: slow_conv3d_forward_cpu -- func: slow_conv_dilated2d(Tensor self, Tensor weight, int[2] kernel_size, Tensor? bias=None, int[2] stride=1, int[2] padding=0, int[2] dilation=1) -> Tensor +- func: slow_conv_dilated2d(Tensor self, Tensor weight, int[2] kernel_size, Tensor? bias=None, int[2] stride=1, SymInt[2] padding=0, int[2] dilation=1) -> Tensor python_module: nn dispatch: CPU: slow_conv_dilated2d_cpu CUDA: slow_conv_dilated2d_cuda autogen: slow_conv_dilated2d.out -- func: slow_conv_dilated3d(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias=None, int[3] stride=1, int[3] padding=0, int[3] dilation=1) -> Tensor +- func: slow_conv_dilated3d(Tensor self, Tensor weight, int[3] kernel_size, Tensor? bias=None, int[3] stride=1, SymInt[3] padding=0, int[3] dilation=1) -> Tensor python_module: nn dispatch: CPU: slow_conv_dilated3d_cpu CUDA: slow_conv_dilated3d_cuda autogen: slow_conv_dilated3d.out @@ -11450,10 +12218,11 @@ - func: col2im(Tensor self, SymInt[2] output_size, int[2] kernel_size, int[2] dilation, int[2] padding, int[2] stride) -> Tensor python_module: nn dispatch: CPU: col2im_cpu CUDA: col2im_cuda + tags: core - func: column_stack(Tensor[] tensors) -> Tensor - func: column_stack.out(Tensor[] tensors, *, Tensor(a!) out) -> Tensor(a!) @@ -11482,10 +12251,11 @@ CompositeExplicitAutograd: isinf SparseCPU, SparseCUDA: isinf_sparse SparseMeta: isinf_sparse_meta SparseCsrCPU, SparseCsrCUDA: isinf_sparse_csr autogen: isinf.out + tags: core - func: record_stream(Tensor(a!) self, Stream s) -> () variants: method dispatch: CUDA: record_stream_cuda @@ -11494,33 +12264,37 @@ variants: function, method structured_delegate: isposinf.out dispatch: SparseCPU, SparseCUDA: isposinf_sparse SparseCsrCPU, SparseCsrCUDA: isposinf_sparse_csr + tags: pointwise - func: isposinf.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: isposinf_out SparseCPU, SparseCUDA: isposinf_sparse_out SparseCsrCPU, SparseCsrCUDA: isposinf_sparse_csr_out + tags: pointwise - func: isneginf(Tensor self) -> Tensor variants: function, method structured_delegate: isneginf.out dispatch: SparseCPU, SparseCUDA: isneginf_sparse SparseCsrCPU, SparseCsrCUDA: isneginf_sparse_csr + tags: pointwise - func: isneginf.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: isneginf_out SparseCPU, SparseCUDA: isneginf_sparse_out SparseCsrCPU, SparseCsrCUDA: isneginf_sparse_csr_out + tags: pointwise # NOTE [_add_batch_dim and _remove_batch_dim] # _add_batch_dim and _remove_batch_dim are meant to be used in the implementation # of the vmap frontend API (see torch/_vmap_internals.py). They are not # user-facing, hence the leading underscore. Please don't use them them anywhere else. @@ -11540,44 +12314,50 @@ - func: special_entr(Tensor self) -> Tensor structured_delegate: special_entr.out python_module: special variants: function + tags: pointwise - func: special_entr.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase python_module: special variants: function dispatch: CPU, CUDA: special_entr_out + tags: pointwise - func: special_ndtri(Tensor self) -> Tensor structured_delegate: special_ndtri.out python_module: special variants: function + tags: pointwise - func: special_ndtri.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase python_module: special variants: function dispatch: CPU, CUDA: special_ndtri_out + tags: pointwise - func: special_log_ndtr(Tensor self) -> Tensor structured_delegate: special_log_ndtr.out python_module: special variants: function + tags: pointwise - func: special_log_ndtr.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) structured: True structured_inherits: TensorIteratorBase python_module: special variants: function dispatch: CPU, CUDA: special_log_ndtr_out + tags: pointwise - func: special_expm1(Tensor self) -> Tensor python_module: special variants: function @@ -11634,17 +12414,19 @@ - func: special_erfcx(Tensor self) -> Tensor python_module: special variants: function structured_delegate: special_erfcx.out + tags: pointwise - func: special_erfcx.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) python_module: special structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: special_erfcx_out + tags: pointwise - func: special_erfinv(Tensor self) -> Tensor python_module: special variants: function @@ -11662,47 +12444,53 @@ - func: special_xlog1py(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator python_module: special variants: function structured_delegate: special_xlog1py.out + tags: pointwise - func: special_xlog1py.self_scalar(Scalar self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator python_module: special variants: function dispatch: CompositeExplicitAutograd: special_xlog1py + tags: pointwise - func: special_xlog1py.other_scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator python_module: special variants: function dispatch: CompositeExplicitAutograd: special_xlog1py + tags: pointwise - func: special_xlog1py.