Intel® oneAPI Deep Neural Network Developer Guide and Reference
A newer version of this document is available. Customers should click here to go to the newest version.
Visible to Intel only — GUID: GUID-B66D1C2C-1C0E-4239-95CB-CFDA2DCA209E
Visible to Intel only — GUID: GUID-B66D1C2C-1C0E-4239-95CB-CFDA2DCA209E
enum dnnl::graph::op::attr
Overview
Attributes of operations. More…
#include <dnnl_graph.hpp> enum attr { undef = dnnl_graph_op_attr_undef, alpha = dnnl_graph_op_attr_alpha, beta = dnnl_graph_op_attr_beta, epsilon = dnnl_graph_op_attr_epsilon, max = dnnl_graph_op_attr_max, min = dnnl_graph_op_attr_min, momentum = dnnl_graph_op_attr_momentum, scales = dnnl_graph_op_attr_scales, axis = dnnl_graph_op_attr_axis, begin_norm_axis = dnnl_graph_op_attr_begin_norm_axis, groups = dnnl_graph_op_attr_groups, axes = dnnl_graph_op_attr_axes, dilations = dnnl_graph_op_attr_dilations, dst_shape = dnnl_graph_op_attr_dst_shape, kernel = dnnl_graph_op_attr_kernel, order = dnnl_graph_op_attr_order, output_padding = dnnl_graph_op_attr_output_padding, pads_begin = dnnl_graph_op_attr_pads_begin, pads_end = dnnl_graph_op_attr_pads_end, shape = dnnl_graph_op_attr_shape, sizes = dnnl_graph_op_attr_sizes, src_shape = dnnl_graph_op_attr_src_shape, strides = dnnl_graph_op_attr_strides, weights_shape = dnnl_graph_op_attr_weights_shape, zps = dnnl_graph_op_attr_zps, exclude_pad = dnnl_graph_op_attr_exclude_pad, keep_dims = dnnl_graph_op_attr_keep_dims, keep_stats = dnnl_graph_op_attr_keep_stats, per_channel_broadcast = dnnl_graph_op_attr_per_channel_broadcast, special_zero = dnnl_graph_op_attr_special_zero, transpose_a = dnnl_graph_op_attr_transpose_a, transpose_b = dnnl_graph_op_attr_transpose_b, use_affine = dnnl_graph_op_attr_use_affine, use_dst = dnnl_graph_op_attr_use_dst, auto_broadcast = dnnl_graph_op_attr_auto_broadcast, auto_pad = dnnl_graph_op_attr_auto_pad, coordinate_transformation_mode = dnnl_graph_op_attr_coordinate_transformation_mode, data_format = dnnl_graph_op_attr_data_format, mode = dnnl_graph_op_attr_mode, qtype = dnnl_graph_op_attr_qtype, rounding_type = dnnl_graph_op_attr_rounding_type, weights_format = dnnl_graph_op_attr_weights_format, };
Detailed Documentation
Attributes of operations.
Different operations support different attributes. Check the document of each operation for what attributes are supported and what are the potential values for them. Missing required attribute or illegal attribute value may lead to failure when adding the operation to a graph.
Enum Values
undef
Undefined op attribute.
alpha
Specifies an alpha attribute to an op.
beta
Specifies an beta attribute to an op.
epsilon
Specifies an epsilon attribute to an op.
max
Specifies a max attribute to an op.
min
Specifies a min attribute to an op.
momentum
Specifies a momentum attribute to an op.
scales
Specifies a scales attribute to an op.
axis
Specifies an axis attribute to an op.
begin_norm_axis
Specifies a begin_norm_axis attribute to an op.
groups
Specifies a groups attribute to an op.
axes
Specifies an axes attribute to an op.
dilations
Specifies a dilations attribute to an op.
dst_shape
Specifies an dst_shape attribute to an op.
kernel
Specifies a kernel attribute to an op.
order
Specifies an order attribute to an op.
output_padding
Specifies an output_padding attribute to an op.
pads_begin
Specifies a pads_begin attribute to an op.
pads_end
Specifies a pads_end attribute to an op.
shape
Specifies a shape attribute to an op.
sizes
Specifies a sizes attribute to an op.
src_shape
Specifies an src_shape attribute to an op.
strides
Specifies a strides attribute to an op.
weights_shape
Specifies a weight_shape attribute to an op.
zps
Specifies a zps attribute to an op.
exclude_pad
Specifies an exclude_pad attribute to an op.
keep_dims
Specifies a keep_dims attribute to an op.
keep_stats
Specifies a keep_stats attribute to an op.
per_channel_broadcast
Specifies a per_channel_broadcast attribute to an op.
special_zero
Specifies a special_zero attribute to an op.
transpose_a
Specifies a transpose_a attribute to an op.
transpose_b
Specifies a transpose_b attribute to an op.
use_affine
Specifies an use_affine attribute to an op.
use_dst
Specifies an use_dst attribute to an op.
auto_broadcast
Specifies an auto_broadcast attribute to an op.
The value can be “none” or “numpy”.
auto_pad
Specifies an auto_pad attribute to an op.
The value can be “none”, “same_upper”, “same_lower”, or “valid”.
coordinate_transformation_mode
Specifies an coordinate_transformation_mode attribute to an op.
The value can be “half_pixel” or “align_corners”. The attribute is defined for Interpolate operations.
data_format
Specifies a data_format of an op. The value can be “NCX” or “NXC”.
mode
Specifies a mode attribute of an op.
The value can be “nearest”, “linear”, “bilinear”, or “trilinear”. The attribute is defined for Interpolate operations.
qtype
Specifies a qtype attribute to an op.
The value can be “per_channel” or “per_tensor”. The attribute is defined for quantization operations.
rounding_type
Specifies a rounding_type attribute to an op.
The value can be “ceil” or “floor”.
weights_format
Specifies a weights_format of an op.
The value can be “OIX”, “XIO”, “IOX”, or “XOI”. Different operations may support different values.