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase python_module: special variants: function dispatch: CPU, CUDA: special_xlog1py_out + tags: pointwise - func: special_xlog1py.self_scalar_out(Scalar self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator python_module: special variants: function dispatch: CompositeExplicitAutograd: special_xlog1py_out + tags: pointwise - func: special_xlog1py.other_scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator python_module: special variants: function dispatch: CompositeExplicitAutograd: special_xlog1py_out + tags: pointwise - func: special_xlogy(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator python_module: special variants: function @@ -11735,47 +12523,53 @@ - func: special_zeta(Tensor self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator python_module: special variants: function structured_delegate: special_zeta.out + tags: pointwise - func: special_zeta.self_scalar(Scalar self, Tensor other) -> Tensor device_check: NoCheck # TensorIterator python_module: special variants: function dispatch: CompositeExplicitAutograd: special_zeta + tags: pointwise - func: special_zeta.other_scalar(Tensor self, Scalar other) -> Tensor device_check: NoCheck # TensorIterator python_module: special variants: function dispatch: CompositeExplicitAutograd: special_zeta + tags: pointwise - func: special_zeta.out(Tensor self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator structured: True structured_inherits: TensorIteratorBase python_module: special variants: function dispatch: CPU, CUDA: special_zeta_out + tags: pointwise - func: special_zeta.self_scalar_out(Scalar self, Tensor other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator python_module: special variants: function dispatch: CompositeExplicitAutograd: special_zeta_out + tags: pointwise - func: special_zeta.other_scalar_out(Tensor self, Scalar other, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck # TensorIterator python_module: special variants: function dispatch: CompositeExplicitAutograd: special_zeta_out + tags: pointwise - func: special_i0(Tensor self) -> Tensor python_module: special variants: function @@ -11785,41 +12579,47 @@ - func: special_i0e(Tensor self) -> Tensor python_module: special variants: function structured_delegate: special_i0e.out + tags: pointwise - func: special_i0e.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) python_module: special structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: special_i0e_out + tags: pointwise - func: special_i1(Tensor self) -> Tensor python_module: special variants: function structured_delegate: special_i1.out + tags: pointwise - func: special_i1.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) python_module: special structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: special_i1_out + tags: pointwise - func: special_i1e(Tensor self) -> Tensor python_module: special variants: function structured_delegate: special_i1e.out + tags: pointwise - func: special_i1e.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) python_module: special structured: True structured_inherits: TensorIteratorBase dispatch: CPU, CUDA: special_i1e_out + tags: pointwise - func: special_logit(Tensor self, float? eps=None) -> Tensor python_module: special variants: function @@ -12134,11 +12934,11 @@ - func: linalg_cross.out(Tensor self, Tensor other, *, int dim=-1, Tensor(a!) out) -> Tensor(a!) python_module: linalg structured: True dispatch: - CPU, CUDA: linalg_cross_out + CPU, CUDA, MPS: linalg_cross_out # linalg.lu_factor - func: linalg_lu_factor(Tensor A, *, bool pivot=True) -> (Tensor LU, Tensor pivots) python_module: linalg variants: function @@ -12353,10 +13153,11 @@ - func: linalg_inv_ex.inverse(Tensor A, *, bool check_errors=False, Tensor(a!) inverse, Tensor(b!) info) -> (Tensor(a!) inverse, Tensor(b!) info) python_module: linalg structured: True dispatch: CPU, CUDA: linalg_inv_ex_out + MPS: linalg_inv_ex_out_mps - func: linalg_inv(Tensor A) -> Tensor python_module: linalg - func: linalg_inv.out(Tensor A, *, Tensor(a!) out) -> Tensor(a!) @@ -12713,462 +13514,362 @@ - func: _fw_primal_copy(Tensor self, int level) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: _fw_primal_copy tags: view_copy + autogen: _fw_primal_copy.out - func: _make_dual_copy(Tensor primal, Tensor tangent, int level) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: _make_dual_copy tags: view_copy + autogen: _make_dual_copy.out - func: view_as_real_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: view_as_real_copy tags: view_copy + autogen: view_as_real_copy.out - func: view_as_complex_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: view_as_complex_copy tags: view_copy + autogen: view_as_complex_copy.out - func: _conj_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: _conj_copy tags: view_copy + autogen: _conj_copy.out - func: _neg_view_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: _neg_view_copy tags: view_copy + autogen: _neg_view_copy.out - func: as_strided_copy(Tensor self, SymInt[] size, SymInt[] stride, SymInt? storage_offset=None) -> Tensor variants: function dispatch: - CompositeExplicitAutogradNonFunctional: as_strided_copy + CompositeExplicitAutogradNonFunctional: as_strided_copy_symint tags: view_copy + autogen: as_strided_copy.out - func: _sparse_broadcast_to_copy(Tensor self, int[] size) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: _sparse_broadcast_to_copy tags: view_copy + autogen: _sparse_broadcast_to_copy.out - func: diagonal_copy(Tensor self, int offset=0, int dim1=0, int dim2=1) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: diagonal_copy tags: view_copy + autogen: diagonal_copy.out - func: expand_copy(Tensor self, SymInt[] size, *, bool implicit=False) -> Tensor variants: function dispatch: - CompositeExplicitAutogradNonFunctional: expand_copy + CompositeExplicitAutogradNonFunctional: expand_copy_symint tags: view_copy + autogen: expand_copy.out - func: permute_copy(Tensor self, int[] dims) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: permute_copy tags: view_copy + autogen: permute_copy.out - func: _reshape_alias_copy(Tensor self, SymInt[] size, SymInt[] stride) -> Tensor variants: function dispatch: - CompositeExplicitAutogradNonFunctional: _reshape_alias_copy + CompositeExplicitAutogradNonFunctional: _reshape_alias_copy_symint tags: view_copy + autogen: _reshape_alias_copy.out -- func: select_copy.int(Tensor self, int dim, int index) -> Tensor +- func: select_copy.int(Tensor self, int dim, SymInt index) -> Tensor variants: function dispatch: - CompositeExplicitAutogradNonFunctional: select_copy_int + CompositeExplicitAutogradNonFunctional: select_copy_symint + SparseCsrCPU, SparseCsrCUDA: select_copy_sparse_csr tags: view_copy + autogen: select_copy.int_out - func: detach_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: detach_copy tags: view_copy + autogen: detach_copy.out - func: slice_copy.Tensor(Tensor self, int dim=0, SymInt? start=None, SymInt? end=None, SymInt step=1) -> Tensor variants: function dispatch: - CompositeExplicitAutogradNonFunctional: slice_copy_Tensor + CompositeExplicitAutogradNonFunctional: slice_copy_Tensor_symint tags: view_copy + autogen: slice_copy.Tensor_out -- func: split_copy.Tensor(Tensor self, int split_size, int dim=0) -> Tensor[] +- func: split_copy.Tensor(Tensor self, SymInt split_size, int dim=0) -> Tensor[] variants: function dispatch: - CompositeExplicitAutogradNonFunctional: split_copy_Tensor + CompositeExplicitAutogradNonFunctional: split_copy_Tensor_symint tags: view_copy -- func: split_with_sizes_copy(Tensor self, int[] split_sizes, int dim=0) -> Tensor[] +- func: split_with_sizes_copy(Tensor self, SymInt[] split_sizes, int dim=0) -> Tensor[] variants: function dispatch: - CompositeExplicitAutogradNonFunctional: split_with_sizes_copy + CompositeExplicitAutogradNonFunctional: split_with_sizes_copy_symint tags: view_copy - func: squeeze_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: squeeze_copy tags: view_copy + autogen: squeeze_copy.out - func: squeeze_copy.dim(Tensor self, int dim) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: squeeze_copy_dim tags: view_copy + autogen: squeeze_copy.dim_out +- func: squeeze_copy.dims(Tensor self, int[] dim) -> Tensor + variants: function + dispatch: + CompositeExplicitAutogradNonFunctional: squeeze_copy_dims + tags: view_copy + autogen: squeeze_copy.dims_out + - func: t_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: t_copy tags: view_copy + autogen: t_copy.out - func: transpose_copy.int(Tensor self, int dim0, int dim1) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: transpose_copy_int tags: view_copy + autogen: transpose_copy.int_out - func: unsqueeze_copy(Tensor self, int dim) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: unsqueeze_copy tags: view_copy + autogen: unsqueeze_copy.out - func: _indices_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: _indices_copy tags: view_copy + autogen: _indices_copy.out - func: _values_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: _values_copy tags: view_copy + autogen: _values_copy.out - func: indices_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: indices_copy tags: view_copy + autogen: indices_copy.out - func: values_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: values_copy tags: view_copy + autogen: values_copy.out - func: crow_indices_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: crow_indices_copy tags: view_copy + autogen: crow_indices_copy.out - func: col_indices_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: col_indices_copy tags: view_copy + autogen: col_indices_copy.out - func: ccol_indices_copy(Tensor self) -> Tensor variants: function dispatch: - CompositeExplicitAutograd: ccol_indices_copy + CompositeExplicitAutogradNonFunctional: ccol_indices_copy tags: view_copy autogen: ccol_indices_copy.out - func: row_indices_copy(Tensor self) -> Tensor variants: function dispatch: - CompositeExplicitAutograd: row_indices_copy + CompositeExplicitAutogradNonFunctional: row_indices_copy tags: view_copy autogen: row_indices_copy.out - func: unbind_copy.int(Tensor self, int dim=0) -> Tensor[] variants: function dispatch: CompositeExplicitAutogradNonFunctional: unbind_copy_int tags: view_copy +- func: unbind_copy.int_out(Tensor self, int dim=0, *, Tensor(a!)[] out) -> () + variants: function + dispatch: + CompositeExplicitAutograd: unbind_copy_int_out + +- func: split_copy.Tensor_out(Tensor self, SymInt split_size, int dim=0, *, Tensor(a!)[] out) -> () + variants: function + dispatch: + CompositeExplicitAutograd: split_copy_Tensor_out + + +- func: split_with_sizes_copy.out(Tensor self, SymInt[] split_sizes, int dim=0, *, Tensor(a!)[] out) -> () + variants: function + dispatch: + CompositeExplicitAutograd: split_with_sizes_copy_out + - func: view_copy(Tensor self, SymInt[] size) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: view_copy_symint tags: view_copy + autogen: view_copy.out - func: view_copy.dtype(Tensor self, ScalarType dtype) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: view_copy_dtype tags: view_copy + autogen: view_copy.dtype_out - func: unfold_copy(Tensor self, int dimension, int size, int step) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: unfold_copy tags: view_copy + autogen: unfold_copy.out - func: alias_copy(Tensor self) -> Tensor variants: function dispatch: CompositeExplicitAutogradNonFunctional: alias_copy tags: view_copy + autogen: alias_copy.out -- func: _fw_primal_copy.out(Tensor self, int level, *, Tensor(a!) out) -> Tensor(a!) - variants: function +- func: to_padded_tensor(Tensor self, float padding, SymInt[]? output_size=None) -> Tensor + variants: method dispatch: - CompositeExplicitAutograd: _fw_primal_copy_out + NestedTensorCPU: NestedTensor_to_padded_tensor_generic + NestedTensorCUDA: NestedTensor_to_padded_tensor_cuda + autogen: to_padded_tensor.out - -- func: _make_dual_copy.out(Tensor primal, Tensor tangent, int level, *, Tensor(a!) out) -> Tensor(a!) - variants: function +- func: _nested_tensor_softmax_with_shape(Tensor self, Tensor query) -> Tensor dispatch: - CompositeExplicitAutograd: _make_dual_copy_out + NestedTensorCPU: NestedTensor_softmax_dropout + NestedTensorCUDA: NestedTensor_softmax_dropout_cuda - -- func: view_as_real_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) +# Apparently, putting "forward" in the name will cause Python bindings to be skipped, so "fwd" it is. +- func: _transformer_encoder_layer_fwd(Tensor src, int embed_dim, int num_heads, Tensor qkv_weight, Tensor qkv_bias, Tensor proj_weight, Tensor proj_bias, bool use_gelu, bool norm_first, float eps, Tensor norm_weight_1, Tensor norm_bias_1, Tensor norm_weight_2, Tensor norm_bias_2, Tensor ffn_weight_1, Tensor ffn_bias_1, Tensor ffn_weight_2, Tensor ffn_bias_2, Tensor? mask=None, int? mask_type=None) -> Tensor variants: function dispatch: - CompositeExplicitAutograd: view_as_real_copy_out + CPU, CUDA, NestedTensorCPU, NestedTensorCUDA: transformer_encoder_layer_forward + autogen: _transformer_encoder_layer_fwd.out - -- func: view_as_complex_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) +- func: _native_multi_head_attention(Tensor query, Tensor key, Tensor value, int embed_dim, int num_head, Tensor qkv_weight, Tensor qkv_bias, Tensor proj_weight, Tensor proj_bias, Tensor? mask=None, bool need_weights=True, bool average_attn_weights=True, int? mask_type=None) -> (Tensor, Tensor) variants: function dispatch: - CompositeExplicitAutograd: view_as_complex_copy_out + CPU, NestedTensorCPU: native_multi_head_attention_cpu + CUDA, NestedTensorCUDA: native_multi_head_attention_cuda + autogen: _native_multi_head_attention.out - -- func: _conj_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) +- func: scaled_dot_product_attention(Tensor query, Tensor key, Tensor value, Tensor? attn_mask=None, float dropout_p=0.0, bool is_causal=False) -> Tensor + python_module: nn variants: function - dispatch: - CompositeExplicitAutograd: _conj_copy_out + autogen: scaled_dot_product_attention.out - -- func: _neg_view_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) +# TODO: THIS NEEDS TO BE REMOVED BUT PEOPLE HAVE TRAINED THEIR MODELS WITH THIS OP BUILTIN +- func: _scaled_dot_product_attention(Tensor query, Tensor key, Tensor value, Tensor? attn_mask=None, float dropout_p=0.0, bool need_attn_weights=False, bool is_causal=False) -> (Tensor, Tensor) + python_module: nn variants: function - dispatch: - CompositeExplicitAutograd: _neg_view_copy_out + autogen: _scaled_dot_product_attention.out - -- func: as_strided_copy.out(Tensor self, SymInt[] size, SymInt[] stride, SymInt? storage_offset=None, *, Tensor(a!) out) -> Tensor(a!) - variants: function +# This aten function is kept so that we can test the choice function from Python +- func: _fused_sdp_choice(Tensor query, Tensor key, Tensor value, Tensor? attn_mask=None, float dropout_p=0.0, bool is_causal=False) -> int dispatch: - CompositeExplicitAutograd: as_strided_copy_out_symint + Meta: _fused_sdp_choice_meta + CPU, NestedTensorCPU: _fused_sdp_choice_cpp + CUDA, NestedTensorCUDA: _fused_sdp_choice_cuda - -- func: _sparse_broadcast_to_copy.out(Tensor self, int[] size, *, Tensor(a!) out) -> Tensor(a!) +- func: _scaled_dot_product_attention_math(Tensor query, Tensor key, Tensor value, Tensor? attn_mask=None, float dropout_p=0.0, bool is_causal=False, Tensor? dropout_mask=None) -> (Tensor, Tensor) variants: function - dispatch: - CompositeExplicitAutograd: _sparse_broadcast_to_copy_out - -- func: diagonal_copy.out(Tensor self, int offset=0, int dim1=0, int dim2=1, *, Tensor(a!) out) -> Tensor(a!) - variants: function +- func: _scaled_dot_product_flash_attention(Tensor query, Tensor key, Tensor value, float dropout_p=0.0, bool is_causal=False, bool return_debug_mask=False) -> (Tensor ouput, Tensor logsumexp, Tensor cum_seq_q, Tensor cum_seq_k, int max_q, int max_k, int philox_seed, int philox_offset, Tensor debug_attn_mask) dispatch: - CompositeExplicitAutograd: diagonal_copy_out + CUDA: _scaled_dot_product_flash_attention_cuda + NestedTensorCUDA: _scaled_dot_product_flash_attention_nestedtensor_cuda - -- func: expand_copy.out(Tensor self, SymInt[] size, *, bool implicit=False, Tensor(a!) out) -> Tensor(a!) +- func: _scaled_dot_product_flash_attention_backward(Tensor grad_out, Tensor query, Tensor key, Tensor value, Tensor out, Tensor logsumexp, Tensor cum_seq_q, Tensor cum_seq_k, int max_q, int max_k, float dropout_p, bool is_causal, int philox_seed, int philox_offset) -> (Tensor grad_query, Tensor grad_key, Tensor grad_value) variants: function dispatch: - CompositeExplicitAutograd: expand_copy_out_symint + CUDA: _scaled_dot_product_flash_attention_backward_cuda - -- func: permute_copy.out(Tensor self, int[] dims, *, Tensor(a!) out) -> Tensor(a!) - variants: function +- func: _scaled_dot_product_efficient_attention(Tensor query, Tensor key, Tensor value, bool compute_log_sumexp, bool is_causal=False) -> (Tensor, Tensor) dispatch: - CompositeExplicitAutograd: permute_copy_out + CUDA: _scaled_dot_product_efficient_attention_cuda + NestedTensorCUDA: _scaled_dot_product_efficient_attention_nestedtensor_cuda - -- func: _reshape_alias_copy.out(Tensor self, SymInt[] size, SymInt[] stride, *, Tensor(a!) out) -> Tensor(a!) - variants: function +- func: _scaled_dot_product_efficient_attention_backward(Tensor grad_out_, Tensor query, Tensor key, Tensor value, Tensor out, Tensor logsumexp, bool is_causal=False, bool chunk_grad_outputs=False) -> (Tensor, Tensor, Tensor) dispatch: - CompositeExplicitAutograd: _reshape_alias_copy_out + CUDA: _scaled_dot_product_efficient_attention_backward_cuda - -- func: select_copy.int_out(Tensor self, int dim, int index, *, Tensor(a!) out) -> Tensor(a!) - variants: function +- func: _chunk_grad_outputs_efficient_attention(Tensor query, Tensor key, Tensor value, bool is_causal=False) -> bool dispatch: - CompositeExplicitAutograd: select_copy_int_out + CUDA: _chunk_grad_outputs_efficient_attention - -- func: detach_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) +- func: _flash_attention_forward(Tensor query, Tensor key, Tensor value, Tensor cum_seq_q, Tensor cum_seq_k, int max_q, int max_k, float dropout_p, bool is_causal, bool return_debug_mask) -> (Tensor output, Tensor softmax_logsumexp, int philox_seed, int philox_offset, Tensor debug_attn_mask) variants: function dispatch: - CompositeExplicitAutograd: detach_copy_out + CUDA: _flash_attention_forward - -- func: slice_copy.Tensor_out(Tensor self, int dim=0, SymInt? start=None, SymInt? end=None, SymInt step=1, *, Tensor(a!) out) -> Tensor(a!) +- func: _flash_attention_backward(Tensor grad_out, Tensor query, Tensor key, Tensor value, Tensor out, Tensor logsumexp, Tensor cum_seq_q, Tensor cum_seq_k, int max_q, int max_k, float dropout_p, bool is_causal, int philox_seed, int philox_offset) -> (Tensor, Tensor, Tensor) variants: function dispatch: - CompositeExplicitAutograd: slice_copy_Tensor_out + CUDA: _flash_attention_backward - -- func: split_copy.Tensor_out(Tensor self, int split_size, int dim=0, *, Tensor(a!)[] out) -> () +# Returns ouput, logsumexp if compute_logsumexp +- func: _efficient_attention_forward(Tensor query, Tensor key, Tensor value, Tensor? cu_seqlens_q, Tensor? cu_seqlens_k, int? max_seqlen_q, bool compute_log_sumexp=False, bool causal=False) -> (Tensor, Tensor) variants: function dispatch: - CompositeExplicitAutograd: split_copy_Tensor_out + CUDA: _efficient_attention_forward - -- func: split_with_sizes_copy.out(Tensor self, int[] split_sizes, int dim=0, *, Tensor(a!)[] out) -> () +- func: _efficient_attention_backward(Tensor grad_out_, Tensor query, Tensor key, Tensor value, Tensor out, Tensor logsumexp, bool is_causal=False, bool chunk_grad_outputs=False) -> (Tensor, Tensor, Tensor) variants: function dispatch: - CompositeExplicitAutograd: split_with_sizes_copy_out + CUDA: _efficient_attention_backward - -- func: squeeze_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: squeeze_copy_out - - -- func: squeeze_copy.dim_out(Tensor self, int dim, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: squeeze_copy_dim_out - - -- func: t_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: t_copy_out - - -- func: transpose_copy.int_out(Tensor self, int dim0, int dim1, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: transpose_copy_int_out - - -- func: unsqueeze_copy.out(Tensor self, int dim, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: unsqueeze_copy_out - - -- func: _indices_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: _indices_copy_out - - -- func: _values_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: _values_copy_out - - -- func: indices_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: indices_copy_out - - -- func: values_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: values_copy_out - - -- func: crow_indices_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: crow_indices_copy_out - - -- func: col_indices_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: col_indices_copy_out - - -- func: unbind_copy.int_out(Tensor self, int dim=0, *, Tensor(a!)[] out) -> () - variants: function - dispatch: - CompositeExplicitAutograd: unbind_copy_int_out - - -- func: view_copy.out(Tensor self, SymInt[] size, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: view_copy_out_symint - - -- func: view_copy.dtype_out(Tensor self, ScalarType dtype, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: view_copy_dtype_out - - -- func: unfold_copy.out(Tensor self, int dimension, int size, int step, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: unfold_copy_out - - -- func: alias_copy.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) - variants: function - dispatch: - CompositeExplicitAutograd: alias_copy_out - -- func: to_padded_tensor(Tensor self, float padding, int[]? output_size=None) -> Tensor - variants: method - dispatch: - NestedTensorCPU: NestedTensor_to_padded_tensor_generic - NestedTensorCUDA: NestedTensor_to_padded_tensor_cuda - autogen: to_padded_tensor.out - -- func: _nested_tensor_softmax_with_shape(Tensor self, Tensor query) -> Tensor - dispatch: - NestedTensorCPU: NestedTensor_softmax_dropout - NestedTensorCUDA: NestedTensor_softmax_dropout_cuda - -- func: _nested_tensor_layer_norm(Tensor self, Tensor? weight, Tensor? bias, float eps) -> Tensor - variants: method - dispatch: - NestedTensorCPU, NestedTensorCUDA: NestedTensor_layer_norm - autogen: _nested_tensor_layer_norm.out - -# Apparently, putting "forward" in the name will cause Python bindings to be skipped, so "fwd" it is. -- func: _transformer_encoder_layer_fwd(Tensor src, int embed_dim, int num_heads, Tensor qkv_weight, Tensor qkv_bias, Tensor proj_weight, Tensor proj_bias, bool use_gelu, bool norm_first, float eps, Tensor norm_weight_1, Tensor norm_bias_1, Tensor norm_weight_2, Tensor norm_bias_2, Tensor ffn_weight_1, Tensor ffn_bias_1, Tensor ffn_weight_2, Tensor ffn_bias_2, Tensor? mask=None, int? mask_type=None) -> Tensor - variants: function - dispatch: - CPU, CUDA, NestedTensorCPU, NestedTensorCUDA: transformer_encoder_layer_forward - autogen: _transformer_encoder_layer_fwd.out - -- func: _native_multi_head_attention(Tensor query, Tensor key, Tensor value, int embed_dim, int num_head, Tensor qkv_weight, Tensor qkv_bias, Tensor proj_weight, Tensor proj_bias, Tensor? mask=None, bool need_weights=True, bool average_attn_weights=True, int? mask_type=None) -> (Tensor, Tensor) - variants: function - dispatch: - CPU, CUDA, NestedTensorCPU, NestedTensorCUDA: native_multi_head_attention - autogen: _native_multi_head_attention.out - -- func: _scaled_dot_product_attention(Tensor query, Tensor key, Tensor value, Tensor? attn_mask=None, float dropout_p=0.0, bool need_attn_weights=False, bool is_causal=False) -> (Tensor, Tensor) - python_module: nn - variants: function - autogen: _scaled_dot_product_attention.out - -# Register the math kernel for cpu -- func: _scaled_dot_product_attention_forward(Tensor query, Tensor key, Tensor value, Tensor? attn_mask=None, float dropout_p=0.0, bool need_attn_weights=False, bool is_causal=False) -> (Tensor, Tensor) - variants: function - dispatch: - CUDA: _scaled_dot_product_attention_forward_cuda - CPU: _scaled_dot_product_attention_forward_math - NestedTensorCPU, NestedTensorCUDA: _scaled_dot_product_attention_forward_math - Meta: _scaled_dot_product_attention_forward_math - -- func: _scaled_dot_product_attention_math(Tensor query, Tensor key, Tensor value, Tensor? attn_mask=None, float dropout_p=0.0, bool need_attn_weights=False, bool is_causal=False) -> (Tensor, Tensor) - variants: function - - func: _triton_scaled_dot_attention(Tensor q, Tensor k, Tensor v, float dropout_p=0.0) -> Tensor variants: function dispatch: CUDA: triton_scaled_dot_attention autogen: _triton_scaled_dot_attention.out @@ -13181,24 +13882,21 @@ - func: special_airy_ai(Tensor x) -> Tensor python_module: special structured_delegate: special_airy_ai.out variants: function + tags: pointwise - func: special_airy_ai.out(Tensor x, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_airy_ai_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise -- func: _flash_scaled_dot_product_attention(Tensor query, Tensor key, Tensor value, Tensor cum_seq_q, Tensor cum_seq_k, int max_q, int max_k, float dropout_p, bool is_causal) -> Tensor - variants: function - dispatch: - CUDA: flash_scaled_dot_product_attention - - func: _transformer_decoder_only_layer_fwd(Tensor src, int embed_dim, int num_heads, Tensor qkv_weight, Tensor qkv_bias, Tensor proj_weight, Tensor proj_bias, bool use_gelu, bool norm_first, float eps, Tensor norm_weight_1, Tensor norm_bias_1, Tensor norm_weight_2, Tensor norm_bias_2, Tensor ffn_weight_1, Tensor ffn_bias_1, Tensor ffn_weight_2, Tensor ffn_bias_2, Tensor? mask=None, Tensor? incr_key=None, Tensor? incr_value=None) -> (Tensor, Tensor, Tensor) variants: function dispatch: CPU, CUDA, NestedTensorCPU, NestedTensorCUDA: transformer_decoder_only_layer_forward autogen: _transformer_decoder_only_layer_fwd.out @@ -13211,592 +13909,686 @@ - func: special_bessel_j0(Tensor self) -> Tensor python_module: special structured_delegate: special_bessel_j0.out variants: function + tags: pointwise - func: special_bessel_j0.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_bessel_j0_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_bessel_j1(Tensor self) -> Tensor python_module: special structured_delegate: special_bessel_j1.out variants: function + tags: pointwise - func: special_bessel_j1.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_bessel_j1_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_bessel_y0(Tensor self) -> Tensor python_module: special structured_delegate: special_bessel_y0.out variants: function + tags: pointwise - func: special_bessel_y0.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_bessel_y0_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_bessel_y1(Tensor self) -> Tensor python_module: special structured_delegate: special_bessel_y1.out variants: function + tags: pointwise - func: special_bessel_y1.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_bessel_y1_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_chebyshev_polynomial_t(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_chebyshev_polynomial_t.out variants: function + tags: pointwise - func: special_chebyshev_polynomial_t.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_t.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_t.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_chebyshev_polynomial_t_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_chebyshev_polynomial_t.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_t.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_chebyshev_polynomial_t_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_u(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_chebyshev_polynomial_u.out variants: function + tags: pointwise - func: special_chebyshev_polynomial_u.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_u.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_u.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_chebyshev_polynomial_u_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_chebyshev_polynomial_u.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_u.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_chebyshev_polynomial_u_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_v(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_chebyshev_polynomial_v.out variants: function + tags: pointwise - func: special_chebyshev_polynomial_v.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_v.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_v.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_chebyshev_polynomial_v_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_chebyshev_polynomial_v.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_v.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_chebyshev_polynomial_v_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_w(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_chebyshev_polynomial_w.out variants: function + tags: pointwise - func: special_chebyshev_polynomial_w.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_w.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_w.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_chebyshev_polynomial_w_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_chebyshev_polynomial_w.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_chebyshev_polynomial_w.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_chebyshev_polynomial_w_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_hermite_polynomial_h(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_hermite_polynomial_h.out variants: function + tags: pointwise - func: special_hermite_polynomial_h.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_hermite_polynomial_h.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_hermite_polynomial_h.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_hermite_polynomial_h_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_hermite_polynomial_h.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_hermite_polynomial_h.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_hermite_polynomial_h_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_hermite_polynomial_he(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_hermite_polynomial_he.out variants: function + tags: pointwise - func: special_hermite_polynomial_he.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_hermite_polynomial_he.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_hermite_polynomial_he.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_hermite_polynomial_he_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_hermite_polynomial_he.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_hermite_polynomial_he.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_hermite_polynomial_he_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_laguerre_polynomial_l(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_laguerre_polynomial_l.out variants: function + tags: pointwise - func: special_laguerre_polynomial_l.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_laguerre_polynomial_l.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_laguerre_polynomial_l.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_laguerre_polynomial_l_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_laguerre_polynomial_l.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_laguerre_polynomial_l.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_laguerre_polynomial_l_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_legendre_polynomial_p(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_legendre_polynomial_p.out variants: function + tags: pointwise - func: special_legendre_polynomial_p.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_legendre_polynomial_p.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_legendre_polynomial_p.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_legendre_polynomial_p_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_legendre_polynomial_p.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_legendre_polynomial_p.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_legendre_polynomial_p_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_modified_bessel_i0(Tensor self) -> Tensor python_module: special structured_delegate: special_modified_bessel_i0.out variants: function + tags: pointwise - func: special_modified_bessel_i0.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_modified_bessel_i0_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_modified_bessel_i1(Tensor self) -> Tensor python_module: special structured_delegate: special_modified_bessel_i1.out variants: function + tags: pointwise - func: special_modified_bessel_i1.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_modified_bessel_i1_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_modified_bessel_k0(Tensor self) -> Tensor python_module: special structured_delegate: special_modified_bessel_k0.out variants: function + tags: pointwise - func: special_modified_bessel_k0.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_modified_bessel_k0_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_modified_bessel_k1(Tensor self) -> Tensor python_module: special structured_delegate: special_modified_bessel_k1.out variants: function + tags: pointwise - func: special_modified_bessel_k1.out(Tensor self, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_modified_bessel_k1_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_scaled_modified_bessel_k0(Tensor x) -> Tensor python_module: special structured_delegate: special_scaled_modified_bessel_k0.out variants: function + tags: pointwise - func: special_scaled_modified_bessel_k0.out(Tensor x, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_scaled_modified_bessel_k0_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_scaled_modified_bessel_k1(Tensor x) -> Tensor python_module: special structured_delegate: special_scaled_modified_bessel_k1.out variants: function + tags: pointwise - func: special_scaled_modified_bessel_k1.out(Tensor x, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_scaled_modified_bessel_k1_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_t(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_shifted_chebyshev_polynomial_t.out variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_t.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_t.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_t.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_shifted_chebyshev_polynomial_t_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_t.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_t.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_shifted_chebyshev_polynomial_t_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_u(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_shifted_chebyshev_polynomial_u.out variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_u.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_u.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_u.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_shifted_chebyshev_polynomial_u_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_u.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_u.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_shifted_chebyshev_polynomial_u_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_v(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_shifted_chebyshev_polynomial_v.out variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_v.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_v.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_v.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_shifted_chebyshev_polynomial_v_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_v.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_v.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_shifted_chebyshev_polynomial_v_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_w(Tensor x, Tensor n) -> Tensor device_check: NoCheck python_module: special structured_delegate: special_shifted_chebyshev_polynomial_w.out variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_w.x_scalar(Scalar x, Tensor n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_w.n_scalar(Tensor x, Scalar n) -> Tensor device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_w.out(Tensor x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck dispatch: CPU, CUDA: special_shifted_chebyshev_polynomial_w_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_w.x_scalar_out(Scalar x, Tensor n, *, Tensor(a!) out) -> Tensor(a!) device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_shifted_chebyshev_polynomial_w.n_scalar_out(Tensor x, Scalar n, *, Tensor(a!) out) -> Tensor(a!) dispatch: CompositeExplicitAutograd: special_shifted_chebyshev_polynomial_w_out device_check: NoCheck python_module: special variants: function + tags: pointwise - func: special_spherical_bessel_j0(Tensor x) -> Tensor python_module: special structured_delegate: special_spherical_bessel_j0.out variants: function + tags: pointwise - func: special_spherical_bessel_j0.out(Tensor x, *, Tensor(a!) out) -> Tensor(a!) dispatch: CPU, CUDA: special_spherical_bessel_j0_out python_module: special structured_inherits: TensorIteratorBase structured: True variants: function + tags: pointwise # Aux function used in the test TestPythonDispatch.test_kwarg_only_and_positional_default # within test/test_python_dispatch.py - func: _foobar(Tensor self, bool arg1=True, bool arg2=True, *, bool arg3=True) -> Tensor dispatch: @@ -13808,5 +14600,12 @@ # Unlike "foreach" functions, lists of tensors should be guaranteed to be on the same device (for now). variants: function dispatch: CUDA: _fused_adam_kernel_cuda_ autogen: _fused_adam, _fused_adam.out + +- func: _fused_adamw_(Tensor(a!)[] self, Tensor(b!)[] grads, Tensor(c!)[] exp_avgs, Tensor(d!)[] exp_avg_sqs, Tensor(e!)[] max_exp_avg_sqs, Tensor[] state_steps, *, float lr, float beta1, float beta2, float weight_decay, float eps, bool amsgrad, bool maximize, Tensor? grad_scale=None, Tensor? found_inf=None) -> () + # Unlike "foreach" functions, lists of tensors should be guaranteed to be on the same device (for now). + variants: function + dispatch: + CUDA: _fused_adamw_kernel_cuda_ + autogen: _fused_adamw, _fused_adamw.out